GLUCOCORTICOIDS IN ASSOCIATIVE LEARNING

GLUCOCORTICOIDS IN ASSOCIATIVE LEARNING AND MEMORY CONSOLIDATION

Michael Zorawski, BSc (Hons)

Dissertation submitted for the higher degree of Philosophiae Doctor (PhD)

School of Psychology Cardiff University, United Kingdom March / May 2002

1

CONTENTS

page I

Contents Declaration and statements

IV

Supervisor-, dissertation submission-, viva- and graduation information

V

Synopsis

VI

Foreword and acknowledgements

VII

Conference presentations and publications

IX

List of abbreviations used in this dissertation

XI XIV

Animal welfare statement

CHAPTER 1 About glucocorticoids and their role in learning and memory 1.1 NEUROBIOLOGY OF GLUCOCORTICOIDS

1 4

1.1.1 The general concept of brain–pituitary–visceral organ axes

4

1.1.2 The hypothalamic-pituitary-adrenal axis

8

1.1.3 Diurnal glucocorticoid secretion rhythm

10

1.1.4 Negative feedback from glucocorticoids

11

1.1.5 Corticosteroid receptor types and the actions of glucocorticoids in the brain

12

1.1.6 Stress-induced neuronal activation of the HPA axis

15

1.1.7 Individual differences in stress levels of glucocorticoids

18

1.2 THE EFFECTS OF GLUCOCORTICOIDS ON MEMORY 1.2.1 The acute effects of glucocorticoid on memory acquisition and consolidation (animal studies)

19 21

1.2.1.1 Early studies

22

1.2.1.2 Studies employing appetitive experimental procedures

23

1.2.1.3 Studies employing selective MR- and GR- agonists and antagonists, or glucocorticoid synthesis inhibitors

24

1.2.1.4 Studies investigating interactions between glucocorticoid and training intensity levels

28

1.2.1.5 Studies comparing effects on discrete-cue and contextual Pavlovian conditioning paradigms

30

1.2.1.6 Studies investigating glucocorticoid memory modulation on an anatomical, system interaction-, and molecular level

32

I

1.2.1.7 Studies manipulating glucocorticoid levels by stressor exposure

42

1.2.1 The acute effects of glucocorticoids on memory acquisition and consolidation (human studies)

44

1.2.2.1 Studies involving the administration of exogenous glucocorticoids or the measurement of endogenous glucocorticoid levels

44

1.2.2.2 Studies comparing memory for emotional and neutral items

49

1.2.3 Effects of chronically elevated glucocorticoid levels on memory

53

1.2.4 Effects of glucocorticoids on memory retrieval

54

1.2.5 Effects of glucocorticoids in electrophysiological studies

56

1.2.6 Summary, limitations of previous work, outline of the empirical components of this project

60

CHAPTER 2 Memory modulation by glucocorticoids in aversive and appetitive discrete-cue Pavlovian conditioning paradigms 2.1 GLUCOCORTICOIDS IN DISCRETE-CUE FEAR CONDITIONING I 2.2 GLUCOCORTICOIDS IN DISCRETE-CUE FEAR CONDITIONING II 2.3 GLUCOCORTICOIDS IN APPETITIVE DISCRETE-CUE CONDITIONING I 2.4 GLUCOCORTICOIDS IN APPETITIVE DISCRETE-CUE CONDITIONING II 2.5 GENERAL DISCUSSION

66 72 79 85 92 96

CHAPTER 3 Do glucocorticoids modulate associations lacking a discrete US ?

106

3.1 GLUCOCORTICOIDS IN LATENT INHIBITION

107

3.2 GLUCOCORTICOIDS IN EXTINCTION

119

3.3 GLUCOCORTICOIDS IN SENSORY PRECONDITIONING

129

3.4 GLUCOCORTICOIDS IN SECOND-ORDER CONDITIONING

139

3.5 GENERAL DISCUSSION

148

CHAPTER 4 Differential effects of glucocorticoids in the memory modulation of sensory and motivational reinforcer properties 4.1 GLUCOCORTICOIDS IN PAVLOVIAN-INSTRUMENTAL TRANSFER I (PILOT) 4.2 GLUCOCORTICOIDS IN PAVLOVIAN-INSTRUMENTAL TRANSFER II 4.3 GENERAL DISCUSSION

155 156 168 176

II

CHAPTER 5 Individual differences in learning due to differences in HPA-reactivity ? 5.1 LOCOMOTOR ACTIVITY TO NOVELTY IN A POPULATION OF LISTER-HOODED RATS 5.2 HR VS. LR IN APPETITVE DISCRETE-CUE PAVLOVIAN CONDITIONING 5.3 HR VS. LR IN AVERSIVE DISCRETE-CUE PAVLOVIAN CONDITIONING

182 187 194 201

5.4 HR VS. LR IN INSTRUMENTAL LEARNING

210


226

CHAPTER 6 Overall summary, limitations and conclusions 6.1 OVERALL SUMMARY

230 232

6.1.1 Discrete-cue vs. context learning modulation

232

6.1.2 Modulation of appetitive learning

235

6.1.3 Glucocorticoids in non-US learning

237

6.1.4 Differential modulation of different reinforcer properties

242

6.1.5 Individual differences in learning between high and low responders to stress

246

6.2 LIMITATIONS

251

6.3 CONCLUDING REMARKS

253

APPENDIX

256

Pilots, parameters and drug selection A.1 FEAR CONDITIONING: (Pilot-) Experiment A.1

257

A.2 FEAR CONDITIONING: (Pilot-) Experiment A.2

264


271

A.4 WHICH DRUG TO SELECT ?

275


281

REFERENCES

287

Author information, contact details, word count

332

III

DECLARATION

This work has not previously been accepted in substance for any degree and is not being concurrently submitted in candidature for any degree.

Signed ...................................................... (candidate) Date ...........................................................

STATEMENT 1 This thesis is the result of my own investigations, except where otherwise stated. Other sources are acknowledged by footnotes giving explicit references. A bibliography is appended.


STATEMENT 2 I hereby give consent for my thesis, if accepted, to be available for photocopying and for inter-library loan, and for the title and summary to be made available to outside organisations.


IV

This PhD project has been supervised by Dr. Simon Killcross.

The dissertation was originally submitted on March 8, 2002. The final version will be bound on May 9, 2002.

The viva voce took place in the School of Psychology at Cardiff University on May 7, 2002. Professor John Pearce was the Chair of the examination board. The examiners were Professor John Aggleton (internal) and Dr. Verity Brown (external, St. Andrews University).

The graduation ceremony will take place in St Davids Hall, Cardiff, on July 16, 2002.

V

Synopsis Glucocorticoids have been shown to fulfil a modulatory function in memory consolidation. This has primarily been investigated in animals, using aversive, context-dependent, emotionally salient tasks. The aim of the current project was to investigate more closely the role of glucocorticoids in associative learning and memory consolidation by conducting a range of behavioural tests, using rats as subjects, and post-training administration of a glucocorticoid receptor agonist (dexamethasone) as pharmacological manipulation. Glucocorticoid memory modulation was demonstrated in discrete-cue Pavlovian conditioning procedures of both appetitive and aversive nature. No such modulation was found in a range of procedures that lack a discrete primary reinforcer and that vary in their inherent emotional significance (latent inhibition, extinction, sensory preconditioning and second-order conditioning). A Pavlovian-instrumental transfer design was used to examine whether glucocorticoid memory modulation is specific for particular properties of motivationally significant events. Glucocorticoids were found to impair associations between a stimulus and the sensory properties of a reinforcer. As glucocorticoids enhanced overall Pavlovian conditioning, a selective role for memory enhancement of the motivational properties of the reinforcer is proposed. Finally, an attempt was made to employ an endogenous manipulation of glucocorticoid-release, making use of existing individual differences in stress reactivity. High and low responders were identified from a population of rats, based on their locomotor activity in response to novelty. They were then compared in a number of learning procedures. High responders were found to show greater levels of both appetitive and aversive discrete-cue Pavlovian conditioning. Further, the groups differed in the way they performed in an instrumental learning procedure. The results are summarised to add to the current picture of glucocorticoids in memory consolidation, and an attempt is made to integrate them into a broader picture with possible implications on learning processes and the role of stress and emotion in human mental disorders.

VI

Foreword and acknowledgements I began my PhD in October 1998 at the University of York where I had also been an undergraduate psychology student. During my final year as an undergraduate, I was, like many fellow students, particularly interested in clinical psychology, which I was hoping to study in the United States. But while I spent a lot of time, effort and (mostly my parents’) money on applications and aptitude tests to pursue this goal, I also began to find out more about areas of psychology that I had yet been less aware of. I realised that fields like behavioural neuroscience and animal learning theory, which were concerned with the basic mechanisms of both normal and pathological human behaviour, might be just as or even more interesting and important to study than clinical psychology. I first met my supervisor, Dr. Simon Killcross, when he was lecturing one of my classes at York. He was very helpful in advising me on future possibilities and on how to prepare for interviews, but was eventually looking for a PhD student himself. I was fortunate enough to be offered the place and had to decide between working with him at York and going to Cardiff. Even though I had a really pleasant impression of Cardiff after talking to Professors Hadyn Ellis and John Pearce, I eventually decided to stay in York for a number of reasons. However, Fate seemed to want me to go to Cardiff, and eventually I did end up there when Simon and his lab transferred there in September 1999. Approximately two years later, I finally began writing up my dissertation. At least ever since then, I have been longing for the moment when the only thing left to do was to write a foreword and an acknowledgement section. Now that the time has come, I perceive it as much less spectacular and exciting than anticipated back then. I think this might be due to being very tired and worn out from the recent, very hectic, weeks. Completing this dissertation turned out to take much longer than I had originally believed it would, and now that it is almost done, I am immediately facing a lot of other important events during the remainder of this month, such as job interviews and temporarily moving back to my home town. Therefore, writing down these final lines takes place in much less a celebratory way than I had once envisaged. Nevertheless, I feel very happy and relieved. This PhD project played a very dominant role in my life over the last three-and-a-half years. I perceived it as a long, uncertain journey along a winding road that was hopefully leading somewhere meaningful. Even though I was well taken care of by my supervisor and others, quite often I felt rather lonely travelling down this road. It is difficult to conceive that I now seem to have arrived at a place where this particular journey ends for me. All in all, doing this project constituted a great experience from which I learnt a lot and which I would not have wanted to miss. There are many people who have helped me with it, who have been there for me during those years and/or who made my time in York and Cardiff a more pleasant one. It is them I would like to acknowledge in the following paragraphs.

Michael Zorawski Cardiff, March 08, 2002

I would like to thank… …Simon Killcross who has been a great, accessible and patient supervisor whose clever insights and good advice I much relied on, and who kept reminding me “not to worry too much.” …my parents, Heidrun and Jürgen Zorawski, who have always been there for me in every respect. Without them, I wouldn’t have been able to study abroad. They provided me with a safe haven where I regularly restored my energy reserves and acquired new motivation. Vielen lieben Dank fuer alles, Mapa ! Ihr seid die Besten! …my favourite cousin and best-friend-in-the-world, Martin Zorawski, without whom the last few years would have been much more difficult. Danke dafuer, dass Du da bist, Njauge!!

VII

…my very special friend Georgia Samolada who I am very glad to have met in Cardiff (stats class!). Thank you, kale Saur, for putting up with me all that time! xxx … Etienne Coutureau, my “academic older brother”, who was always there when I needed advice and who recently became more and more of a good friend as well. Good luck in Bordeaux and keep in touch!! …Pam Blundell, my “academic older (although younger) sister”, whose help I much appreciated. …everybody in the lab and department at York and Cardiff who helped me a lot and made me feel comfortable: Thank you, Angie Morgan (“lab-angel!”), Vincent Casteras (my other French source of advice), Sophie Dix (who taught me surgery), Mark Good (who tried to explain LTP to me), Ana da Costa (who showed me what a lesion looks like), John Pearce (who made me feel at home in Cardiff much before I joined the lab), Jasper Ward-Robinson, Dave George, Mark Haselgrove, Rob Honey, John Aggleton, Eman Amin, Kerry Housler, Janice Muir, Trish Jenkins, Seralynne Vann, Kristina Wilton, Jo Oswald, Mark Day, Anthony McGregor, Andy Hayward, Elaine Sweet (It’s a pity you quit!), Geoffrey Hall (who used to be in my research committee), Euan Macphail, Amanda Harrison, Charlotte Bonardi (Friday afternoon football!), Ian Shevill, Erica Setzu (who took over my nice desk and office in York), Gavin Phillips, Michelle Symonds, Maggie Snowling (my undergraduate supervisor), all my Cardiff office mates in 10.20 (Mark Isaac, Mike Dunn, Geoff Duggan, Jenny Naji, Fiona Harrison, Alex Payne, Stuart Huddart, Alastair Barrowcliff, Nick Row and Amelia Woodward), the secretarial and technical staff at York and Cardiff (especially Brenda Westrope, Dave Griffiths, Hilmar Jay, Den Simmonds, Dave Johnson, Alan Jones and Lorraine Awcock), and anyone I might have accidentally forgotten... Last but not least I would like to thank Tanya Patel, Josie Monagham, Simon Moore, Fiona Phelps, Hanz Neth and Suzy Charman. …my other friends at York, especially Steve Graham (who has not only been the best officemate I can possibly imagine but who has also become a very dear friend; Say hello to Tomasina and Hugo Alexander ), Susi Niederhuber (Alles liebe, meine HFK!!!), Fred Hardy (Juventus-Hamburg 1-3!), Ingebjorg Tønnesen (Will you provide me with free therapy?), Aisha Khan (Ribbit!), Sahori Watanabe, Patrick Rinke, Simon McCabe and all the other Yorkies. Thanks also to my Cardiff housemates Doug Boyd and Joyceline Cuenco. …all my other friends and family, especially outside the UK, who I have not seen much of in the last few years and who I am hoping to catch up with soon! …all those academics and fellow postgrads from other labs who encouraged and inspired me, gave me advice, let me visit their labs, socialised with me at conferences or simply answered my e-mails or reprint requests. Thank you, James McGaugh & Larry Cahill (whose work inspired and encouraged me, and whose hospitality when visiting Irvine much impressed me), Edmund Fantino (my first animal learning teacher who I kept in touch with ever since being at UCSD), Michael Koch, Colin Hendrie, Irit Akirav, Marcel van Gaalen, Barry Setlow, Benno Roozendaal, Melly Oitzl, Nektarios Mazarakis, Desiree Gonzalo, Ines Goerendt, Chris Lowry, Thomas Steckler, Dominique de Quervain, Athina Markou, Ralph Adolphs, Tony Buchanan, Turhan Canli, Kevin LaBar, George Koob, Stephen Hamann, Pier Vincenzo Piazza, John Glowa, Marc Geyer, Alexander Cools, Hans Markowitsch, Bruce McEwen, David Diamond, Markus Fendt, Mathew Martin-Iverson, Mohammed Kabbaj, Vincent Stretch, John Polich, Lucianne Groenink, Adrie Bruijzeel, Catherine Harmer, Daniel Reisel, Paul Soseng, Mathew Holalan and Liz Sepulveda. …the BNA, the BAP, the Welsh branch of the BPS, the journal “Brain”, Psypag, and the School of Psychology (Cardiff University) for providing funding and travel grants for my conference visits. ...Professors Verity Brown (University of St. Andrews) and John Aggleton who have agreed to be my examiners on this project, and who might therefore be the only people ever to have read this dissertation.

VIII

Conference presentations and publications The data reported in this dissertation have so far resulted in the following conference presentations and publications.

Oral conference presentations Summer Meeting of the British Association for Psychopharmacology (BAP) in Harrogate (UK), July 2001: Zorawski, M., & Killcross, S. (2001a). Posttraining glucocorticoids modulate CS-US but not CS-nothing associations. Journal of Psychopharmacology, Supplement to Volume 15 Number 3, A69.

Summer Meeting of the Psychology Postgraduate Affairs Group (PsyPAG) in Sheffield (UK), July 2001: Zorawski, M., & Killcross, S. Posttraining Glucocorticoids in aversive and appetitive pavlovian discrete cue conditioning. Unpublished.

Poster presentations Animal Learning Symposium IV in Gregynog (UK) in April 2000: Zorawski, M. & Killcross, S. Posttraining glucocorticoid agonist enhances memory in appetitive and in aversive Pavlovian discrete-cue conditioning paradigms. Unpublished.

Forum of European Neuroscience (FENS) in Brighton (UK), June 2000: Zorawski, M., & Killcross, S. (2000a). Posttraining glucocorticoid receptor agonist enhances memory in an appetitive discrete cue Pavlovian conditioning paradigm. European Journal of Neuroscience, 12(Supplement 11), 78.3.

Summer Meeting of the British Association for Psychopharmacology (BAP) in Cambridge (UK), July 2000: Zorawski, M., & Killcross, S. (2000b). Effects of posttraining/postpreexposure glucocorticoid receptor agonist in appetitive and aversive discrete cue Pavlovian conditioning paradigms and latent inhibition. Journal of Psychopharmacology, Supplement to Volume 14 Number 3, A51.

IX

30th Annual Meeting of the Society for Neuroscience (SFN) in New Orleans (USA), November 2000: Zorawski, M., & Killcross, S. (2000c). Post-training dexamethasone in appetitive and aversive cue conditioning and latent inhibition. Society for Neuroscience Abstracts, 26(2), 562.3.

Animal Learning Symposium V in Gregynog (UK) in April 2001: Zorawski, M., & Killcross, S. Glucocorticoids modulate emotional but not non-emotional memories. Unpublished.

Oxford University Cognitive Neuroscience Autumn School in Oxford (UK) in September 2001: Zorawski, M., & Killcross, S. Glucocorticoids modulate CS-US but not CS-nothing memories. Unpublished.

7th conference on the neurobiology of learning and memory: “Orchestration of cells and systems. Making memories in the brain.” at the University of California Irvine (USA) in November 2001. Zorawski, M., & Killcross, S. Habit formation in high- and lowresponder rats. Unpublished. 31st Annual Meeting of the Society for Neuroscience (SFN) in San Diego (USA), November 2001: Zorawski, M., & Killcross, S. (2001b). Memory modulation by glucocorticoids is selective for the motivational properties of the US evidence from a Pavlovian-instrumental transfer study. Society for Neuroscience Abstracts, 27, 84.11.

Journal publication Zorawski, M., & Killcross, S. (accepted). Post-training glucocorticoid receptor agonist enhances memory in appetitive and aversive Pavlovian discrete-cue conditioning paradigms. Neurobiology of learning and memory.

X

List of abbreviations used in this dissertation 5-HT A AA ACTH ADX AN ANI ANOVA A-O ATP AVP BDNF BLA C ºC CA1 cAMP CEA CFC Ch CON COND CORT CR CRF CRH CS DA dB DCFC DEX DG DHEA DHEA-S DIFF E EXP EXT F FI FSH g GH GR

serotonin agonist active avoidance adrenocorticotrophic hormone adrenalectomy antagonist anisomycin analysis (analyses) of variance action-outcome adenosine triphosphate arginine vasopressin brain-derived neurotrophic factor basolateral nucleus of the amygdala conditioning (trial or session) degrees Celsius Cornu Ammonia 1 (area 1 of the Ammon’s Horn of the hippocampus) cyclic adenosine monophosphate central nucleus of the amygdala contextual fear conditioning chapter control group conditioning corticosterone conditioned response corticotropin-releasing factor (US-American term for CRH) corticotropin-releasing hormone conditioned stimulus dopamine decibel discrete-cue fear conditioning dexamethasone dentate gyrus (of the hippocampus) dehydroepiandrosterone dehydroepiandrosterone sulfate different (stimulus) extinction (trial or session) experimental group extinction Fisher's F ratio fixed interval (reinforcement schedule) follicle-stimulating hormone gram growth hormone glucocorticoid (type II) receptor XI

GRHAB HPA HR Hz ic icv ip ISI kg kHz LA LH LI LiCl LL LP LR LTD LTP M mA ME MEA MeR mg ml min mpPVN MR MRMWM n N NA NAC NaCl NCAM ng NGF NT-3 NTS OC OSD p p.

glucocorticoid receptor-selective habituation hypothalamic-pituitary-adrenal high responders hertz intracerebral intracerebroventricular intraperitoneal inter-stimulus-interval kilogram kilohertz lateral nucleus of the amygdala luteinising hormone latent inhibition lithium chloride left lever lever presses low responders long-term depression long-term potentiation mean milliampere magazine entries medial nucleus of the amygdala medium responders milligram milliliter minute(s) medial parvicellular part of the paraventricular nucleus mineralocorticoid (type I) receptor mineralocorticoid receptor-selective Morris water maze number of subjects in a subgroup total number of subjects noradrenaline nucleus accumbens sodium chloride (saline) neural cell adhesion molecule nanogram nerve growth factor neurotrophin-3 nucleus tractus solitaris (nucleus of the solitary tract) operant (instrumental) conditioning operant successive discrimination probability page XII

PA PBP PET PIT PKA POMC pp. PRL PTSD PVN r RI RL RNF s SAL sc SEM SOC SPC S-R S-S ST TRH TSH US VI VT WAIS-R

passive avoidance primed burst potentiation positron emission tomography Pavlovian-instrumental transfer protein kinase A proopiomelanocortin pages prolactin post-traumatic stress disorder paraventricular nucleus of the hypothalamus Pearson's product moment correlation random interval (reinforcement schedule) right lever reinforcements second(s) saline subcutaneous standard error of the mean second-order conditioning sensory preconditioning stimulus-response stimulus-stimulus stria terminalis thyrotropin-releasing hormone thyroid-stimulating hormone unconditioned stimulus variable interval (reinforcement schedule) variable time (reinforcement schedule) Wechsler Adult Intelligence Scale – Revised

XIII

Animal welfare statement

All animals were treated in accordance with the UK Animals (Scientific Procedures) Act 1986. Experiments were carried out under Home Office personal licence PIL 60/7313 (MZ) and project licence PPL 60/2470.

…and some thoughts on animal research: In this project, I have used rats as subjects. Animal experiments in general are often being regarded as cruel by many members of the general public, and scientist who work with animals are sometimes portrayed as sadistic psychopaths who enjoy torturing animals. This could not be further from the truth. Everybody I met working with animals did so in a most responsible and caring fashion, minimising all sources of discomfort for the animals, while always appreciating the ethical issues concerned with their work. People rely on animal research everyday without noticing. Given the role of animals in most people’s diets, I hope that scientists who carry out animal research for medical purposes, directly or indirectly, will be given more credit in the future for what they do. At the same time, I want to stress that it is nevertheless very important to be critical towards the unnecessary use of animals in research, and to always appreciate that laboratory animals are living creatures that need to be treated with respect.

XIV

CHAPTER 1 About glucocorticoids and their role in learning and memory

1

Certain events in life seem to be of greater importance than others, that is events that trigger emotional responses. Emotions are central motive states arising when an organism responds to events of biological significance, such as danger, threat, or reward. Fear, for example, is a normal adaptive response which enables and motivates the organism to react to a threatening stimulus by either withdrawing from it, fighting it or in some cases complete behavioural inhibition, in order to maximise chances for survival. In order to be best prepared for threatening situations in the future, it is important for an organism to learn about stimuli that signal potential threat. This also applies to stimuli signalling, for example, food or a potential mate. Hence, it makes sense that learning and memory are enhanced for emotionally salient events, as compared with non-significant ones. Most people will remember better about when they ran away from a vicious dog or when they were asked to get married, even if that was many decades ago, than they will about what colour of shirt they wore yesterday. Similarly, most people (depending, of course, on their age and origin) will remember well where they were and what they were doing when they found out about Kennedy’s assassination, the fall of the Berlin wall, Lady Diana’s accident, or the terrorist attack on the World Trade Center in New York. In addition to anecdotal evidence occurring in everyday life, it has also been experimentally demonstrated in humans that memory for emotional events is generally better than for neutral ones (Bradley, Greenwald, Petry, & Lang, 1992; Burke, Heuer, & Reisberg, 1992; Christianson & Loftus, 1991; Cahill, Babinsky, Markowitsch, & McGaugh, 1995; Cahill et al., 1996; Cahill & McGaugh, 1995; Cahill, Prins, Weber, & McGaugh, 1994; van Stegeren, Everaerd, Cahill, McGaugh, & Gooren, 1998). Despite the intuitively adaptive value of good memory for survival, it might also seem difficult to

2

keep one’s memory organised if everything was to be remembered to the same extent. A quote by William James (1890/1950) illustrates this beautifully: Selection is the very keel on which our mental ship is built. And in the case of memory its utility is obvious. If we remembered everything, we should on most occasions be as ill off as if we remembered nothing. (p. 680)

On the contrary, the overconsolidation of emotional memory can also trigger the inappropriate and excessive occurrence of otherwise adaptive emotions such as fear, and lead to anxiety disorders such as post-traumatic stress disorder (PTSD) (Rosen & Schulkin, 1998). James McGaugh developed a technique to assess the effects of drugs and other treatments on memory consolidation unconfounded by possible effects on acquisition or retrieval by administering them shortly after training, and found that certain treatments can enhance memory consolidation (Breen & McGaugh, 1961; McGaugh, 1966, 1973). Based on McGaugh’s ideas about memory modulation, it has been suggested that an endogenous system might regulate the degree of memory consolidation (Gold & McGaugh, 1975). Like the adrenal catecholamine adrenaline (Gold & van Buskirk, 1975, 1976, for a review see Gold, 1989), glucocorticoids are thought to constitute a neurobiological substrate that is involved in this selective memory modulation system. Glucocorticoids are adrenocortical stress hormones that are released in increased amounts during an emotional event, via activation of the hypothalamic-pituitary-adrenal (HPA) axis. During such an event, the HPA axis is activated alongside the sympathetic branch of the autonomic nervous system which prepares an organism for the so-called fight-or-flight-response. The purpose of this introductory chapter is first to describe the neurobiological mechanisms by which glucocorticoids are released and act in the brain. Second, and more importantly, its aim is to review the existing literature on the role of the

3

adrenocortical systems in memory, both in animal and human subjects. Certain important distinctions between studies are made in an attempt to clarify the often conflicting results reported in this area of research, and the general overall conclusion of previous work is conveyed. Finally, some of the possible limitations of the existing studies are outlined and discussed. Certain questions are raised that have not yet been answered and which form the theoretical foundation of this PhD project. These questions, which are pursued in the empirical chapters to follow, deal with issues such as the applicability of glucocorticoid memory modulation to a certain form of learning, discrete-cue Pavlovian conditioning, in different directions of emotional valence (Chapter 2), its applicability to forms of learning that lack the occurrence of a discrete emotional event (Chapter 3), the specific properties of the emotional event that are modulated by glucocorticoids (Chapter 4), and finally possible individual differences in glucocorticoid memory modulation and how they might have interesting implication regarding predispositions to certain psychopathological conditions (Chapter 5).

1.1 THE NEUROBIOLOGY OF GLUCOCORTICOIDS

1.1.1 The general concept of brain–pituitary–visceral organ axes In neural communication, neuroactive substances such as neurotransmitters are usually released at the synapse. The situation is different in neuroendocrine regulation where neurons release peptides into the bloodstream which act at a distance as neurohormones. Neuroendocrine regulation is, for example, involved in homeostasis of cellular metabolism, growth and development, osmoregulation,

4

reproduction, and response to stress. There are five regulatory neuroendocrine axes that all originate in, and are controlled by, the brain, and which are then relayed through the anterior pituitary, the master gland, to finally produce hormones in one of several different visceral organs. Figure 1.1.1.1 provides a schema of these five neuroendocrine axes. They can be separated on the basis of these different organs or glands, constituting their endpoints: the adrenal cortex, the thyroid, the gonads (ovary in females, testis in males), the mammary glands, and the liver, as an example for an organ stimulated by growth hormone (GH). Neural and hormonal feedback controls the secretion of either a releasing hormone, or an inhibitory hormone, from neurons of the hypothalamus into the hypothalamopituitary portal system, a specialised vasculature (Akil et al., 1998; Kalat, 1995). Neurons located in several hypothalamic nuclei, e.g. the paraventricular nucleus (PVN), the preoptic area, the arcuate nucleus and the periventricular nucleus, release these hormones through their axon terminals, located in the median eminence, one of six midline circumventricular organs. It expands from the floor of the hypothalamus to the pituitary gland and consists of three regions: the ependymal zone that surrounds the infundibular recess of the third ventricle; the internal zone which surrounds the ependymal zone and which contains axons of vasopressin- and oxytocin-producing neurons (these neurons are located in the supraoptic and paraventricular nuclei which cross the hypothalamohypophyseal system to terminate in the posterior pituitary); and the external zone which contains axons and terminals of inhibitory and releasing hormone neurons and vessels of the portal plexus. The portal vessels make the external zone of the median eminence highly vascular. The portal system consists of a complex array of capillaries and veins which arise from the

5

Brain (Hypothalamus-PVN) stimulating hormones (e.g. CRH, TRH etc.)

Pituitary ACTH

Adrenal Cortex

Glucocorticoids

TSH

Thyroid

GH

Liver (but growth effect on most cells of body)

Thyroxine Triiodothyronine

Insulinlike growth factor I

LH & FSH

PRL

?

?

Ovary

Testis

Mammary glands (breast)

Testosterone Estrogen Progesterone

Figure 1.1.1.1. Schema of the brain – pituitary – organ axes.

Abbreviations: ACTH = adrenocorticotropic hormone; CRH = corticotropin-releasing hormone; GH = growth hormone (somatotropin); FSH = follicle-stimulating hormone; LH = luteinising hormone; PRL = prolactin; PVN = paraventricular nucleus; TRH = thyrotropin-releasing hormone; TSH = thyroid-stimulating hormone.

6

superior hypophyseal artery, a branch of the internal carotid. The releasing or inhibitory hormones are secreted from the axon terminals of the neurons of the tuberoinfundibular system. They then enter the portal capillaries and long portal veins by which they reach venous sinusoids surrounding anterior pituitary cells. There, these different kinds of specialised cells secrete stimulating hormones only in response to a specific releasing or inhibitory hormone. Thyrotopes, for example, only respond to thyrotropin-releasing hormone (TRH) and secrete thyroid-stimulating hormone (TSH), whereas corticotrophs only respond to corticotropin-releasing hormone (CRH; also known as corticotropin-releasing factor or CRF) and secrete adrenocorticotropic hormone (ACTH). The pituitary-stimulating hormones then cross the systemic circulation to interact with receptors of glandular, endocrine cells of a number of visceral organs and glands (e.g. in the adrenal cortex) from where further kinds of hormones are released into the systemic circulation (e.g. glucocorticoids) (Akil et al., 1998; Kalat, 1995). The secretion of hypothalamic and pituitary hormones occurs in brief ultradian pulses, the frequency often being about one pulse every one to three hours. The pulsatile secretion by the hypothalamus runs the pulsatile secretion of the anterior pituitary. Circadian control is also important for the secretion of hormones. In the hypothalamic-pituitary-adrenal axis, for example, there is a daily secretion rhythm at all three levels, in the case of cortisol (in humans) peaking in the morning, and reaching a trough at night (see Section 1.1.3) (Akil et al., 1998). The

secretion

of

the

releasing

and

inhibitory

hormones

by

the

tuberoinfundibular neurons takes place as a function of neuronal firing from neurons in different brain areas (see Section 1.1.6). Like the release of pituitary hormones, it is also regulated by feedback from the secreted hormones upstream on the axis, e.g.

7

hormones released by the visceral organs act on the pituitary to down-regulate the release of trophic hormones and some of these feed back to the hypothalamus (see Section 1.1.4).

1.1.2 The hypothalamic-pituitary-adrenal axis As with the adrenal medulla that produces and releases noradrenaline (NA) and adrenaline, and the autonomic nervous system (ANS) that modulates physiologic functions via neurotransmitters, the hypothalamic-pituitary-adrenal or HPA axis plays a crucial role in an animal’s response to stressful stimuli. Figure 1.1.1.2 outlines the HPA axis. The hypothalamus, particularly the medial parvicellular part of the PVN (mpPVN), expresses a 41-amino-acid peptide called CRH, which has been identified by Vale, Spiess, Rivier, and Rivier (1981), and which has recently been found to be involved in anxiety-related behaviour (Timpl et al., 1998; Radulovic, Ruehmann, Liepold, & Spiess, 1999). CRH is also thought to play a major role in depression and is a focus of attention in the development of new anti-depressants (Holsboer, 2001). Through a projection of the CRH neurons in the mpPVN of the hypothalamus to the external layer of the median eminence, peptides, including both CRH and arginine vasopressin (AVP), are secreted into the portal bloodstream via which they travel to the anterior pituitary. Corticotropes are specialised pituitary cells that only respond to CRH and AVP. The CRH and AVP receptors of corticotropes belong to the superfamily of Gprotein coupled receptors. CRH and AVP have a synergistic effect on corticotropes as they activate different signal transduction pathways. When stimulated by AVP and

8

Neuronal activation from various brain sites (see figure 1.1.6.1)

throughout brain

ypothalamus (PVN)

(-) (blood-brain barrier)

ituitary

(-) ACTH

drenal cortex

GLUCOCORTICOIDS

CRH (& AVP)

throughout body

Figure 1.1.2.1. Simple schema of the HPA axis.

Abbreviations and symbols: ACTH = adrenocorticotropic hormone; AVP = arginine vasopressin; CRH = corticotropin- releasing hormone; PVN = paraventricular nucleus, (-) indicates negative feedback

9

CRH, corticotropes synthesise and release ACTH. ACTH’s precursor, from which it is synthesised, is proopiomelanocortin (POMC). POMC is also the precursor of βendorphin and was the first mammalian endocrine- or neuronal precursor to be successfully cloned. ACTH is derived from the middle region of POMC by activity of multiple processing enzymes in the anterior pituitary corticotropes, whereas βendorphin derives from the C-terminal region. The function of the N-terminal region of POMC is yet unknown (Akil et al., 1998). After ACTH is released, it reaches the adrenal cortex through the systemic circulation, there to bind to and activate its receptors. ACTH receptor activation leads to the synthesis of glucocorticoids from cholesterol through a number of enzymatic processes. The term glucocorticoids refers to a group of various adrenocortical hormones (corticosteroids) such as hydrocortisone (cortisol) and corticosterone, the primary glucocorticoids in humans and rats respectively, as well as cortisone. They are then rapidly secreted into the blood stream. At resting level, glucocorticoids are not stored by the adrenal cortical cells.

1.1.3 Diurnal glucocorticoid secretion rhythm Glucocorticoids are not only secreted in acute stress situations. There is a baseline secretion rhythm with different oscillations across the day which differ between species, often depending on whether they are diurnal or nocturnal. Hence, this rhythm depends on activity rather than on light-dark constellations. Humans (diurnal) show their peak levels of glucocorticoids at about 7 am with a steady decline thereafter, until the nadir is reached in the late evening and secretion starts rising again. In contrast, nocturnal animals like rats show a peak in the evening with the nadir in the morning. Consistent with glucocorticoid secretion rhythms, circulating

10

ACTH concentration shows a similar pattern, preceding the glucocorticoid rhythm by about 1-2 h while the CRH mRNA activity pattern in the parvicellular PVN of the hypothalamus precedes it by several hours, with CRH mRNA levels starting to rise when glucocorticoid levels are low, and to drop when those levels are high. This suggests a negative-feedback loop between glucocorticoids and CRH mRNA activity. However, even though removal of the adrenal cortices, adrenalectomy (ADX), increases general CRH mRNA levels, it does not change the diurnal pattern and leaves the characteristic afternoon drop of CRH levels in rats intact (Kwak, Young, Morano, Watson, & Akil, 1993), indicating further steroid-independent mechanisms. Krieger and Hauser (1978) dissociated the light-dark phase shift from a concomitant shift of time-of-eating, and found that the latter was a more potent synchroniser of the phase of plasma corticosteroid levels. The diurnal rhythm not only affects baseline glucocorticoid levels but also the responsiveness and sensitivity to stress. The HPA-axis is particularly sensitive to stressful stimuli at the nadir of the rhythm (Dallman et al., 1991). Glucocorticoids also alter the glucose metabolism and affect energy use. In humans, as well as in rats, peak levels of glucocorticoids occur just before activity increases and food intake takes place while they are lowest at resting levels. This is also when the individual is most sensitive to stress. It is important therefore to evaluate the impact of a stressor not only by its absolute magnitude but also within the context and timing it occurs.

1.1.4 Negative feedback from glucocorticoids Glucocorticoids play a role in the negative-feedback regulation of the HPA axis. ADX prevents glucocorticoid secretion and also removes any negative-feedback effects. CRH and AVP mRNA levels in the mpPVN increase, leading to increased

11

CRH and AVP release, and enhanced ACTH secretion. The fact that CRH and AVP mRNA levels remain normal in neighbouring magnocellular neurons demonstrate the target specificity of the glucocorticoid feedback loop. Exogenous corticosteroid treatment, e.g. with the synthetic glucocorticoid dexamethasone (DEX), reverses the effects of ADX, and, in intact animals, leads to a decrease of pituitary POMC and ACTH and parvicellular AVP and CRH (Akil et al., 1998).

1.1.5 Corticosteroid receptor types and the actions of glucocorticoids in the brain While the adrenomedullary catecholaminergic component of a stress response can be triggered within seconds, the glucocorticoid response can be regarded as its second wave, taking effect minutes after secretion and exhibiting a great time range of action. There are two types of corticosteroid receptors (Veldhuis, van Koppen, van Ittersum, de Kloet, 1982): the mineralocorticoid receptor (MR or Type I receptor) and the glucocorticoid receptor (GR or Type II receptor). These are intracellular receptors that mediate slow genomic actions. When a glucocorticoid hormone or agonist binds to the receptor, a conformational change in the receptor is induced and leads to the dissociation of the receptor from its attached protein. Specific nuclear translocation signal activity leads to a dimerisation of the receptor complex and the receptor dimer binds to the so-called corticosteroid responsive element of the DNA which initiates transcription. This affects mRNA translation to certain proteins and eventually results in steroid-induced alteration (Lupien & McEwen, 1997; Sandi, 1998). Both MRs and GRs have been cloned and were found to be the product of different genes, despite sharing some homology in the DNA binding domain. In addition to the slow genomic intracellular actions of corticosteroids, rapid steroid effects have also been reported, implicating membrane- rather than

12

intracellular receptors (Schumacher, 1990). It has been postulated that corticosteroids are metabolised during uptake into the cell and interact with membrane-associated receptor proteins, leading to a modulation of membrane characteristics or transmitter response, and giving rise to rapid corticosteroid actions. Binding to a putative membrane corticosteroid receptor has also been suggested (Puia et al., 1990), and such a receptor has indeed been identified in amphibians (Orchinik, Murray, & Moore, 1991). MRs are more heterogeneously expressed in the brain than GRs. Highest expression levels are found in the limbic system and certain brainstem motor nuclei. GRs are widely expressed in most brain regions, e.g. in the PVN and other hypothalamic nuclei, the limbic system including the hippocampus, the cerebral cortex and most brainstem monoaminergic nuclei (Lupien & McEwen, 1997). Within the amygdala, both MRs and GRs are highly expressed in the central (CEA) and medial (MEA) nuclei of the amygdala. The basolateral nucleus of the amygdala (BLA) possesses a moderate density of GRs, while the lateral nucleus of the amygdala (LA) has a only a very low density of these receptors (Roozendaal & McGaugh, 1996a; de Kloet, 1991). Corticosteroid receptors have different affinities to agonist compounds that elicit glucocorticoid responses. MRs have a high affinity for endogenous glucocorticoids (corticosterone in rats, cortisol in humans) and the mineralocorticoid aldosterone (recognised in MRs of the kidneys). GRs have a lower affinity to aldosterone, as well as to endogenous glucocorticoids (Kd of GRs for corticosterone: ~ 2.5 – 5 nM; Kd of MRs for corticosterone: ~ 0.5 nM), but a higher affinity to the synthetic glucocorticoids DEX and RU28362 (Reul & de Kloet, 1985). MRs are often regarded as low-threshold receptors with a high affinity and low capacity (Reul et al.,

13

2000). In the brain, they are heavily occupied by basal levels of endogenous glucocorticoids so that acute stress and enhanced glucocorticoid levels occupy the lower-affinity GRs (McEwen & Sapolsky, 1995). For example, hippocampal MR occupancy is greater than 70% whereas GR occupancy is only about 10%, but this can rise during stress or the circadian peak to levels of up to about 90% (Sandi, 1998). Synthetic glucocorticoids like DEX can therefore mimic an acute stress response. There are other steroids that bind to these receptors without eliciting glucocorticoid responses. These antagonistic compounds compete with agonists for binding-sites and therefore block the agonistic response. A number of corticosteroid antagonists have been synthesised and show specificity for the two receptors. MR antagonists include spironolactone and the more potent RU26752 and RU28318. GRs, in contrast, have a high affinity to the antagonist RU38486 (Lupien and McEwen, 1997). The level of circulating glucocorticoids influences their binding to the corticosteroid receptors. Low or zero-levels (following ADX) of circulating glucocorticoids enhance binding and up-regulate the receptor affinity and population (Reul, Pearce, Funder, Krozowski, 1989). High circulating levels, in contrast, result in a down-regulation of corticosteroid receptors. Endogenous and exogenous glucocorticoids regulate MRs and GRs differentially: high levels of exogenous GR agonists or corticosterone lead to a down regulation of GRs but increase MR capacity population (Reul et al., 1989). High levels of mineralocorticoids (e.g. aldosterone) will down-regulate both types of receptors, whereas MR antagonists (e.g. spironolactone) will up-regulate both.

14

1.1.6 Stress-induced neuronal activation of the HPA axis In addition to the regular pattern of baseline glucocorticoid secretion, the HPA axis is activated by acute stress, e.g. in emergency situations. As mentioned above, the PVN of the hypothalamus mediates the downstream activation of the axis. However, whether a stress response occurs is determined by neuronal input to the PVN from different loci of the brain. These inputs can be broadly divided into five different categories: (1) Input from the brainstem is mainly catecholaminergic and transmits visceral information. (2) Somatic and special sensory information is transmitted from midbrain and pons. (3) Cognitive, emotional and affective stressors reach the PVN via the limbic system. Based on anatomical studies, it is thought that the prefrontal cortex (PFC), septum, amygdala and hippocampus all lack significant direct projections to the PVN. The bed nucleus of the stria terminalis and several sites in the hypothalamus are believed to be relay areas for limbic inputs to the PVN. (4) Intrahypothalamic projections reach the PVN to convey motivational information or to integrate other stress-specific signals from other input areas. (5) Blood-borne chemosensory information is transmitted from circumventricular organs such as the subfornical organ and the vascular organ of the lamina terminalis (Akil et al., 1998). Figure 1.1.1.6 outlines the neuronal input into the PVN schematically. It is not yet clear what the precise functional roles of the different pathways are. Some pathways might be activated by many different stressors and aid labelling an event as stressful. It also remains to be clarified whether there is a hierarchical structure among the different afferents to the PVN, e.g. visceral catecholaminergic input preceding intrahypothalamic motivational signals, and how the severity of a stressor and the degree it causes deviation from the equilibrium, may be conveyed.

15

Cortex Thalamus

(3) Cognitive and emotional information (2) Somatic & specialized sensory information

PFC

Midbrain periaqueductal gray

Caudal thalamus

peripenducular nuclei

Septum

Amygdala

dorsomedial hypothalamus

Hippocampus GABA

numerous forebrain areas

Pontine central gray

Midbrain

Hypothalamus

Limbic system

GABA Sensory stimuli

posterior intralaminar nuclei

Several hypothalamic sites

Bed nucleus of stria terminalis

several hypothalamic areas adjacent to PVN arcuate nucleus

Beta-Endorphin & ACTH / NPY & DA

Auditory Stimuli

(4) Motivational & integrative information from other systems

B6, B7, B8 of raphe nucleus 5HT (somatic & sensory information)

PVN (5) Blood-borne chemosensory information

vis./aud. vis./aud.

Lateraldorsal tegmental nucleus

soma Spinal Chord

soma

cholinergic (vis/aud/somatosens. stimuli)

AVP & CRH

Pedunculopontine nucleus NA A2 cells of caudal NTS

Brainstem

GABA & Angiotensin II

A1 cells of ventrolateral medulla Locus ceruleus C1 cells of ventrolateral medulla

Anterior Pituitary (Corticotropes)

GABA & Angiotensin II

Subfornical organ

Organ of lamina terminalis

Circumventricular organs

ACTH Adrenaline

Blood-borne signals (e.g. Angiotensin II)

Adrenal Cortex

GLUCOCORTICOIDS

C2 cells of NTS C3 cells of dorsomedial medulla

HPA-AXIS

(1) Visceral & sensory information

Figure 1.1.6.1. Schema of the neuronal input into the PVN

16

It is known however that some areas of the brain (e.g. cortex, PVN) are activated by most stressors (e.g. noise, pain, swimming) while others are selectively activated to encode the sensory modality involved (e.g. auditory pathways during the startle response) or the subtle aspects of a stressor (e.g. the dominant or subordinate status of an animal during an aggressive encounter) (Akil et al., 1998). The most relevant type of stress-induced HPA activation with respect to this review is the emotional one, which is thought to be mediated by the limbic system. Specifically, it has been found that the CEA seems to be crucial in mediating the HPA activation. Lesions of the CEA attenuated the neuroendocrine response, as measured by plasma corticosterone and prolactin levels, to both an aversive unconditioned stimulus (US), i.e. a single footshock, (Roozendaal, Koolhaas, & Bohus, 1991) and an aversive conditioned stimulus (CS) (Roozendaal, Koolhaas, & Bohus, 1992). The former study also revealed an attenuation of behavioural (immobility behaviour or freezing) and cardiovascular (interbeat interval) responses after the US. Recently however, Dayas, Buller, and Day (1999) proposed that the MEA rather than the CEA is involved in the neuroendocrine response to an unconditioned emotional stressor. Using brief restraint, amygdala lesions, c-fos expression and retrogradetracing experiments, they found that the stressor elicited more c-fos expression in the MEA than in the CEA. Furthermore, lesions of the MEA but not the CEA suppressed HPA activation as measured by c-fos expression in the tuberoinfundibular CRH cells of the mpPVN, the apex of HPA axis, as well as in oxytocinergic cells of the supraoptic and paraventricular nuclei, which, in rodents, commonly accompany HPA activation. Retrograde-tracing techniques were used to find that even though there are direct neural connections between the MEA and the PVN, their number is very small. The authors supported the idea of an indirect route between these two sites that

17

includes the bed nucleus of the stria terminalis as a relay station. They also pointed out that the reason that other studies found the CEA to be a crucial site (e.g. Roozendaal et al., 1991; Roozendaal et al., 1992) could be the use of excitotoxic lesions which might have interrupted cells of the MEA. It remains to be seen if the MEA is also involved in mediating conditioned emotional stressors.

1.1.7 Individual differences in stress levels of glucocorticoids The magnitude and pattern of a stress response with its release of glucocorticoids has been found to differ considerably between individuals of a population. Piazza et al. (1991) divided a population of rats into so-called high (HR) and low (LR) responders, according to their locomotor activity in a novel environment. This measure proved to be a marker for a number of individual differences. HR were found to show greater levels of plasma corticosterone in response to a stressor such as novelty, elevated for a longer period of time, than LR. This finding was confirmed by Kabbaj, Devine, Savage, and Akil (2000), employing a light/dark anxiety test or restraint as stressor. Similarly, Cools, Rots, Ellenbroek, and de

Kloet

(1993)

reported

intrastrain-differences

in

neuroendocrine

stress

responsiveness, measured as plasma ACTH levels in response to novelty or a conditioned emotional stimulus. There are other differences of both physiological and behavioural nature (Cools & Gingras, 1998; Kabbaj et al., 2000; for reviews see Cools et al., 1993; Dellu, Piazza, Mayo, Le Moal, & Simon, 1996a; Piazza & Le Moal, 1996), for example in the concentration of mesolimbic dopamine (DA) in response to stress (Rouge-Pont, Deroche, Le Moal, & Piazza, 1998; Rouge-Pont et al., 1993), in the susceptibility to apomorphine effects on gnawing (Cools et al., 1993), in the exploration of a Y-maze (Dellu, Mayo, Piazza, Le Moal, & Simon, 1993), and in self-

18

administration of amphetamine (Piazza, Deminiere, Le Moal, & Simon, 1989; Piazza et al., 1990). While the magnitude of a corticosterone response to stress generally increases with age, the individual differences between HR and LR persist (Dellu et al., 1996b). Such individual differences in stress reactivity have been implicated in the development of certain human personality characteristics, such as the trait of sensation-seeking, and psychopathological conditions, such as drug addiction or PTSD.

1.2 THE EFFECTS OF GLUCOCORTICOIDS ON MEMORY

The effects of glucocorticoids on memory have been investigated in numerous studies, both in animals and humans, and a number of recent reviews have been dedicated to this topic (Belanoff, Gross, Yager, & Schatzberg, 2001; de Kloet, Oitzl, & Joëls, 1999; Korte, 2001; Lupien & McEwen, 1997; McEwen & Sapolsky, 1995; Roozendaal, 2000; Rose, 1995; Sandi, 1998). The studies usually involve the administration of either natural glucocorticoids or synthetic MR- or GR- agonists or antagonists, given either before or after training, or before test. Another common manipulation in animals is ADX, preventing the synthesis and release of glucocorticoids. In short-term ADX, surgery is carried out relatively shortly before training, in order to prevent cell death in the dentate gyrus (DG) of the hippocampus, a long-term consequence of ADX. Sometimes, ADX and pharmacological manipulations are combined. The results have often appeared contradictory and there is a lot of confusion about whether or not glucocorticoids affect memory in a beneficial or detrimental way. In order to overcome this confusion, it is important to

19

clarify exactly what questions are being asked. Two important distinctions need to be made. Is the aim to investigate the effects of an acute dose of glucocorticoids, as it would occur in a particular emotional or stressful situation, or are chronic effects being studied, mimicking the occurrence of constant stress? Secondly, it is important to consider what exact process is being affected in the study. For this purpose, it is useful to divide the construct of learning and memory into three phases: the acquisition phase during which the learning event is experienced, the consolidation phase during which information is stored (into long-term memory), and finally the retrieval phase during which a learned event influences subsequent behaviour. This PhD project is concerned only with the acute effects of glucocorticoids on the consolidation of memory. Hence, in the following review of the existing literature, an emphasis is put on those studies investigating acute effects on consolidation, although sometimes they cannot be dissociated from effects on acquisition. One section is dedicated to animal studies (Section 1.2.1) and another one to studies employing human subjects (Section 1.2.2). Further, in order to convey the complex nature of the relationship between glucocorticoids and memory more clearly, and to elucidate why the literature often seems controversial, some studies concerned with chronic effects (Section 1.2.3) or effects on memory retrieval (Section 1.2.4), as well electrophysiological studies (Section 1.2.5), are briefly discussed in extra sections. The state of existing literature is briefly summarised in Section 1.2.6, addressing some of its limitations and setting up a framework for the experimental chapters to follow. An overview of the most important experiments investigating the acute modulation of memory consolidation by glucocorticoids can also be found in Table 2.0.1 of Chapter 2.

20

1.2.1 The acute effects of glucocorticoid on memory acquisition and consolidation (animal studies) Due to the extensive animal literature on the acute effects of glucocorticoids on memory, the current section is divided into a number of subsections. It begins with an overview of the original studies in the field, going back to the 1970s, which established the role of glucocorticoids in memory modulation (Section 1.2.1.1). The next section refers to the very few studies that have examined the memory modulatory effect of glucocorticoids in appetitive procedures (Section 1.2.1.2). A large part of the literature is concerned with studies employing the administration of selective MRand GR- agonists and antagonists, or glucocorticoid synthesis inhibitors. These studies are discussed in Section 1.2.1.3. A number of studies that examined the effects of glucocorticoids on memory consolidation in the light of different training intensities is discussed next (Section 1.2.1.4). A specific question that is particularly important to the work in this thesis concerns glucocorticoid effects on contextual vs. discrete-cue conditioning (Section 1.2.1.5). An extensive section (1.2.1.6) is dedicated to those studies that aimed to elucidate the neural bases of glucocorticoid memory modulation, ranging from work on the involvement of different brain regions, to studies concerned with the mechanisms of how glucocorticoids affect memory, proposing particular interactions with neurotransmitter systems, as well as direct molecular effects on protein synthesis. Finally, the last section (1.2.1.7) discusses those studies that manipulated glucocorticoid levels by stressor exposure rather than pharmacologically.

21

1.2.1.1 Early studies Following some earlier studies investigating the effects of glucocorticoids on the extinction of avoidance behaviour (Bohus & Lissak, 1968; de Wied, 1967; van Wimersma Greidanus, 1970; see Chapter 3) and some null-result-studies, in most of which glucocorticoid treatment was given several hours prior to acquisition trials (Beatty, Beatty, Bowman, & Gilchrist, 1970; Gold & van Buskirk, 1976; Hennessy, Cohen, & Rosen, 1973; Garrud, Gray, Rickwood, & Coen, 1977), the first effects on memory consolidation were reported in the late 1970s. Using rats in a passive avoidance task, Kovacs, Telegdy, and Lissak (1977) found that corticosterone affects memory in an inverted U-shaped fashion. While the lowest dose facilitated passive avoidance, an intermediate dose had no effect and the highest dose led to an impairment. The effect was thought to be mediated, at least partly, by changed serotonergic metabolism, as serotonin (5-HT) levels were affected by corticosterone in the same way as behaviour. In this study, however, corticosterone was administered prior to both the conditioning session and the retention test, so that no conclusions about the role of corticosterone on memory consolidation per se can be drawn. The same goes for a study by Mormede and Dantzer (1977) who found a facilitation of fear conditioning in pigs when administering DEX 2 hr before conditioning or test. The first study that investigated effects on memory consolidation more explicitly was carried out by Cottrell and Nakajima (1977) and found that post-training hydrocortisone reversed the memory impairing effects of pre-training administration of the protein synthesis inhibitor cycloheximide. It was also the first study to investigate in more detail the neural substrates of glucocorticoid effects on memory, as the results were achieved both with subcutaneous (sc) and intrahippocampal administrations. Using both active and passive avoidance tasks in mice and a post-

22

training administration procedure, Flood et al. (1978) discovered that different glucocorticoids, both natural (corticosterone, hydrocortisone) and synthetic (DEX), enhance memory in mice in a retrograde manner. Effects were both dose- and timedependent, with DEX enhancing memory when administered up to 150 min after training. In mice treated with the protein synthesis inhibitor anisomycin, glucocorticoids had an anti-amnesic effect. The authors suggested that glucocorticoids “…may modulate arousal and thus provide an optimal level of CNS [central nervous system] excitability for long-term memory trace formation” (Flood et al., 1978, p. 87). Coover, Sutton, Welle, and Hart (1978) carried out a study primarily concerned with state-dependency effects of DEX. However, a superior performance in Pavlovian conditioning of a control group that received post-training DEX, as compared to pretraining DEX, could also be interpreted in terms of a memory modulatory role of glucocorticoids.

1.2.1.2 Studies employing appetitive experimental procedures The first and to-date only investigations of glucocorticoid memory modulation using appetitive procedures were carried out by Micheau and colleagues (Micheau, Destrade, & Soumireau-Mourat, 1981, 1985). They found that intracerebroventricular (icv) and intrahippocampal post-training infusions of corticosterone enhanced retention performance in an operant successive-discrimination procedure in mice. In this paradigm, an operant response (lever pressing) was reinforced only during presentations of the positive stimulus, a combination of a light and a buzzer, but not the negative stimulus, diffuse illumination and white noise. The effect was timedependent and only occurred if administration took place within 6 hr of training. The authors did not find an effect in an operant continuous reinforcement task.

23

1.2.1.3 Studies employing selective MR- and GR- agonists and antagonists, or glucocorticoid synthesis inhibitors Using the Porsolt forced-swimming task in which retention of acquired immobility is tested, Veldhuis, Korte, and de Kloet (1985) found that DEX, as well as the specific GR agonist RU28362, fully restored ADX-impaired retention performance while the mineralocorticoids aldosterone and progesterone did not. Corticosterone was only effective in a 500 times higher dose than the GR agonists. Drugs were administered sc 15 min after the end of the initial swimming exposure, 24 hr before the retention test. They also found that the specific GR antagonist RU38486 impaired retention in intact rats. The results point to a specific role of the GR system in the consolidation of acquired immobility. Similar data have been reported by Jefferys and colleagues (Jefferys, Copolov, Irby, & Funder, 1983; Jefferys & Funder, 1987). Mitchell and Meaney (1991), however, only found a partial reversal of ADXinduced impairment of acquired immobility with post-training corticosterone treatment. A complete reversal was only achieved when corticosterone was also present at test, suggesting that glucocorticoids play a role not only in the consolidation but also in the retrieval or expression of acquired immobility. However, the dose may not have been sufficient, and given that corticosterone treatment occurred prior to training, a state-dependency effect cannot be ruled out. Further evidence for a role of glucocorticoids in the expression of behavioural inhibition comes from a study by Corodimas, LeDoux, Gold, and Schulkin (1994). They found that high doses of corticosterone, in form of implanted hormone pellets, enhanced expression of conditioned freezing, compared to low doses and placebo. Oitzl and de Kloet (1992) investigated the role of glucocorticoids in learning and memory specifically with respect to the two different receptor types, MR and GR.

24

First, they examined the effects of ADX on spatial memory using the Morris water maze. Short-term ADX impaired memory, whereas selective removal of the adrenal medulla did not produce any deficits, suggesting a glucocorticoid-regulated function. ADX rats did not differ from controls when escaping onto a visible (cued) rather than a hidden platform, ruling out possible sensory or motor effects. Oitzl and de Kloet (1992) then went on to directly investigate the effects of both MR- (spironolactone) and GR- (RU38486) antagonists at different stages of the learning paradigm, in order to differentially evaluate the function of the two receptors in spatial memory. They administered the drugs either before the first training session to assess the effects on acquisition, directly after the first training session to assess the effects on consolidation, or before the second training session to assess the effects on retrieval. While the MR antagonist had no effect on the latency to find the platform when administered at any of the three time points, the GR antagonist impaired performance when given directly before or after the first training session, leading to the assumption that GRs play a role in the memory consolidation process. The MR antagonist, however, seemed to have led to a change in escape strategy. On the second and third test session, control and GR antagonist animals exhibited a practice effect, whereas animals that received the MR antagonist before retrieval neither improved nor deteriorated in their performance, but spent more time swimming around the maze. Oitzl and de Kloet (1992) concluded that MRs might be involved in processes such as situation evaluation, (appropriate) response selection, and sensory integration. In another study by Oitzl, Fluttert, Sutanto and de Kloet (1998b), the effects of continuous and phasic GR blockade by icv administration of RU38486 in different doses were compared. Phasically treated animals received the drug prior to training over three consecutive days, whereas continuously treated animals were given an

25

infusion each hour over 10 days. It was found that phasic administration impaired performance in a dose-dependent way, whereas the high dose of continuous administration facilitated performance. However, it should be noted here that the GR antagonist in the phasic group was also present during retention test. As the drug was an antagonist, the results support the hypothesis that acute stress can enhance cognitive function while chronic stress can impair it (but see Oitzl, Fluttert, & de Kloet, 1998a). In line with the idea that GRs but not MRs modulate memory consolidation, Conrad, Lupien, and McEwen (1999) found that GR- but not MRagonists and antagonists influenced spatial memory performance in the Y-maze. They proposed an inverted U-shaped relationship between corticosterone and cognitive performance, depending on GRs only, a claim that was supported by their finding that GR occupancy in ADX rats was significantly linked to spatial memory performance, following a non-linear inverted U-shaped function. MR occupancy did not correlate with performance (Conrad, Lupien, & McEwen, 1996). Sandi and Rose (1994a) found that both MR (RU28318) and GR (RU38486) antagonists, administered intracerebrally (ic) shortly before training, impaired memory in a passive avoidance task in one-day old chicks when tested 4 hr and 24 hr after training. Young chicks have the tendency to peck spontaneously at small salient objects, for example bright-coloured beads. A bead of a particular colour was coated in methylanthranilate, an aversive bitter-tasting substance, and learning was reflected by avoidance of this bead subsequent to a single taste-experience with it. For technical reasons, drugs were administered pre-training so that the effects could have been either on acquisition or consolidation. However, as no impairment was observed when retention was tested 30 min after training (in contrast to tests at 4 hr and 24 hr), an effect on consolidation was suggested. Administration of the GR antagonist resulted

26

in U-shaped effects, with memory being impaired at the intermediate dose but not the low and high doses. This is in line with the inverted U-shaped effects reported with GR agonists (e.g. Conrad et al., 1996). In contrast, only the highest dose of the MR antagonist was effective. State-dependency effects were ruled out as the results did not change when the drug was present during the retention test. Like Oitzl and de Kloet (1992), Sandi and Rose (1994a) suggested that MRs and GRs may influence different aspects of learning and memory. Since the MR but not the GR antagonist increased responses to novelty, Sandi and Rose (1994a) proposed an MR function of regulating reactivity to non-specific aspects of training, and a GR function of memory consolidation. In another study by same group (Sandi & Rose, 1994b), again using young chicks in a weak version of the passive avoidance task, it was found that corticosterone administered ic either 15 min before, or 5, 30, or 60 min after training, enhanced retention as tested 24 hr later. The effect was dose-dependent such that only the medium dose of corticosterone was effective, supporting the previously observed inverted U-shaped function (Kovacs et al., 1977). The GR antagonist RU38486 but not the MR antagonist RU28318 blocked the effects of corticosterone when administered concurrently 30 min after training, suggesting that GRs but not MRs are involved in the modulation of memory consolidation. The protein synthesis inhibitor anisomycin only partly attenuated the effects of corticosterone, suggesting that other cellular processes mediate the effects as well. Further support for a role of glucocorticoids in the modulation of memory comes from studies using corticosteroid synthesis inhibitors. Loscertales, Rose, and Sandi (1997) found that such inhibitors, either suppressing basal (aminoglutethimide) or stress-induced (metyrapone) levels of corticosterone, impaired passive avoidance learning in young chicks. Similarly, Roozendaal, Bohus, and McGaugh (1996a) found

27

that metyrapone impaired spatial learning of rats in the water maze. This effect was blocked by the synthetic GR agonist DEX but not by an equivalent dose of corticosterone.

The overall conclusion of the work discussed in this subsection is that the modulation of memory consolidation by glucocorticoids seems to be mediated selectively by GRs, whereas MRs seem to play a different role in learning, primarily concerned with strategy- and response selection.

1.2.1.4 Studies investigating interactions between glucocorticoid and training intensity levels Sandi and colleagues (Cordero, Merino, & Sandi, 1998; Cordero & Sandi, 1998; Sandi, Loscertales, & Guaza, 1997; Sandi & Rose, 1997) linked the effects of glucocorticoids on memory consolidation to the learning- or training experience. Sandi and Rose (1997) found that only the strong version of the passive avoidance task in young chicks, using undiluted methylanthranilate as the aversive reinforcer, resulted in increased plasma corticosterone levels. In a weak version of the task, using a 10% methylanthranilate solution, levels were not increased. Exogenous corticosterone enhanced memory in the weak and impaired memory in the strong version of the task. This biphasic action of glucocorticoids on memory consolidation supports the idea of an inverted U-shaped relationship (Kovacs et al., 1977) and such concentration-dependent effects of glucocorticoids have also been reported in electrophysiological studies (e.g. Bennett, Diamond, Fleshner, & Rose, 1991; Diamond, Fleshner, Bennett, & Rose, 1992). Whether the deleterious effects of high doses of glucocorticoids rely on the same mechanisms as the enhancing effects, for

28

example glycoprotein synthesis (see Section 1.2.6), is not yet known. The interaction between the role of the training experience itself and effects of glucocorticoids on memory consolidation have also been investigated in rats. Sandi et al. (1997) found that rats performed better in a spatial-learning Morris water-maze task if the water temperature was colder and the task presumably more aversive. Post-training corticosterone levels were greater when the water was 19°C rather than 25°C, and exogenous corticosterone only facilitated the less aversive version of the task. Similarly, Cordero et al. (1998), using a contextual fear-conditioning procedure in rats, showed that there was a correlational relationship between the intensity of footshock, levels of plasma corticosterone and the extent of conditioned fear expressed by freezing. This was the case directly after the conditioning session, as well as during retention tests 24 hr and 7 days afterwards. Higher shock intensities therefore seem to produce more freezing behaviour and higher corticosterone levels. Corticosterone could therefore be the modulator between the other two measures, supporting the assumption that glucocorticoids play an important role in the memory consolidation process. The fact that the positive correlation between freezing and corticosterone levels was still present during re-exposure to the context (extinction) supports the idea that these hormones might play a role in the long-term expression of freezing also. In a further contextual fear conditioning study, Cordero and Sandi (1998) found that pre-training administration of a GR- (RU38486) but not an MR(RU28318) antagonist impaired conditioned freezing in a later retention test. This was, however, only the case with an intermediate (0.4 mA) but not with a high (1.0 mA) shock intensity. Post-training corticosterone enhanced conditioned freezing. While the results support the idea of GR-mediated memory-modulatory effects of glucocorticoids, the dependence on shock intensity for an effect of GR blockade could

29

be interpreted in a number of ways. Cordero and Sandi (1998) speculated that the dose may not have been sufficient to result in an impairment but favoured the explanation that more intense training conditions would be more resistant to experimental manipulations at a specific physiological level due to possible additive effects of different physiological systems mediating such aversive experiences. One might also note that different, seemingly contradictory, results have been obtained in other studies of the effects of acute GR blockade (Oitzl et al., 1998a,b), suggesting complex interactions of a number of factors such as training intensity, context, timeof-administration and neural site-of-action. 1.2.1.5 Studies comparing effects on discrete-cue and contextual Pavlovian conditioning paradigms Pugh and colleagues (Pugh, Fleshner, & Rudy, 1997a; Pugh, Tremblay, Fleshner, & Rudy, 1997b) investigated the role of glucocorticoids in auditory-cue and contextual fear conditioning in juvenile rats. It has been shown that contextual but not auditory-cue conditioning depends on an intact hippocampus (Phillips & LeDoux, 1992, 1994; Selden, Everitt, Jarrard, & Robbins, 1991), an area rich in GRs (Reul & de Kloet, 1985). In one study, Pugh et al. (1997a) found that pre- or post-training administration of the GR antagonists RU38486 and RU40555 impaired contextual fear conditioning while leaving auditory-cue fear conditioning intact. The authors concluded that glucocorticoids contribute to the consolidation of some aspects of contextual fear conditioning. The authors acknowledged the study’s limitation in using juvenile rats where the results may not necessarily generalise to adult rats. Pugh et al. (1997b) investigated this issue further. In line with their first study, they found that short-term ADX impaired contextual but not auditory-cue fear conditioning. In support of a role in memory consolidation, this was only found in tests of long-term

30

but not short-term memory, i.e. when they took place 24 hr rather than immediately after training; and the effect could be blocked by corticosterone replacement. Corticosterone replacement restored performance even when it was given after conditioning, providing further support for a role of glucocorticoids in one of the two processes contextual fear conditioning is hypothesised to depend on (Fanselow, 1990; Rudy & Pugh, 1996; Kiernan & Westbrook, 1993; Young, Brohenek, & Fanselow, 1994), either the construction and consolidation of a representation of the context, or the association of this representation with the aversive US. The fact that preexposure to the context 24 hr before ADX reversed its impairing effects led Pugh et al. (1997b) to favour the former process, suggesting that glucocorticoids are not involved in the association process. In line with other studies (Kovacs et al., 1977; Sandi & Rose, 1997), Pugh et al. (1997b) finally observed an inverted U-shape relationship between post-training corticosterone administration and contextual fear conditioning in ADX rats, with freezing to context being lower when no or a high dose of corticosterone was administered. From these results, Pugh et al. (1997b) suggested that the role of glucocorticoids in fear conditioning might be selective for the construction and consolidation of a representation of the context. Since parallel results were found with hippocampal lesions (Phillips & LeDoux, 1992, 1994; Selden et al., 1991) and since the hippocampus is rich in GRs (Reul & de Kloet, 1985), the authors suggested that the hippocampus was likely to be the site-of-action of glucocorticoids in contextual fear conditioning. In a further study of this group, Fleshner, Pugh, Tremblay, & Rudy (1997) treated rats prior to training over 5 days with the neurosteroid dehydroepiandrosterone sulfate (DHEA-S), which also elevates circulating levels of the adrenal steroid dehydroepiandrosterone (DHEA). DHEA, which is thought to act as a functional antiglucocorticoid (Kalimi et al., 1994), like ADX impaired long-term

31

contextual fear conditioning but had no effect on auditory-cue or short-term contextual fear conditioning. However, when administered acutely pre- and posttraining, DHEA-S enhanced retention in a passive avoidance task by exerting a central effect, probably the facilitation of NMDA-mediated glutamatergic neurotransmission (Reddy & Kulkarni, 1998).

1.2.1.6 Studies investigating glucocorticoid memory modulation on an anatomical, system interaction-, and molecular level A great number of studies investigating the neural loci of glucocorticoid memory modulation have been carried out by Roozendaal and colleagues (for a review see Roozendaal, 2000). In summary, they have proposed a model of glucocorticoid memory modulation according to which glucocorticoids dose- and time-dependently modulate memory consolidation by activating GRs in different brain areas, e.g. in noradrenergic soma, the BLA and the hippocampus. As a result, glucocorticoids facilitate stress-induced noradrenergic neurotransmission and formation of cyclic adenosine monophosphate (cAMP) / protein kinase A (PKA) in the BLA, which coordinates and influences memory storage in other brain areas, such as the hippocampus, possibly involving the nucleus accumbens (NAC) as a site of convergence, through projections via the stria terminalis (ST). Roozendaal and colleagues were primarily interested in the role of the amygdala, which has been shown to be critically involved in memory modulatory effects of emotional arousal (Cahill, 2000a,b; McGaugh, Ferry, Vazdarjanova, & Roozendaal, 2000), adrenaline, and other substances (McGaugh et al., 1996; McGaugh, 2000). For example, Roozendaal and McGaugh (1996a) found that lesions of the BLA and MEA, but not CEA, blocked the memory-enhancing effects of post-

32

training administration of the GR agonist DEX in an inhibitory (passive) avoidance task. Corticosterone did not affect memory, likely due to its low dose and its 10-fold lower affinity to GRs than DEX. Lesions of the CEA but not the BLA or MEA impaired acquisition of the task. The study not only provided further evidence that glucocorticoid memory modulation is mediated by GRs, but also suggested that the BLA and MEA constitute critical neural loci of this modulation. The study was also interesting for another reason. The fact that CEA but not BLA lesions impaired acquisition of the task opposes the prominent view that aversive memories are formed in the amygdala after CS and US information converge at the level of the LA, which forms part of the BLA (e.g. Davis, 1992; Fanselow & LeDoux, 1999; LeDoux, 1994, 1995; Maren & Fanselow, 1996). Instead, the results support the idea that the amygdala modulates memory in other brain regions (Cahill, Weinberger, Roozendaal, & McGaugh, 1999; Parent, Quirarte, Cahill, & McGaugh, 1995) and that its different nuclei serve different independent functions (Killcross, Robbins, & Everitt, 1997; Vazdarjanova & McGaugh, 1998). In another study, Roozendaal, Portillo-Marquez, and McGaugh (1996b) investigated the role of the different amygdala nuclei in a Morris water-maze task. As there was no enhanced performance from systemic administration of exogenous glucocorticoids in this stressful task (Roozendaal et al., 1996a), possibly due to high (in terms of memory consolidation optimal) levels of endogenously circulating glucocorticoids, they applied short-term ADX to one group of animals. In sham-ADX animals, lesions of the CEA (but not BLA or MEA) impaired acquisition of the task, whereas lesions of either the CEA or MEA (but not BLA) impaired retention performance. In the case of CEA lesions, this might have been due to impaired acquisition. The results are in line with the previously mentioned finding that the CEA

33

but not BLA is important for the acquisition of the inhibitory avoidance paradigm (Roozendaal and McGaugh, 1996a). While ADX impaired memory, the effect was reversed by post-training DEX but not corticosterone. Lesions of the BLA but not CEA or MEA blocked the effects of ADX and DEX, again pointing to the BLA in mediating glucocorticoid influences on memory consolidation. This view is further supported by the fact that post-training infusions of the GR agonist RU28362 into the BLA enhanced retention in a passive avoidance task, whereas treatment with the GR antagonist RU38486 into the BLA impaired memory in the Morris water maze. Infusions into the CEA were without effect in either task (Roozendaal & McGaugh, 1997a). However, since the BLA only possesses a moderate density of GRs, Roozendaal and McGaugh (1996a, 1997a) argued that in addition to GR binding in the BLA, the memory-modulatory effects of glucocorticoids could be due to GR activation of ascending aminergic fibre pathways, e.g. presynaptic noradrenergic cell groups A1-A7, or projections from the BLA to the enthorinal cortex and the DG, arguing for a BLA-based modulation of hippocampal memory. Roozendaal, Sapolsky, and McGaugh (1998) investigated the effects of BLA lesions and corticosterone supplement on long-term ADX and dissociated the cognitive impairments and the neurodegeneration of granule cells in the DG. Whereas low doses of corticosterone supplement did not block the impairments in spatial learning and memory, they blocked the degeneration of the DG. The authors suggested that the low plasma corticosterone levels were insufficient to activate GRs at the time of learning. BLA lesions, in contrast, did not prevent neurodegeneration but blocked the cognitive deficit. Therefore, the BLA is not only necessary for GR agonist-induced memory enhancement but also mediates ADX-induced memory impairment. A similar dissociation between cognitive impairment and hippocampal

34

degeneration was reported by McCormick, McNamara, Mukhopadhyay, and Kelsey (1997) who found that acute CORT treatment in the days before testing blocked the learning impairments of long-term ADX despite degeneration in the DG. Roozendaal and colleagues also investigated the involvement of other brain areas in glucocorticoid memory modulation, and their interactions with the BLA. For example, Roozendaal and McGaugh (1996b) lesioned the ST, a structure considered to be a major afferent-efferent pathway of the amygdala. They found that, as with lesions of the BLA (Roozendaal & McGaugh, 1996a; Roozendaal et al., 1996b), lesions of the ST did not impair memory in the Morris water maze nor in the inhibitory avoidance task. In sham-lesioned control animals, post-training systemic administration of the GR agonist DEX blocked the impairing effects of ADX in the water maze and enhanced memory in the inhibitory avoidance. As was the case for BLA lesions (Roozendaal & McGaugh, 1996a; Roozendaal et al., 1996b), both DEX effects were blocked by ST lesions, suggesting an integral role of the ST in glucocorticoid memory modulation. The authors pointed to findings that lesions of the ST also blocked the memory-enhancing effects of amygdala-specific treatments, such as electrical stimulation (Liang & McGaugh, 1983a) or infusions of adrenaline (Liang & McGaugh, 1983b) and the ß-adrenergic agonist clenbuterol (Introini-Collison, Miyazaki, & McGaugh, 1991), and hypothesised that the ST might mediate amygdala efferents that have memory-modulatory influence over other brain areas. Another area that has been implicated to be involved in the modulation of memory by glucocorticoids for a number of reasons is the hippocampus. It is rich in GRs; some learning tasks affected by glucocorticoids were thought to be hippocampal-dependent (e.g. Oitzl & de Kloet, 1992; Pugh & Rudy, 1997a,b); and an earlier study, which demonstrated memory-enhancing effects of glucocorticoids,

35

administered them directly into the hippocampus (Cottrell & Nakajima, 1977). In line with previous studies, Roozendaal and McGaugh (1997b) found that in an inhibitory avoidance procedure, post-training intrahippocampal infusions of a GR agonist enhanced, and pre-training infusions of a GR antagonist impaired, retention performance. These effects were blocked by lesions of the BLA but not the CEA. Given that the BLA projects to the entorhinal cortex and from there to the DG (Thomas, Assaf, & Iversen, 1984), the results give further support to the claim that the BLA mediates memory-modulatory hormonal influences on the hippocampus (Roozendaal and McGaugh, 1996a, 1997a). Roozendaal and McGaugh’s (1996a, 1997a, 1997b) ideas find support from a study by Packard, Cahill and McGaugh (1994), who found that post-training ic microinfusions of d-amphetamine into the caudate nucleus enhanced retention in a cued- but not in a spatial water-maze task, whereas the opposite pattern of results was found with hippocampal infusions. Intraamygdaloid infusions enhanced retention in both tasks, and pre-retention amygdala inactivation with lidocaine did not block any of the memory-enhancing effects. Hence, the amygdala seems to modulate memory consolidation in other brain regions but does not seem to play a direct role in retention. In line with this, Packard and McGaugh (1996) found that inactivation of the hippocampus with lidocaine blocks place learning, whereas the same treatment in the caudate nucleus blocks response learning. These studies, despite not involving glucocorticoids, show how the hippocampus and the caudate nucleus are involved in different types of memory, and, more importantly, emphasise a memory-modulatory role of the amygdala, involving different brain structures and different types of memory. The modulatory effect is time-dependent and the amygdala itself does not seem to serve as a storage site of neural plasticity, or even be needed for acquisition. In line with this are also

36

electrophysiological studies which found that electrical stimulation of the BLA enhanced the induction of population-spike long-term potentiation (LTP) in the DG (in vivo) while BLA lesions or temporary inactivations impaired it (Akirav & RichterLevin, 1999a,b; Ikegaya, Saito, & Abe, 1994, 1995; Kim, Lee, Han, & Packard, 2001; for a review see Richter-Levin & Akirav, 2001). Employing an elegant crossed-lesion design, Setlow, Roozendaal, and McGaugh (2000) investigated the role of the NAC and the BLA in glucocorticoid memory modulation. Bilateral lesions of the NAC blocked the memory-enhancing effects of post-training DEX. Ipsilateral combined lesions of the BLA and the NAC did not have this effect. Contralateral lesions of the two structures, however, blocked the DEX effects on memory, suggesting that the BLA-NAC pathway is important for glucocorticoids to modulate memory, and supporting the view that the BLA modulates memory storage in other brain regions. Given that there are projections from both the BLA and the hippocampus to the NAC (Groenewegen et al., 1991; Wright, Beijer, & Groenewegen, 1996), the authors suggested that the NAC might be a site of convergence for memory-modulatory information from the BLA and memory-mediating information from the hippocampus. In support of this view, Roozendaal, de Quervain, Ferry, Setlow, & McGaugh (2001a) found that posttraining infusions of the GR agonist RU28362 into either the BLA or the hippocampus dose-dependently enhanced retention performance in an inhibitory avoidance task. The effects were blocked by bilateral lesions of either the NAC or ST. The authors suggested that the BLA might interact with hippocampal effects on memory consolidation at the level of the NAC via a pathway passing through the ST. Roozendaal and colleagues were not only interested in the neural structures involved in glucocorticoid memory modulation, but also in the question whether

37

particular neurotransmitter systems interact with glucocorticoids in the modulation of memory. A primary candidate for such an interaction is NA, a neurotransmitter that itself is known to modulate memory in a similar fashion as glucocorticoids (Liang, McGaugh, & Yao, 1990). NA and other β-adrenergic agonists such as clenbuterol have been found to enhance retention in a dose-dependent fashion when administered into the BLA immediately after training (Ferry & McGaugh, 1999; Ferry, Roozendaal, & McGaugh, 1999a; Hatfield & McGaugh, 1999; for a review see Ferry, Roozendaal, & McGaugh, 1999b), while β-adrenergic antagonists like propranolol impair retention (Liang, Chen, & Huan, 1995; Gallagher, Kapp, Musty, & Driscoll, 1977; Introini-Collison, Saghafi, Novack, & McGaugh, 1991; Hatfield & McGaugh, 1999). Memory modulation by β-adrenergic agonists is thought to be related to an increase in cAMP. The β-adrenoceptor is directly coupled to the enzyme adenylate cyclase via the guanine-nucleotide-binding regulatory G-protein (Pfeuffer, 1977). Using microdialysis, Galvez, Mesches, & McGaugh (1996) found that a single footshock-stimulation of medium intensity (0.55 mA, 1.0 s) increases NA release in the amygdala. NA also mediates the memory-modulatory effects (Gold & van Buskirk, 1975) of the primary adrenal catecholamine adrenaline (Costa-Miserachs, Portell-Cortes, Aldavert-Vera, Torras-Garcia, & Morgado-Bernal, 1993, 1994; Gold & van Buskirk, 1976; Liang, Juler, & McGaugh, 1986; Introini-Collison et al., 1992). Adrenaline, which is rapidly released from the adrenal medulla during emotional situations, does not readily enter the brain but has been shown to increase NA levels in the brain (Gold & van Buskirk, 1978), for example in the amygdala (Williams, Men, Clayton, & Gold, 1998). The amygdala is thought to be a locus of the central noradrenergic mediation of peripheral memory-modulatory effects of adrenaline (Liang et al., 1986). Evidence for an interaction between the adrenocortical and the

38

adrenergic system comes from a study by Roozendaal, Carmi, & McGaugh (1996c) who found that the memory-enhancing effects of adrenaline were blocked by pretraining administration of the corticosterone synthesis inhibitor metyrapone. Quirarte, Roozendaal, and McGaugh (1997) investigated a possible interplay of glucocorticoids and the central noradrenergic system at the level of the BLA. They looked at the effects of either specific (β1-receptor and β2-receptor) or non-specific β-adrenergic blockade on the memory-modulating effects of glucocorticoids. Using the inhibitory avoidance paradigm, they found that the memory-enhancing effects of post-training DEX were blocked by microinfusions of either the non specific β-blocker propranolol, the specific β1-adrenergic antagonist atenolol or the β2-adrenergic antagonist zinterol into the BLA but not into the CEA. They also found that atenolol blocked the memory-enhancing effects of the specific GR agonist RU28362 when microinfused concurrently into the BLA. The results of this study suggest that an intact β-adrenergic system is required for the memory-modulatory effects of glucocorticoids, and that a locus of interaction between these two system is the BLA. Incidentally, the highest dose of RU28362 was ineffective in enhancing memory when being administered alone, supporting the aforementioned idea of an inverted Ushape function between glucocorticoid levels and memory. Since there was no memory-enhancing effect of that dose in the presence of concurrent atenolol microinfusions, the authors came to the important conclusion that β-adrenergic blockade does not simply shift the dose-response function of glucocorticoids to the right but that β-adrenergic activation in the BLA is necessary for any level of glucocorticoid memory modulation. Quirarte et al. (1997) proposed the idea that glucocorticoids bind postsynaptically in the BLA’s noradrenergic receptor system, possibly on the same neurons that express β-adrenergic receptors. Being intracellular

39

and intranuclear receptors (Joëls & de Kloet, 1994), GRs are not present presynaptically on the noradrenergic terminals of the BLA. However, GRs are activated in many brain regions after systemic administration of glucocorticoids, and they occur in high densities in the cell bodies of the noradrenergic cell bodies A1-A7. Furthermore, there are dense innervations from the locus coeruleus and the nucleus of the solitary tract (NTS). Although most of the noradrenergic terminals innervating the amygdala reach the CEA, there is also quite a high density in the BLA. It is therefore possible that glucocorticoids modulate memory by enhancing activity or efficacy of noradrenergic transmission in the BLA both pre- and postsynaptically. Support for this idea comes from two studies by Roozendaal and colleagues (Roozendaal, Nguyen, Power, & McGaugh, 1999a; Roozendaal, Williams, & McGaugh, 1999b). Regarding presynaptic interactions, they found that post-training administration of the GR agonist RU28362 into the NTS enhances retention in the inhibitory avoidance task, an effect which is blocked by atenolol-induced βadrenergic blockade in the BLA. Administration of the GR antagonist RU38486 into the NTS, while not impairing retention, shifted the memory-enhancing dose-response effects of systemic post-training DEX injections to the right (Roozendaal et al., 1999b). Investigating postsynaptic interactions further, Roozendaal et al. (1999a) found that blockade of β-adrenergic receptors in the BLA by direct infusions of atenolol also blocked the memory-enhancing effects of unilateral dorsal hippocampal infusions of the GR agonist RU28632. With unilateral atenolol infusions, the effect only occurred if atenolol was given ipsilaterally but not contralaterally to the hippocampal treatments, indicating that the BLA influence on the hippocampal memory modulation is mediated by direct neural connections rather than from peripheral stress responses resulting in BLA activation.

40

The idea that glucocorticoids influence the noradrenergic system has also been proposed by a number of other authors. McEwen (1987) found that glucocorticoids activate noradrenergic neurons in the locus coeruleus during emotional arousal. A permissive effect by glucocorticoids on the noradrenergic system has also been described by de Kloet (1991). Joëls and de Kloet (1989) reported that NA-induced excitability of hippocampal neurons can be affected by GR activation, and glucocorticoids have been found to increase the conversion of tyrosine into NA (Iuvone, Morasco, & Dunn, 1977). How exactly the influences of glucocorticoids on the noradrenergic system occur is not yet known. Both genomic and non-genomic actions have been implicated. Roozendaal (2000) hypothesised about a possible postsynaptic influence of glucocorticoids on ß-adrenoceptor activation-induced cAMP production and cAMP-dependent PKA formation via a1-adrenoceptors. This influence might occur somewhere between the membrane-bound ß-adrenoceptor and the intracellular cAMP production site, since Ferry, Roozendaal, and McGaugh (1999c) found that, as for the GR antagonist RU38486, administration of the selective a1-adrenoceptor antagonist prazosin into the BLA shifted the dose-response curve of the noradrenergic agonist clenbuterol to the right, while it did not affect the memoryenhancing effects of BLA infusions of the synthetic cAMP analog 8-bromo-cAMP. Furthermore, the memory enhancing effect of post-training RU28362 was blocked by the PKA inhibitor Rp-cAMPS but not by the a1-adrenoceptor antagonist prazosin (Roozendaal, Quirarte, & McGaugh, 2001b). Another way in which glucocorticoids might modulate memory is by affecting protein synthesis (Rose, 1995; Sandi, 1998). For example, glucocorticoids might affect second-wave glycoprotein synthesis, a process induced by corticosterone at a memory-facilitating dose in untrained chicks (Rose, 1995). Antibodies against the

41

glycoprotein neural cell adhesion molecule (NCAM) blocked the memory-facilitating effects of corticosterone (Sandi, Rose, Mileusnic, & Lancashire, 1995). The synthesis of NCAM has shown to be important in some forms of learning (e.g. Cremer et al., 1994). Furthermore, ADX leads to a reduction of kainic-acid mediated increases in mRNA levels of substances like nerve growth factor (NGF), brain-derived neurotrophic factor (BDNF), and neurotrophin-3 (NT-3), both in terms of the number of neurons producing a detectable amount of mRNA, as well as the labelling intensity of individual neurons. The synthetic GR agonist DEX can restore the attenuated effect of kainic acid on NGF, BDNF and NT-3 mRNA levels in adrenalectomised rats (Barbany & Persson, 1993). Administration of NGF has shown to improve performance in the Morris water maze (Markowska, Koliatsos, Breckler, Price, & Olton, 1994) while antibodies to NGF impair learning of passive avoidance in mice (Ricceri, Alleva, & Calamandrei, 1994).

1.2.1.7 Studies manipulating glucocorticoid levels by stressor exposure Finally, there have been a number of studies investigating the effects of stress rather than glucocorticoids on learning and memory. Already 30 years ago, Lovely, Pagano, and Paolini (1972) found that rats housed individually had not only higher plasma corticosterone levels but also showed a faster acquisition of an active avoidance task than rats housed in pairs. Most extensively, the effects of acute stress exposures have been investigated by Shors and colleagues (for reviews see Shors, 1998; Shors, Beylin, Wood, & Gould, 2000). Shors, Weiss, and Thompson (1992) found that restraint stress facilitates classical conditioning. However, in their procedure, the stressor is presented 24h before conditioning. Even though corticosterone was found to be necessary for the effect to occur (Beylin & Shors,

42

1999), it was not found to be sufficient. Corticosterone levels at the time of conditioning were back at baseline (Shors, Pickett, Wood, & Paczynski, 1999). Even though the effect could be observed if conditioning followed shortly after the stressor, retrograde stress did not affect conditioning (Shors, 2001). Hence, these effects are not really comparable to the aforementioned effects of glucocorticoids on memory consolidation. Shors proposed a sensitisation-like mechanism mediated by activation of NMDA receptors in the LA/BLA but not CEA (Shors & Matthew, 1999). Interestingly, the stress-induced facilitation of learning was limited to male rats, whereas in females an impairment was observed (Wood & Shors, 1998). The great majority of studies reported in this thesis, including those of my own, utilized male rats. Even though this issue is not addressed in the current project, the role of sexual dymorphism in the brain and its consequences on cognition and behaviour deserves more attention in the future. The effects of acute pre- and post-training psychosocial stress have been investigated by Liu, Tsuji, Takeda, Takada, & Matsumiya (1999), using a passive avoidance task. Psychosocial stress was defined as watching and hearing the jumping, struggling, vocalising and defecation of shocked fellow rats while no shocks were administered to the subjects themselves. Liu et al. (1999) found that psychosocial stress led to increased levels of corticosterone as well as to enhanced learning. Both effects were blocked by prior administration of the corticosterone synthesis inhibitor metyrapone, supporting the idea that glucocorticoids enhance memory consolidation. In line with this, Kaneto (1997) found that pre- and post-training footshock enhanced later retention performance in a passive avoidance task. Other researchers have reported memory impairments following post-training stressor exposure. For example, Rudy (1996) found that isolation for several hours after training reduced contextual fear conditioning. However, Rudy et al. (1999)

43

proposed a role for endogenous opioids rather than glucocorticoids in this isolation effect. Similarly, Diamond and colleagues (Diamond, Fleshner, Ingersoll, & Rose, 1996; Diamond, Park, Heman, & Rose, 1999) found that post-training stressors such as exposure to a novel environment or to a predator led to spatial memory impairments in a 14-arm radial maze and a radial-arm water maze respectively. It should be noted though that retention tests occurred right after stressor presentation so that the impairing effects might have been on the retrieval process.

1.2.2 The acute effects of glucocorticoids on memory acquisition and consolidation (human studies) The acute effects of glucocorticoids on memory acquisition and consolidation have been investigated to a considerably lesser extent in humans than in animals. This section is divided into two parts. The first part (1.2.2.1) discusses those studies in the human literature that involved the administration of glucocorticoids and/or the measurement of cortisol levels, and looked at memory tasks. The second part of the section (1.2.2.2) is concerned with studies pursuing the question of whether emotional arousal has an effect on memory.

1.2.2.1 Studies involving the administration of exogenous glucocorticoids or the measurement of endogenous glucocorticoid levels There have been considerably fewer studies concerning the acute effects of glucocorticoids on memory consolidation in humans than in animals, and the designs and results of these seem on the whole less conclusive. Nevertheless, they need to be reviewed here. It should also be pointed out that memory assessment often involves non-emotional declarative material, whereas in animals, learning procedures are in

44

most instances emotional, i.e. involving the presentation of either an aversive or, occasionally, an appetitive reinforcer. This issue will be revisited in Chapter 3. Hence, a series of important studies, not specifically concerned with the effects of glucocorticoids but rather with emotional arousal in general, follows this section (1.2.2.2). Beckwith, Petros, Scaglione, and Nelson (1986) found that when different doses of hydrocortisone were administered 60 min before the presentation of lists of words, recall for the first of those lists was enhanced. Since recall was tested immediately after training, the design of the study did not permit any conclusions regarding the effects of glucocorticoids on consolidation of long-term memory. An interaction was found between drug dose and practice. All doses enhanced performance for the first list, but only the highest dose enhanced memory for subsequent lists, whereas the lowest dose impaired it. Beckwith et al. (1986) therefore proposed an indirect effect of hydrocortisone on memory via arousal and motivation. Kirschbaum, Wolf, May, Wippich, and Hellhammer (1996) induced elevated cortisol levels by psychosocial stress (public speaking and mental arithmetic). Cortisol levels negatively correlated with performance in a subsequent declarative memory task, and in a later study this correlation was only found in men but not women (Wolf, Schommer, Hellhammer, McEwen, & Kirschbaum, 2001b). Pretreatment with hydrocortisone impaired declarative memory as compared to placebo while leaving procedural memory intact. However, Lupien, Gillin, and Hauger (1999) did not find an effect of hydrocortisone on declarative memory and proposed that working memory was more sensitive to the effects of glucocorticoids. As in the study by Beckwith et al. (1986), short-term- rather than long-term memory was assessed, and cortisol levels were elevated both during training and test. Hence, no conclusions

45

about possible effects on long-term memory consolidation can be made from these studies. The effects could be attributed to any phase of learning and memory. Using a similar design, Wolf et al. (1997) found that acute administration of the neurosteroid DHEA-S decreased cortisol levels but had no effect on memory. Fehm-Wolfsdorf, Reutter, Zenz, Born, and Lorenz-Fehm (1993) considered the diurnal rhythm of endogenous glucocorticoids. In humans, levels are at peak in the morning and lowest in the evening. The researchers found better performance in a free recall task in the morning than at night. It is possible, however, that other confounds such as different levels of tiredness might have contributed to this pattern of results. Administration of hydrocortisone reduced morning performance to evening levels. However, evening administrations had no effect. Lupien and McEwen (1997) suggested that the endogenous glucocorticoid levels in the morning could have corresponded to the performance peak in the inverted U-shaped function so that hydrocortisone impaired performance by causing a shift to the descending end of the inverted U. However, evening administrations of hydrocortisone during low endogenous levels may not have been sufficient to shift performance toward the peak of the function. It should be noted that again memory was tested immediately after training so that no real conclusions on the role of glucocorticoids on long-term memory consolidation can be made. Lupien, Gillin, Frakes, Soefje, and Hauger (1995) administered different doses of hydrocortisone to subjects while they learned and recalled a word-pair list. Four hours later, they learned and recalled a second list. The authors found that cortisol levels measured during exposure to the first list (during the infusion), but not the second list, were positively correlated with the degree of memory impairment four days later (as compared to original recall performance). From these data, Lupien et al.

46

(1995) proposed that glucocorticoids may have a detrimental effect on memory acquisition and consolidation. Lupien and McEwen (1997), in a review article, proposed a model of the acute effects of glucocorticoids on human cognition. According to this model, MRs mediate a dose-dependent inverted U-shaped function between glucocorticoids and selective attention, arousal and sensory integration. These processes can affect down-stream aspects of memory like acquisition and consolidation. Furthermore, they suggested a detrimental effect of glucocorticoids on memory acquisition and consolidation, likely to be mediated by GRs. Vedhara, Hyde, Gilchrist, Tytherleigh, and Plummer (2000) compared students’ perceived self-reported degree of stress, cortisol levels and performance on a number of cognitive tasks during putatively less (non-exam period), and more, stressful (exam period) times. Self-reported stress was higher in the exam period but contrary to the authors’ expectations, cortisol levels, measured by saliva sampling at different times of the day, were lower during that time. Short-term memory performance was enhanced in the exam-period during which cortisol levels were lower. However, the authors acknowledged the possibility of a practice effect since the exam period followed the non-exam period for all subjects, and the same items, albeit in a different order, were used in both versions of the task. Acute cortisol levels at the time of testing were not used for the analysis, and the study did not include a long-term memory task. Therefore no conclusion can be made regarding memory consolidation. Most human studies reported above have found detrimental effects of glucocorticoids on memory. While this seems to be in contrast to the many findings in the animal literature where glucocorticoids proved beneficial to memory, it has to be

47

noted that none of these studies specifically investigated the effects of glucocorticoids on memory consolidation. Instead, most studies included short-term memory tasks where retention tests occurred shortly after training with glucocorticoid levels being manipulated at all phases of learning and memory. Secondly, the material used in these memory tasks was generally neutral, rather than emotionally salient as in the animal studies. The only study comparable to the animal literature that investigated the effects of glucocorticoids on memory consolidation was carried out by Buchanan and Lovallo (2001). Subjects were administered either hydrocortisone or placebo and then shown a set of sixty emotionally arousing or neutral pictures. They were asked to rate all stimuli on 9-point Likert scales of emotional arousal and emotional valence. One week later, subjects returned for a free- and cued recall as well as a recognition test. It was found that hydrocortisone-treated subjects performed better than placebo subjects in the cued recall test. When stimuli were labelled as either low arousal or high arousal stimuli, based on a median-split according to the perceived ratings of each group, it was found that high arousal stimuli were remembered better than low arousal stimuli. When stimuli were labelled as pleasant, unpleasant and neutral, based on tertiary splits, pleasant and unpleasant stimuli were remembered better than neutral ones. Finally and most importantly, it was found that the memory-enhancing effect of hydrocortisone was selective for high arousal stimuli. The results are in line with the general findings reported in the animal literature that glucocorticoids can enhance memory consolidation. However, Buchanan and Lovallo (2001) administered treatment prior to stimulus exposure, so that the effect might have been on acquisition mechanisms, although comparison with the animal literature suggests an effect on consolidation instead. The fact that hydrocortisone only affected memory for emotionally arousing stimuli is in line with the fact that de Quervain, Roozendaal,

48

Nitsch, McGaugh, and Hock (2000) did not find an effect of pre- and post-training cortisone in a non-emotional memory task where these two conditions served as controls for a pre-retrieval cortisone condition (see Section 1.2.3). The specificity of glucocorticoid effects for emotionally arousing material was linked to possible interactions with the central noradrenergic system, in line with the animal work of Roozendaal discussed in the previous section (e.g. Quirarte et al., 1997; Roozendaal, 2000) and the human work of Cahill and colleagues discussed in Section 1.2.2.2 below. The fact that stimuli of either emotional valence were remembered better than neutral ones is also interesting, given that parallel experiments in animals, with few exceptions (Micheau et al., 1981, 1985), exclusively employed aversive paradigms. This issue is addressed empirically in Chapter 2 of this thesis.

1.2.2.2 Studies comparing memory for emotional and neutral items Even though Cahill and colleagues have not directly investigated the effects of glucocorticoids on memory consolidation in humans, one could argue that they did so indirectly, and their work is certainly important when discussing glucocorticoid memory modulation. Cahill et al. (1994) demonstrated that acute emotional arousal enhances memory in humans, and that the noradrenergic system is involved in this facilitation. Based on a procedure by Heuer and Reisberg (1990), subjects were presented a narrative, accompanied by a series of slides, in either an emotionally arousing or a neutral version. In the neutral version, a boy visited his father’s workplace in the hospital. In the emotionally arousing version, the boy had a severe accident and had to be treated in the hospital. Both stories were of the same length, containing 12 items, and only the middle section of the slides and narrative differed. Subjects received either placebo or the β-adrenergic receptor antagonist propranolol

49

before hearing the story. Retention was tested after one week. In the placebo condition, the emotionally arousing items were recalled significantly better than the neutral ones, showing that emotional arousal enhances memory. However, this effect was blocked by β-noradrenergic blockade. Cahill et al. (1994) argued that the noradrenergic system is crucial for emotional memory modulation, a finding in line with the aforementioned animal literature on the topic. Cahill & McGaugh (1995) replicated the results of enhanced memory by emotional arousal in an improved procedure where the slides of the middle section were exactly the same and only the accompanied narrative differed. This allowed them to rule out possible confounding effects of presenting different slides, e.g. the possibility that they differ in inherent memorability or novelty. Superior memory for emotionally arousing material was also found in people with disorders that generally compromise memory function, like Alzheimer’s disease (Ikeda et al., 1998; Moayeri, Cahill, Jin, & Potkin, 2000; Kazui et al., 2000; but see Hamann, Monarch, & Goldstein, 2000) or Korsakoff’s syndrome (Hamann, Cahill, McGaugh, & Squire, 1997; Hamann, Cahill, & Squire, 1997). Van Stegeren et al. (1998) investigated whether β-blockers modulated memory centrally or peripherally. The experimental design was almost identical to Cahill et al.’s (1994) but two different β-adrenergic blockers, propranolol and nadolol, were compared. Propranolol crosses the blood-brain-barrier easily, whereas nadolol does not. It was found that only propranolol blocked the memory-enhancing effect of emotional arousal, suggesting that β-blockers act centrally and not peripherally on emotional memory. In contrast, O’Carroll, Drysdale, Cahill, Shajahan, and Ebmeier (1999a) did not find an effect of either propranolol or nadolol on memory, employing only the emotionally arousing narrative of Cahill et al’s (1994) procedure. However, in another study by O’Carroll, Drysdale, Cahill, Shajahan, and Ebmeier (1999b), emotional

50

memory was enhanced by yohimbine and impaired by metoprolol, drugs that stimulate and block central noradrenergic activity respectively. Yohimbine stimulates central noradrenergic activity by blocking alpha-2-adrenergic autoreceptors, while metoprolol is a widely used clinical beta blocker. The results support the idea that during an emotional experience, sympathetic arousal, ultimately in form of central noradrenergic activation, modulates memory consolidation. Not only emotional but also muscle-tension-induced arousal has been shown to enhance memory, depending on ß-adrenergic activity (Nielson & Jensen, 1994). Since glucocorticoid memory modulation is thought to depend on an interaction between glucocorticoids and central NA (e.g. Quirarte et al., 1997; for reviews see Cahill & McGaugh, 1996, 1998; Cahill, Roozendaal, & McGaugh, 1997; McGaugh & Cahill, 1997; Roozendaal, 2000), the studies discussed above, while not directly investigating the effects of glucocorticoids, are nevertheless important in terms of their theoretical implications and their comparability to many animal studies on the effects of emotion on memory consolidation. Cahill et al. (1996) also investigated the neural substrates of memory enhancement by emotional arousal in humans. They showed twelve neutral and twelve emotional films to subjects and tested free recall three weeks later. Emotional films were significantly better recalled than neutral films. Using positron emission tomography (PET), it was found that the glucose metabolic rate of the right amygdala complex during encoding was significantly correlated with the number of emotional, but not with the number of neutral films recalled. This again supports findings reported in the animal domain (e.g. McGaugh, 2000) that the amygdala is critically involved in the memory modulation of emotionally arousing events (for reviews see Cahill, 1997, 1999, 2000a,b; Cahill & McGaugh, 1996, 1998; McGaugh et al., 2000;

51

Phelps & Anderson, 1997). In further support of this, Cahill et al. (1995) reported that a patient with Urbach-Wiethe disease, a rare hereditary condition with bilateral brain damage confined to the amygdala complex, did not show an enhanced memory for emotionally arousing material. Similar reports involving patients with amygdala damage have come from Adolphs and colleagues (Adolphs, Cahill, Schul, & Babinsky, 1997; Adolphs, Tranel, & Denburg, 2000). Both of two patients with bilateral damage of the amygdala failed to show the enhancement of memory for emotional material that was displayed in control subjects (Adolphs et al., 1997). In patients with unilateral damage, the results were less clear. Only patients with left amygdala damage failed to show this enhancement (Adolphs et al., 2000). However, only two subjects with right amygdala damage participated in the study. LaBar and Phelps (1998) found improved memory consolidation for emotionally arousing taboo words over neutral words in control but not in temporal-lobectomy subjects. Hamann, Ely, Grafton, and Kilts (1999) reported enhanced memory for emotionally arousing stimuli of both pleasant and aversive nature, and confirmed an involvement of the amygdala as indicated by PET. Most recently effects of hemispheric laterality (Canli, Zhao, Brewer, Gabrieli, & Cahill, 2000) and gender-laterality interactions (Cahill et al., 2001; Kilpatrick & Cahill, 2001) have been found, linking the left and right amygdala with enhanced memory for emotional material in women and men respectively. Adolphs, Denburg, and Tranel (2001) investigated possible interactions between amygdala damage and different aspects of stimuli to be remembered. They assessed memory for gist and visual detail of photographs displaying realistic scenes that were either aversive and emotionally arousing, or neutral. Gist referred to the salient, central information of a scene. Adolphs et al. (2001) found that in normal patients, brain damage controls and patients with unilateral amygdala damage,

52

emotional arousal enhanced memory for the gist of a scene. Memory for visual detail was generally better for neutral stimuli. This pattern was reversed in a patient with bilateral amygdala damage who displayed impaired memory for the gist of emotional scenes, both relative to controls and in the within-subject comparison to the gist of neutral scenes. Conversely, the subject with bilateral amygdala damage showed superior memory for the visual detail of emotional scenes compared to all control groups. The authors suggested that emotional arousal, mediated by the amygdala, enhances memory for gist at the expense of visual detail.

1.2.3 Effects of chronically elevated glucocorticoid levels on memory As mentioned earlier, there has been considerable confusion and controversy about the role of glucocorticoids in memory. The titles of two recent articles by de Kloet et al. (1999), “Stress and cognition: are corticosteroids good or bad guys?”, and Sapolsky (2000), “Stress hormones: good and bad.”, illustrate this. What applies to many things in life also applies to glucocorticoids with respect not only to memory (McEwen & Sapolsky, 1995) but also to general health (Sapolsky, 2000): excessive and inappropriate occurrence can be detrimental. While glucocorticoids serve an important function in acute stress situations, for example by enhancing memory consolidation as illustrated above, chronic stress, chronic administration of glucocorticoids or chronically elevated endogenous levels of glucocorticoids have been found to impair memory in both animals (e.g. Arbel, Kadar, Silbermann, & Levy, 1994; Dachir, Kadar, Robinzon, & Levy, 1993; Fuchs, et al., 2001; Luine, Spencer, & McEwen, 1993; Luine, Villegas, Martinez, & McEwen, 1994; Ohl, Michaelis, Vollmann-Honsdorf, Kirschbaum, & Fuchs, 2000) and humans (Greendale, Kritz-Silverstein, Seeman, & Barrett-Connor, 2000; Lupien, et al., 1994;

53

Newcomer, Craft, Hershey, Askins, & Bardgett, 1994; Newcomer et al., 1999; Sapolsky, 2000; Starkman, Gebarski, Berent, & Schteingart, 1992; Rubinow, Post, Savard, & Gold, 1984; Wolkowitz, et al., 1990), an effect often associated with hippocampal dysfunction. The human neuropathological condition of chronic hypercortisolemia, i.e. chronically enhanced levels of cortisol, is known as Cushing’s syndrome. Cushing’s syndrome patients have been found to frequently display a number of cognitive deficits, for example in visual memory of geometric designs (Whelan, Schteingart, Starkman, & Smith, 1980). In line with the studies mentioned above on chronic glucocorticoids leading to memory impairments and the finding that prolonged elevation of glucocorticoid levels are toxic to hippocampal cells (Sapolsky, 1986), Starkman et al. (1992) found a positive correlation between verbal memory functions and hippocampal volume, as well as a negative correlation between mean plasma cortisol levels and hippocampal volume in 12 patients with Cushing’s syndrome. The latter was, however, not correlated to general measures of cognitive function like the WAIS-R Fullscale IQ.

1.2.4 Effects of glucocorticoids on memory retrieval Given the inverted U-shaped dose-response function observed in many studies reported in the previous section (Conrad et al., 1996; Kovacs et al., 1977; Pugh & Rudy, 1997b; Quirarte et al., 1997), too much acute stress and excessively high glucocorticoid levels can also be detrimental to memory processes. Furthermore, it is important to point out that the concept of memory also comprises the retrieval phase during which previously learned information is remembered. Anecdotal evidence tells us that stress has a detrimental effect on memory retrieval. Most people have experienced problems remembering certain information when they were nervous and

54

stressed, for example in test situations or job interviews. Experimental evidence confirming this comes from the work of de Quervain and colleagues (de Quervain, Roozendaal, & McGaugh, 1998) who found that footshock 30 min prior to a retention test impaired rats’ performance in the Morris water maze. This effect was blocked by prior administration of the corticosterone synthesis inhibitor metyrapone. Pre-test administration of corticosterone also impaired performance. De Quervain et al. (2000) showed the same effect in humans. Pre-test administration of cortisone impaired delayed free-recall of a word list in a long-term declarative memory study, as compared to the placebo condition. Wolf et al. (2001a) found a similar cortisolinduced deficit on retrieval of more recently learned material. In other studies, the effects of glucocorticoids on memory retrieval cannot easily be dissociated from effects on extinction and/or expression of memory. Quite a number of studies have investigated the effects of glucocorticoids on extinction in animals (e.g. Bohus, 1970; van Wimersma Greidanus, 1970; Micco, McEwen, & Shein, 1979; Micheau, Destrade, & Soumireu-Mourat, 1982; Port, Sisak, Finamore, Soltrick, & Seybold, 1998; see Chapter 3) and even though the overall impression of these studies is that glucocorticoids might facilitate extinction (but see Micco et al., 1979), not all studies may be comparable. Extinction is probably not just a process of retrieval failure but also constitutes new learning. In some studies, glucocorticoids were administered before the extinction session, also measuring retrieval in the same session (Bohus & Lissak, 1968; Kovacs, Telegdy, & Lissak, 1976; Port et al., 1998); in other studies they were given after extinction, with further test sessions following a delay (van Wimersma Greidanus, 1970). Further, there may be overlapping and possibly even opposing effects on retrieval and expression of behavioural inhibition in some studies. For example, some forms of behavioural inhibition such as freezing (Cordero et al.,

55

1998), the ontogenetic development of freezing (Takahashi & Kim, 1994) and acquired immobility in a forced-swimming paradigm (Mitchell & Meaney, 1991) are thought to be directly affected by glucocorticoids. In certain experimental procedures this can lead to difficulties in the interpretation of the results. For example, Corodimas et al. (1994) found enhanced conditioned freezing with implanted corticosterone pellets that were present over several days across the extinction phase. While this might indicate a role of corticosterone in the expression of freezing, it also appears to contradict the finding of de Quervain et al. (1998) that glucocorticoids impair retrieval, as well as those findings mentioned above that reported a facilitative effect on extinction.

Given the opposing effects of acute glucocorticoid administration on memory consolidation on the one hand (Sections 1.2.1 & 1.2.2), and chronic glucocorticoid administration in general (Section 1.2.3) or acute glucocorticoid administration on memory retrieval (Section 1.2.4) on the other hand, it is hardly surprising that some confusion exists when talking about the role of glucocorticoids in memory. This further elucidates the importance of considering the experimental design carefully when interpreting the data.

1.2.5 Effects of glucocorticoids in electrophysiological studies There have been a number of studies investigating the effects of stress and glucocorticoids on a variety of electrophysiological measures in the hippocampus, including LTP, long-term depression (LTD) and primed burst potentiation (PBP). LTP in an electrically induced increase in synaptic transmission whose rapid onset and long duration are reminiscent of memory (Teyler & DiScenna, 1987). Assuming

56

that changes in synaptic efficacy constitute the basis for learning and memory (e.g. Hebb, 1949), these phenomena may constitute physiological models of memory (Brown, Chapman, Kairiss, & Keenan, 1988). PBP is a low-threshold form of LTP that is produced by a total of only five physiologically patterned pulses (Diamond et al., 1992). The effects of stress and glucocorticoids on hippocampal PBP have primarily been investigated by Diamond and colleagues (Diamond & Park, 2000). For example, Diamond, Bennett, Engstrom, Fleshner, and Rose (1989) found that ADX reduced the threshold for hippocampal PBP in anaesthetised rats, suggesting that glucocorticoids exert an inhibitory influence on hippocampal plasticity. Bennett et al. (1991) reported a negative correlation between circulating corticosterone and the magnitude of PBP of the CA1 population spike. However, Diamond et al. (1992) proposed that this observed linear correlation was only part of a non-linear inverted U-shaped relationship between corticosterone levels and hippocampal PBP, finding a positive correlation at low and a negative correlation at high levels of serum corticosterone in anaesthetised ADX rats with implanted corticosterone pellets. Diamond, Fleshner, and Rose (1994) found that psychological stress in form of a novel environment increased serum corticosterone levels and blocked hippocampal PBP in behaving rats. A different form of stressor, exposure to a predator (cat), blocked PBP but not LTP in the hippocampus in vivo, suggesting that stress inhibits but does not completely block the development of hippocampal plasticity. The first study on the effects of stress on LTP was carried out by Foy, Stanton, Levine, and Thompson (1987), who found that tailshock impaired LTP in hippocampal explants. The relationship between adrenocortical stress hormones and LTP and LTD was primarily investigated by Pavlides and colleagues. The hormones

57

seem to have a biphasic effect on LTP and LTD in the hippocampus, involving both MRs and GRs. Pavlides, Watanabe, and McEwen (1993) found that chronic administration of corticosterone impairs LTP in the DG in anaesthetised rats, even 48 hr after cessation of treatment. Acute corticosterone also impaired LTP, but only for as long as hormone levels were elevated. The way in which LTP is modulated by glucocorticoids therefore seems to depend on the receptor type that is targeted. While MRs are usually heavily occupied during the diurnal cycle, GRs start to be increasingly occupied in a stressful situation. It was found that MR activation by aldosterone facilitates and prolongs LTP in the DG in freely behaving ADX rats as well as in hippocampal CA1 slices (Pavlides, Kimura, Magarinos, & McEwen, 1994; Pavlides, Ogawa, Kimura, & McEwen, 1996), an effect that can be blocked, in vivo, by administration of the MR antagonist RU28318 (Pavlides, Watanabe, Magarinos, & McEwen, 1995b). GR activation by RU28362 disrupts LTP in the DG in vivo and in CA1 in vitro (Pavlides, Kimura, Magarinos, & McEwen, 1995a; Pavlides et al., 1996) or even induces LTD (Pavlides et al., 1995a). The disruptive effect of RU28362, in turn, can be blocked by the GR antagonist RU38486 (Pavlides et al., 1995b). In the same way that glucocorticoids differentially mediate LTP, they also have other differential electrophysiological effects, for example changes of excitability of hippocampal neurons. A theory of how glucocorticoids disrupt LTP involves glucocorticoid inhibition of glucose transport in the hippocampus. This is mediated through GRs and seems to involve both translocation of existing glucose transporter molecules from the plasma membranes of the cell to a putative intracellular membrane fraction (Horner, Munck, & Lienhard, 1987), and also a genomic decrease of mRNA levels for the glucose transporter (Garvey, Huecksteadt, Lima, & Birnbaum, 1989). Doyle, Rohner-Jeanrenaud, and Heanrenaud, (1993) found that the

58

hippocampus is particularly vulnerable for glucocorticoid inhibition of glucose transport. LTP is an energy demanding process and the glucocorticoid inhibition of glucose transport accelerates the loss of ATP resulting in impaired efficacy of LTP. Taken together, these electrophysiological results suggest that stress and glucocorticoids impair or block hippocampal synaptic plasticity. However, their relationship seems to follow an inverted U-shape function with low doses of corticosterone, possibly mediated by MRs, exerting a positive effect on PBP or LTP. The results stand in contrast to the majority of the behavioural studies discussed above, as these predominantly showed enhancing effects of glucocorticoids on memory

consolidation.

However,

as

Sandi

(1998)

pointed

out,

such

electrophysiological studies always manipulate steroid levels before the induction of LTP or PBP and hence mimic exposure to the learning situation under circumstances of previously induced stress or glucocorticoid elevation. This may not be ecologically valid. In fact, in the first study to investigate changes in LTP induction in freely behaving rats during exposure to the stressor, Bramham, Southard, Ahlers, and Sarvey (1998) found that acute cold stress (4ºC) led to elevated levels of corticosterone but did not impair LTP in the DG. It is also a matter of debate whether PBP and LTP really constitute a valid physiological model of learning in the first place (for a discussion see Shors & Matzel, 1997). Further, it should be noted that the electrophysiological effects linked to stress and glucocorticoids have been studied exclusively in the hippocampus although other brain areas may be involved in memory consolidation. First and foremost, the amygdala is thought to play an important role in mediating stress responses and modulating memory elsewhere in the brain, for example in the hippocampus (Packard et al., 1994; Cahill & McGaugh, 1998). Lesions of the BLA impair LTP in the DG in vivo (Ikegaya et al., 1994) while

59

priming stimulation of the BLA enhances LTP in the DG (Akirav & Richter-Levin, 1999a). Kim et al. (2001) found that the amygdala is critical for stress-induced modulation of hippocampal LTP and learning. In line with this, Akirav and Richter Levin (1999b) found evidence for, and proposed, a biphasic modulation of hippocampal LTP by behavioural stress and BLA stimulation according to which ...the emotionally activated amygdala in its fast excitatory phase serves as a marker for important events, processed by the hippocampus, to be stabilized and thus remembered. The activation of the slower inhibitory phase may be beneficial to reduce masking effects of following events during the initial consolidation stage. (p.10535)

1.2.6 Summary, limitations of previous work, outline of the empirical components of this project The literature on the effects of glucocorticoids on memory is extensive and controversial. Often effects of chronic and acute administration, as well as effects on acquisition, consolidation, and retention are confounded and a confusing picture arises. This project is concerned only with the acute effects of glucocorticoids on memory consolidation. A great deal of evidence, particularly from animal studies, suggests that glucocorticoids enhance memory consolidation in a time- and dosedependent way. This memory modulation seems to be mediated selectively by GRs. MRs, in contrast, have been linked to non-specific aspects of learning like strategy selection and sensory integration. While many human studies have reported detrimental effects of glucocorticoids, the small number that have specifically investigated the effects of glucocorticoids on memory consolidation in a way that can be compared with the large and reasonably consistent body of animal research, come to conclusions supporting those of the animal work, proposing a memory modulatory role of glucocorticoids. Figure 1.2.6.1 presents a broad, simplified schema of the way

60

exogenous

CORT

endogenous

HPA activation

memory modulation

experimental treatment

CONTEXT discrete cues?

appropriate response at next encounter (e.g. avoidance) in some cases, excessive fear (PTSD, phobias)

Figure 1.2.6.1. Simple schema of glucocorticoid memory modulation, derived from the current literature. Thick, dotted arrow reflects a strengthened association.

61

glucocorticoid memory modulation could be seen on a behavioural level, according to the existing literature. However, it should be noted that the vast majority of animal studies have employed either a passive (inhibitory) avoidance procedure or a spatial water-maze task. While it is not entirely clear what type of cognitive process the passive avoidance task really requires, it has recently been suggested (Roesler, Berlau, LaLuminiere, Roozendaal, & McGaugh, 2001) that learning this task does not depend on instrumental action. However, the conditioned behaviour of, for example, not entering the shock-associated compartment is maintained by the absence of an aversive US. Similarly, spatial learning in the Morris water maze is not the most straightforward associative procedure. Pavlovian (classical) discrete-cue conditioning, in contrast, has only been employed very rarely as a paradigm. In fact, Pugh et al. (1997a,b) proposed that glucocorticoids do not affect this type of procedure, and that their role in fear conditioning might be selective to the formation and consolidation of context representations instead. It is difficult to see why glucocorticoids, with their proposed adaptive role as a memory modulator, would not affect discrete-cue conditioning, especially as many stressors in an organism’s environment seem to be predicted by such discrete cues. Furthermore some studies (e.g. Sandi & Rose, 1994a,b; Micheau et al., 1981, 1985), albeit not constituting classical discrete-cue conditioning paradigms, have suggested that memory for discrete cues is affected by glucocorticoids. Hence, the first aim of this PhD project was to investigate whether or not glucocorticoid memory modulation could be demonstrated in discrete-cue Pavlovian conditioning. These experiments are reported in Chapter 2. The GR agonist DEX is used as the only pharmacological manipulation in this thesis. The reasons behind choosing DEX are discussed in the Appendix.

62

Further, there has only been one group of investigators (Micheau et al., 1981, 1985) that tested for the effects of glucocorticoids in appetitive procedures, reporting effects only in an operant procedure. The scarcity of such studies is surprising since glucocorticoids are also released in response to pleasant emotional stimuli such as food, sex or drugs of abuse (Caggiula et al., 1991; Fuller & Snoddy, 1981; Honma, Honma, & Hiroshige, 1984; Merali, McIntosh, Kent, Michaud, & Anisman, 1998; Orchinik, Licht, & Crews, 1988; Piazza & Le Moal, 1997). There is evidence that enhanced memory for emotional material is mediated primarily by arousal in general rather than the valence of the material (Bradley et al., 1992). Hence, a further aim for this project was to investigate the effects of glucocorticoids on memory consolidation in an appetitive procedure, which, in line with the aforementioned point, utilises discrete cues. Again, these experiments can be found in Chapter 2. A striking difference between the animal and the human literature is that in the former all paradigms are emotionally salient, while in the latter, almost none are. It is easier to test a human subject (also for ethical reasons) on neutral memory tasks such as delayed recall of a word list. However, with respect to stress hormones, this does not necessarily pose an ecologically valid situation. Acute release of glucocorticoids implies the presence of an emotionally salient event that glucocorticoids may aid one to learn about. On the contrary, it is rather difficult to test a rat on a neutral memory task, merely because it is more difficult to examine non-motivated behaviours in rats. However, there are some learning paradigms that may allow the assessment of associative learning about neutral cues, or learning that is not concerned with a primary reinforcer. Given the interesting findings of de Quervain et al. (2000), and Buchanan and Lovallo (2001), in humans, i.e. the necessary emotionalnature of material for glucocorticoids to exert a memory-enhancing effect on them, it would be

63

of great interest to examine this issue in rats, and investigate if neutral or non-primary emotional learning is equally unaffected by glucocorticoids. Chapter 3 attempts to find an answer to this question. Most of the reported animal studies have in common that learning takes place about the environment and its predictive relationship to either the occurrence of an unpleasant event, most prominently a footshock, or, in the case of the water-maze task, the removal of an unpleasant event, the threat of drowning. These events are of rather abstract nature. Even though a footshock is a real physical event, it is difficult to describe its exact properties, probably because of its high degree of aversiveness and its extremely short duration. Appetitive reinforcers however, which have not much been studied in relation to glucocorticoid memory modulation, can be perceived in more detail, and they can be divided into their motivational and sensory properties. In other words, a food pellet possesses certain motivational properties, describing its role as constituting nutrition and serving the biological need of eating. These express the direction of its emotional valence. However, a food pellet also has sensory properties that make it unique from all other types of food, such as its flavour, its texture, its colour etc. Hence, the question arises what type of reinforcer properties are actually remembered better due to glucocorticoids. Do glucocorticoids enhance associations between stimuli in the environment and the motivational or the sensory properties of the reinforcer? Chapter 4 deals with this question and presents experiments that aim to dissociate the different reinforcer properties in the context of memory modulation by glucocorticoids. In human studies a similar distinction has been made between the gist and the detail of learned material, and the amygdala has been found to play a role in the extent to which these different properties are processed (Adolphs et al., 2001).

64

Finally, it was noted in Section 1.1.7 that within a normal population, there can be substantial individual differences in stress responsivity in form of glucocorticoid release. These differences have been investigated quite extensively with respect to glucocorticoid-dopamine interactions and reward sensitivity, proposing animal models of the human personality trait of sensation-seeking (e.g. Dellu et al., 1996a), as well as of the predisposition for drug addiction (e.g. Piazza & Le Moal, 1996). However, it is rather surprising, given the well-established role of glucocorticoids in memory modulation, that these individual differences have not been investigated with respect to learning and memory, even though this might give alternative, and possibly more parsimonious explanations of some of the observed behavioural differences, and it might possibly also lead to animal models of certain psychiatric disorders, for example PTSD, or drug addiction. Chapter 5 attempts to make the first steps into this direction. At last, Chapter 6 constitutes a brief overall summary and discussion of the results of the experiments reported in this thesis. It attempts to integrate the findings into an overall model in the framework of this introductory chapter.

65

CHAPTER 2 Memory modulation by glucocorticoids in aversive and appetitive discrete-cue Pavlovian conditioning paradigms

66

Post-training administration of glucocorticoids have been shown to modulate memory in a number of studies (see sections 1.2.1 and 1.2.2). Table 2.0.1 gives an overview of some of the most important ones, specifying the type of treatment, i.e. which drugs were used (partly in combination with specific brain lesions or ADX), the type of administration, i.e. how and when in relation to the learning event a drug was given, the type and emotional valence of the behavioural paradigm, and finally the observed effect on memory retention. At first glance, the overall conclusion of studies conducted so far is that posttraining GR agonists reliably enhance memory consolidation, when tested in a later retention test, whereas GR antagonists impair memory consolidation. However, a closer look reveals that not many different behavioural tasks have been used to study the effects of glucocorticoids on memory. Each of these may have different constraints so that one has to be cautious when generalising the results. For example, the great majority of studies employed the inhibitory avoidance paradigm, also known as passive avoidance. It is a matter of debate what type of cognitive task the inhibitory avoidance task actually constitutes. In the Roozendaal-and-McGaugh version of the task (e.g. Roozendaal & McGaugh, 1996a, 1997a), a trough-shaped alley is divided into two compartments by a retractable door. A rat is placed into the smaller, lit, compartment and can freely enter the dark compartment, where it receives a footshock. Days later, in a non-reinforced session, the latency of the rat to enter the dark compartment is taken as a measure of memory. Those who term the paradigm inhibitory avoidance, rather than passive avoidance, argue that the rat has to actively inhibit the desire to enter the naturally preferred dark compartment. However, the question remains if this paradigm constitutes a Pavlovian procedure, or some type of aversive one-trial operant task. On going through the retractable door, the behaviour

67

Table 2.0.1. Some key studies investigating the effects of glucocorticoids on memory consolidation. To be continued, next page. * see abbreviations, overleaf. Administration type time

Authors (Year)

Species

Treatment

Kovacs et al. (1977)

rat

CORT (A)

ip

Cottrell & Nakajima (1977)

rat

Hydrocortisone (A)

sc/ hippo

mouse

Hydrocortisone (A) DEX (A) CORT Anisomycin (ANI) + DEX

sc

Flood et al. (1978)

post

post

mouse

CORT

icv

post

Oitzl & de Kloet (1992)

rat

ADX Spironolactone (MR-A) RU38486 (GR-AN)

icv icv

Sandi & Rose (1994a)

chick

RU28318 (MR-AN) RU38486

Sandi & Rose (1994b)

chick

CORT

Roozendaal & McGaugh (1996a)

Roozendaal et al. (1996a)

rat

rat

CEA / CEA+DEX / CEA+CORT BLA / BLA+ DEX / BLA+CORT MEA / MEA+DEX / MEA+CORT Metyrapone (CORT inhibitor) Metyrapone + DEX Metyrapone + CORT

Valence

Effect on memory

AA

-

low dose facilitated, high dose impaired memory

PA

-

reversed impairing effects of Cycloheximide

AA / PA

-

all drugs enhanced retention (DEX up to 150 min. post) DEX blocked amnesic ANI effects

Micheau et al. (1981)

SHAM/SHAM+DEX/SHAM+C ORT

Paradigm*

OC OSD

+

no effect enhancement

pre & post pre & post

MWM

-

impairment change of strategy impairment

ic

pre

PA

-

impairment impairment

ic

pre post

PA

-

enhancement

sc

--- / enhancement / none impairment / impairm. blocked / none post

PA

-

none / DEX effect blocked / none none / DEX effect blocked / none

sc 68

pre pre + post pre + post

MWM

-

impairment attenuates impairment none

Table 2.0.1 (continued). Some key studies investigating the effects of glucocorticoids on memory consolidation.

Authors (Year)

Species

Treatment RU28362 (GR-A)

Roozendaal & McGaugh (1997a)

rat

Pugh et al. (1997a)

rat

Administration type time

Paradigm

Valence

Effect on memory

-

none enhancement none impairment

-

impairment none

RU38486

CEA BLA CEA BLA

post

RU38486

sc

pre

CFC DCFC

oral

pre

CFC DCFC CFC DCFC

-

impairment none reversed ADX effects none

PA MWM

ADX Pugh et al. (1997b)

rat

Cordero & Sandi (1998)

rat

RU28318 RU38486 CORT

icv icv ip

pre pre post

CFC (high/low shock)

-

none impairment with low shock only enhancement with low shock only

Buchanan & Lovallo (2001)

human

cortisol

oral

pre

free recall cued recall recognition

+/-

none enhancement for arousing items none

ADX + CORT

Abbreviations: + = appetitive; - = aversive; A = agonist; AA = active avoidance; ADX = adrenalectomy; AN = antagonist; ANI = anisomycin; BLA = basolateral nucleus of the amygdala; CEA = central nucleus of the amygdala; CFC = contextual fear conditioning; CORT = corticosterone; DCFC = discretecue fear conditioning; DEX = dexamethasone; GR- = GR selective; hippo = hippocampus; icv = intracerebroventricular; ip = intraperitoneal; MEA = medial nucleus of the amygdala; MR- = MR selective; MWM = Morris water maze; OC = operant conditioning; OSD = operant successive-discrimination; PA = passive avoidance; pre = prior to training; post = after training; sc = subcutaneous

69

of the rat is directly punished by footshock. This interpretation points to an operant component. However, in Pavlovian terms, one might argue that learning occurs about the aversive nature of the overall context of the larger, dark compartment. Similarly, light and dark could also be considered as discrete visual cues. Roesler et al. (2001) recently addressed part of this question, concluding that learned avoidance does not depend on the animal’s operant response of stepping into the shock compartment during training. These distinctions are likely to be important, as they highlight the fact that using this procedure, it is not clear which of these different forms of learning are susceptible to glucocorticoid-induced memory modulation. A further central question is that of the role of glucocorticoids in contextual vs. discrete-cue conditioning. As mentioned above, the nature of inhibitory avoidance is ambiguous. Contextual conditioning might be one of the processes occurring in this task. Rudy and colleagues (Pugh et al., 1997a,b), whose work is discussed in more detail in Chapter 1 (Section 1.2.1.5) and the general discussion of this chapter, addressed this question. Pugh et al.’s (1997a,b) results led them to argue that the role of glucocorticoids in memory is selective for forming a representation of the context, and that the hippocampus is probably the crucial neural substrate at which glucocorticoids modulate memory. The hippocampus is believed to be an important structure not only in context-specific learning (Honey & Good, 1993) and contextual conditioning (e.g. Phillips & LeDoux, 1992, 1994; Selden et al., 1991) but also in spatial learning (e.g. O’Keefe & Nadel, 1978; Morris, Garrud, Rawlins, & O’Keefe, 1982; Packard et al., 1994). However, if the BLA mediates glucocorticoid memory modulation by interacting with different parts of the brain (e.g. McGaugh et al., 2000; Roozendaal,

70

2000), one would probably not expect this modulation to be selective for one form of learning, such as contextual conditioning. Packard et al. (1994), for example, found that the amygdala modulates both hippocampus- and caudate nucleus-dependent processes. To my knowledge, no one has yet conducted a study in which glucocorticoids were administered after training in straightforward discrete-cue conditioning. This chapter describes a series of studies in which the effects of posttraining administration of the GR agonist DEX were investigated in discrete-cue Pavlovian conditioning paradigms. The selection of DEX as the pharmacological manipulation is discussed in the Appendix (Section A.4). Experiment 2.1 is based on the pilot studies reported in the Appendix (Sections A.1, A.2, A.3, A.5) and employs a fear-conditioning paradigm in which conditioned freezing is assessed. Experiment 2.2 employs another fear-conditioning procedure, testing for levels of conditioned suppression of an ongoing instrumental behaviour. Experiments 2.3 and 2.4 constitute appetitive designs and thus tackle another question in the field of memory modulation by glucocorticoids. As far as I am aware, all but few studies, i.e. those of Micheau et al. (1981, 1985) who employed an operant successive-discrimination paradigm, have assessed memory and glucocorticoids exclusively in the aversive domain. Although HPA axis activation and subsequent increases in glucocorticoids often reflect the presence of aversive stressors, the HPA axis is also activated by appetitive stimuli. Merali et al. (1998) found that both aversive and appetitive events evoked the release of CRH in the CEA, as well as increased levels of circulating ACTH and corticosterone. Piazza and Le Moal (1997) reported that glucocorticoids are secreted in response to positive reinforcers such as food, sex and drugs of abuse. Memory modulation by GR activation interacts with phasic changes in arousal caused by presentation of motivationally significant

71

reinforcement. In line with suggestions that a site of action of GR-induced modulation lies in the BLA (Roozendaal & McGaugh, 1996b, 1997a,b), Cahill and McGaugh (1990) found that complete lesions of the amygdala impaired learning in a task using highly aversive, but not mildly aversive or appetitive, reinforcement. Highly arousing stimuli such as footshock may result in amygdala-modulated learning due to the release of endogenous stress hormones. With less arousing stimuli, failure to activate endogenous stress systems may reduce the participation of the amygdala in learning. Even though this is, of course, speculative, it suggests that if the effect of arousal on learning is determined by the level of stress hormone activity rather than the aversive nature of a stimulus per se, memory modulation should follow exogenous GR activation, even in low-arousal tasks, including those of an appetitive nature. It is therefore of great interest whether or not glucocorticoids can modulate memory for both aversive and appetitive learning. Experiments 2.3 and 2.4 investigate the effects of post-training DEX in an appetitve discrete-cue Pavlovian conditioning paradigm.

2.1 GLUCOCORTICOIDS IN DISCRETE-CUE FEAR CONDITIONING I

The aim of this study was to investigate if glucocorticoids can modulate memory in a discrete-cue fear conditioning paradigm. Memory was assessed by conditioned freezing and the parameters used in this study were developed in the pilot studies reported in the Appendix. Two minor changes were instantiated. Compared to the pilot experiments, it was now possible to run two rats at the same time as a second behavioural testing chamber was set up. Further, the frequency of the clicker was decreased slightly from that used in (pilot) Experiment A.5, to 5 Hz.

72

Method Subjects Subjects were 24 male, Lister Hooded rats (Harlan, UK; n = 8/group), weighing between 280 g and 305 g (M = 294 g) at the beginning of the experiment. They were housed in pairs in a temperature-controlled colony room (21±2ºC) and maintained on a 12:12-hr-light-dark cycle (lights on at 08:00 am) with free access to water. Prior and during the experiment, rats were fed approximately 15g of food pellets per day to maintain their body weights. All procedures were carried out between 12:00 and 20:00. Apparatus All behavioural procedures took place in one of two identical modified Campden Instruments chambers (Campden Instruments Ltd., UK; 24.5-cm-wide x 20.5-cm-high x 22-cm-deep) located inside a soundproof cabinet, equipped with a fan providing a 62-dB background noise. The chamber was constructed from aluminium (side walls and ceiling) and plexiglass (hand hinged front door). The floor of the chamber consisted of 16 stainless steel bars spaced 1.5 cm apart centre-to-centre. Bars were wired and connected to a Coulbourn precision-regulated animal shocker (model E13-14, Coulbourn Instruments, USA), set to deliver scrambled footshock as US (0.225 mA, 0.5 s). The shock intensity was based on what is generally considered a mild footshock in the literature. A house-light on the centre of the ceiling was lit during the experiment. A tone (3 kHz, 80 dB, 10 s) served as an auditory CS and was delivered by two speakers in the left wall of the box. A second auditory cue was available in form of a 5-Hz train of clicks (clicker) with a sound level of 75 dB. It was produced by a heavy-duty relay mounted outside the chamber. The background noise in the chamber was 62 dB. The door of the chamber remained open during sessions

73

and the crucial parts of the procedure were filmed with a video camera and later assessed manually. A green and a red light, as well as a lit timer outside the box, indicated the various stages of the experiment. A BBC Master 128 microcomputer (Acorn, UK) equipped with a SPIDER extension for on-line control (Paul Fray Ltd, UK) controlled the equipment and recorded the data. Drugs and injection procedure DEX (Sigma, UK) was injected sc, immediately post-training, at doses of 0.3, 0.6 and 1.2 mg/kg in a volume of 2.0 ml/kg. The lowest dose of 0.3 mg/kg was based on the dose used by Roozendaal and McGaugh (1996a,b) and (pilot-)Experiment A.5. The other doses were derived from that in order to form a dose-response-curve. DEX was first dissolved in absolute ethanol and then diluted in 0.9% NaCl (saline), in order to reach the highest concentration (1.2 mg/kg). The other doses were derived from that by diluting with the appropriate amount of saline and ethanol. For the control condition, a 0.9% saline solution (SAL) with a matched final concentration of ethanol (3.5%) was used. Drugs were prepared freshly for each day of administration. Behavioural procedure Habituation and conditioning. On day 1, rats were placed in the box and received a conditioning session that lasted 25 min. They were given two nonreinforced presentations of the tone after 3 and 6 min, for habituation purposes. The tone was then paired with the administration of a mild footshock (0.225 mA, 0.5 s) that started immediately at the offset of the stimulus. This CS-US presentation occurred only once, after 20 min. The session was terminated after 25 min and rats were given an immediate sc injection of one of three doses of DEX, and then returned to their home cage. On day 2, a second acquisition session was given at the same time of day, only that this time clicker was presented and SAL was administered, so that all

74

subjects were trained with both stimuli and administered both treatments. Stimuli, treatments, order of stimuli, and order of treatments were fully counterbalanced. Extinction. On day 3, rats received an extinction session lasting 60 min during which 10 non-reinforced presentations of each of the two stimuli were given, the length of the ISI being 2 min 50 s. The order of stimulus presentation was pseudo-randomised with no more than two of the same stimuli occurring consecutively. The index of learning was freezing during the CS period. Assessment of freezing. Rats were filmed during all stimulus presentations. Freezing was defined as the absence of any movement other than breathing. In order to assess freezing during the CS period, the rater watched the recordings while listening to a sequence of beeps (1/s) from a metronome. At each beep, the rater made a decision as to whether he thought that freezing took place during that moment or not, and entered the data directly into an excel spreadsheet. See Table 2.1.1 for an outline of the experimental design. Statistical analysis All data were first processed with Microsoft Excel (Microsoft Corporation, USA) on an RM Personal Computer (Research Machines Plc., UK) and then analysed with the statistical software packages Statistica (Statsoft, USA) and SPSS (SPSS Inc. USA). Analysis of variance (ANOVA) and, when appropriate, simple-effect analysis (Levine, 1991) and posthoc tests (Newman-Keuls Pairwise Comparisons) were conducted.

75

Table 2.1.1. Design of Experiment 2.1 (Glucocorticoids in fear conditioning I). A and B are either tone or clicker, and X and Y are either DEX in a dose according to group label or SAL (all counterbalanced). + indicates reinforcement by mild footshock.

PHASE GROUP DEX 0.3 DEX 0.6 DEX 1.2

Conditioning 1

Conditioning 2

Extinction

Day 1

Day 2

Day 3

A+ & Treatment X

B+ & Treatment Y

10 x A 10 x B

Results Figure 2.1.1 shows the effect of post-training DEX on freezing during extinction. For illustration purposes, the results are collapsed across all doses of DEX. Extinction of freezing was less pronounced during the CS that was followed by DEX as compared to SAL after conditioning. A three-way, 3 x 2 x 10, mixed ANOVA with one between groups factor, DOSE (0.3 mg/kg, 0.6 mg/kg, 1.2 mg/kg) and two repeated measures, TREATMENT (SAL, DEX) and TRIAL (extinction trials 1-10), was carried out with freezing during the CS period as the dependent variable. There were no significant main effects of DOSE, F(2, 21) < 1, or TREATMENT, F(1, 21) < 1. There was a significant main effect of TRIAL, F(9, 189) = 3.89, p < .001, with freezing diminishing over trials, providing evidence for extinction taking place. The interaction of TREATMENT x TRIAL, was significant, F(9, 189) = 2.26, p = .02. Simple-effect analysis revealed a significantly higher level of freezing to DEX stimuli (M = 4.63), than SAL stimuli (M = 2.88), on trial 7 of extinction, F(1, 21) = 7.45, p < .02. A trend in the same direction on trial 6 only closely failed to reach significance, F(1, 21) = 3.55, p = .07. Figure 2.1.2 compares the levels of freezing with different

76

Freezing (s) during 10s CS

6

*

5 4 3 2 1

DEX SAL

0 1

2

3

4

5

6

7

8

9

10

Extinction Trial

Figure 2.1.1. Mean time freezing (s) during presentations of aversive auditory conditioned stimuli (CS) in extinction where conditioning was followed either by DEX or SAL administration. * p < .02 (simple effect of drug treatment at trial)

77

.


6 5 4 3 2 1

SAL (n=24) DEX 0.3 mg/kg (n=8) DEX 0.6 mg/kg (n=8) DEX 1.2 mg/kg (n=8)

0 1&2

3&4

5&6

7&8

9&10

Extinction Trial (2-trial bins)

Figure 2.1.2. Dose-response-curve for DEX for mean time freezing (s) across extinction. DEX scores for the different doses are compared with overall SAL scores.

78

doses of DEX, by plotting a dose-response-curve with the different dose groups’ DEX scores in contrast to the overall baseline freezing level to SAL. Although no significant effects involving the between group factor DOSE were found, it appears as if the higher doses were more effective

Discussion DEX enhanced memory in a discrete-cue aversive conditioning paradigm as shown by slowed extinction of freezing behaviour with significantly greater levels of freezing at extinction trial 7 (and marginally, trial 6). The observation that freezing did not differ between the DEX and the SAL condition at the beginning of the extinction session may well be due to a ceiling effect, with this measure not being sensitive enough to pick up a difference at those levels. There was no effect of dose, probably due to a large variance. However, the effect reported seemed to largely occur at the higher doses of DEX, 0.6 mg/kg and 1.2 mg/kg.

2.2 GLUCOCORTICOIDS IN DISCRETE-CUE FEAR CONDITIONING II

In order to confirm the results obtained in Experiment 2.1, it was decided to carry out a further discrete-cue fear conditioning study. Freezing as a measure of memory relies on a rather subjective interpretation by the rater, and often creates substantial variance. There are also debates about whether all absence of movement really reflects freezing as it can easily be confused with almost motionless sniffing and exploring behaviour. At the same time, an anxious rat may display fear-induced movements, particularly at the onset of an aversive CS. In order to replicate and

79

extend the generality of Experiment 2.1, Experiment 2.2 employed a conditioned suppression paradigm. In this study, the measure of memory is suppression of an ongoing instrumental behaviour during the presentation of a stimulus that was previously paired with an aversive reinforcer such as footshock. The highest dose of DEX (1.2 mg/kg) employed in the previous experiment was used, and a betweensubject design was implemented.

Method Subjects Subjects were 16 male, Lister Hooded rats (Harlan, UK; n = 8/group), weighing between 402 g and 470 g (M = 433 g) at the beginning of the experiment (conditioning session). For housing and feeding information, refer to Experiment 2.1. All procedures were carried out between 09:00 and 12:00. Apparatus Eight identical operant chambers (25-cm-wide x 25-cm-high x 22-cm-deep, Paul Fray Ltd., Cambridge, UK), housed in light- and sound-attenuating boxes were used. Each chamber was fitted with a retractable lever, located on the left side of a central, recessed magazine that provided access, via a hinged Plexiglas panel, to food reinforcement (45 mg, Rodent Grain-Base Formula; BIO-SERV, USA) delivered by a pellet dispenser. The floor of the chambers consisted of 18, 5-mm diameter steel rods spaced 1.5 cm apart centre-to-centre, perpendicular to the front wall of the chamber. The grids were connected to a Coulbourn precision-regulated animal shocker (model E13-14; Coulbourn Instruments, USA) set to deliver scrambled footshock as US (0.5 mA, 0.5 s). A 3-kHz tone produced by a lab-manufactured tone generator and delivered through a speaker mounted in the ceiling of the chamber, provided a

80

discrete auditory stimulus in each chamber. This produced a sound level of 75 dB. The background noise in the chamber was 55 dB. Two BBC Master 128 microcomputers (Acorn, UK), equipped with a SPIDER extension for on-line control (Paul Fray Ltd., UK) controlled the equipment and recorded the data. Drugs and injection procedure DEX (Sigma, UK) was injected sc, immediately post-training, at a dose of 1.2 mg/kg in a volume of 6.0 ml/kg. DEX was first dissolved in absolute ethanol and then diluted in 0.9% saline in order to reach its final concentration. For the control condition, a 0.9% saline solution (SAL) with a matched final concentration of ethanol (4%) was used. Drugs were prepared freshly on the day of administration. Behavioural procedure Magazine and lever training. All animals received two, 30-min magazine training sessions, with the levers removed from the chambers, during which food pellet reinforcement was delivered according to a variable time (VT)-60 schedule (i.e. reinforcers were delivered at a rate varying around a mean of 60 s). In the next 30-min session, the lever on the left of the magazine was introduced into the chamber and pressing was reinforced under a variable interval (VI)-2 schedule (i.e. reinforcers were programmed to occur following the first response after intervals of time varying around a mean of 2 s) for 30 min. This pattern was repeated over the next four sessions with increasing reinforcement schedules of VI-15, VI-30 and finally two sessions of VI-60. A VI-60 schedule of reinforcement was maintained throughout the remainder of the experiment. Habituation. On the day prior to conditioning, rats received a 20-min habituation session in which two, non-reinforced, 30-s presentations of the tone occurred after 7 and 17 min. The session served to eliminate unconditioned suppression of lever pressing during the presentation of the tone

81

(Killcross & Robbins, 1993). Conditioning. On the next day, rats received a 10-min conditioning session in which a 30-s tone was presented after 7 min. The final 0.5 s of the tone coincided with the delivery of a 0.5-s footshock, delivered through the grid floor of the chamber. At the end of the session, rats were immediately injected with either DEX or SAL, and then returned to their home cage. On the day following conditioning, rats received a reminder session of lever-training. Extinction. On each of the next six days, rats received one 20-min extinction session. Non-reinforced presentations of a 30-s tone were given after 7 and 17 min. Lever pressing rates were recorded during the 30 s prior to the tone presentations (preCS period) and during the 30-s tone presentations (CS period). A suppression ratio for each trial was calculated by dividing the number of lever presses (LP) during the CS period by the total number of LP during the preCS and CS periods. Hence a value of 0.5 reflects no suppression and a value of 0.0 reflects complete suppression of lever pressing. Table 2.2.1 displays an outline of the experimental design. Statistical analysis Refer to Experiment 2.1.

Table 2.2.1. Design of Experiment 2.2 (Glucocorticoids in fear conditioning II). SAL = SAL-treated control group; DEX = DEX-treated group with dose in mg/kg; MT = magazine training; LT = lever training; VT = variable time schedule; VI = variable interval schedule; LL = left lever. Treatment according to group label.

PHASE GROUP SAL DEX 1.2

Magazine- & Lever Training Pre-experiment, Day 3 MT: VT-60 LT: VI-2, 15, 30, 60 LL ? Food

Habituation

Conditioning

Extinction

Day 1

Day 2 1x Tone ? Shock & Treatment

Day 4-8

2x Tone ? 0

82

2x Tone ? 0

Results Figure 2.2.1 illustrates the effects of post-training administration of DEX on conditioned suppression of lever pressing over the course of extinction. DEX significantly enhanced conditioned suppression. There was no difference between the two treatment groups in baseline lever pressing during the preCS period. PreCS Scores (LP/30 s). A two-way, 2 x 15, mixed ANOVA with the between-group factor TREATMENT (DEX, SAL) and the repeated measure TRIAL (preCS periods of all 15 trials) revealed that the groups did not differ, F(1, 14) < 1, (Ms: DEX = 6.08; SAL = 6.29). Suppression ratios were therefore considered an appropriate measure to assess conditioned suppression of lever pressing and constitute the dependent variable of all further analyses. Suppression ratios. Habituation and conditioning. Groups did not differ prior to treatment during habituation (HAB) and conditioning (COND) (see left panel, Figure 2.2.1). A two-way, 2 x 3, mixed ANOVA with one between groups factor, GROUP (pre-DEX, pre-SAL) and one repeated measure, TRIAL (HAB 1 + 2, single COND trial) found no main effect of GROUP, F(1, 12) < 1, (Ms: pre-DEX = 0.27, pre-SAL = 0.29), nor an interaction of GROUP x TRIAL, F(2, 24) < 1. A significant main effect of TRIAL, F(2, 24) = 8.94, p < .01, reflected a reduction of unconditioned suppression across trials. Newman-Keuls posthoc analysis revealed a lower suppression ratio in habituation trial 1 (M = 0.03) than in habituation trial 2 (M = 0.39) and the conditioning trial (M = 0.43), ps < .01. Extinction. A two-way, 2 x 12, mixed ANOVA with one between groups factor, TREATMENT (DEX, SAL) and one repeated measure, TRIAL (extinction trials 1-12), revealed a significant main effect of TREATMENT, F(1, 13) = 5.19, p < .05, with animals receiving DEX after conditioning displaying greater conditioned

83

Suppression Ratio (CS/preCS+CS)

Habituation & Conditioning

Extinction

0.6

0.6

0.5

0.5

0.4

0.4

0.3

0.3

0.2

0.2

0.1

DEX SAL

DEX SAL

0.0

0.1 0.0

1

2

C

1

Trial

2

3

4

5

6

Blocks of 2 Trials

Treatment

Figure 2.2.1. Suppression ratios (±SEM) of lever pressing in the different posttraining treatment groups (DEX and SAL) during habituation, conditioning and extinction. Treatment, as indicated, occurred after the single conditioning trial (C).

84

suppression, i.e. lower suppression ratios (M = 0.31), than animals receiving SAL (M = 0.42). There was also a significant main effect of TRIAL, F(11, 143) = 9.71, p < .001, with suppression ratios increasing over trials, reflecting extinction. The TREATMENT x TRIAL interaction approached significance, F(11, 143) = 1.79, p = .06, likely to reflect the observation that the difference between the groups was greater at the beginning of extinction and reduced as conditioned suppression in both groups extinguished.

Discussion Experiment 2.2 provides further evidence for a memory-modulatory role of glucocorticoids in aversive discrete-cue Pavlovian conditioning and hence supports the claims made in Experiment 2.1. Only one dose of DEX was applied (1.2 mg/kg) and it led to a significant increase in conditioned suppression of lever pressing, clearly evident on the first three days of extinction. Eventually both groups’ conditioned fear to the tone extinguished and their performance coincided at ceiling.

2.3 GLUCOCORTICOIDS IN APPETITIVE DISCRETE-CUE CONDITIONING I

While Experiments 2.1 and 2.2 demonstrated the role of glucocorticoids in the modulation of aversive learning, Experiments 2.3 and 2.4 were designed to investigate their role in appetitive designs. If glucocorticoid memory modulation is mediated by general arousal levels rather than the aversive nature of a stimulus or event as such, memory modulation should follow exogenous GR activation, even in

85

low-arousal, e.g. appetitive tasks (see introduction). In order to investigate this question, an appetitive discrete-cue Pavlovian conditioning paradigm was set up with auditory stimuli and food rewards as CS and US respectively. Glucocorticoids were administered immediately after both of two conditioning sessions, and memory was assessed during conditioning and extinction on days following treatment.

Method Subjects Subjects were 24 male, Lister Hooded rats (Harlan, UK; n = 6/group), weighing between 247 g and 272 g (M = 262 g) at the start of the experiment. For housing and feeding information, refer to Experiment 2.1. All experiments were carried out between 09:00 and 12:30. Apparatus Eight identical operant chambers (30.5-cm-wide x 24.1-cm-high x 21.0-cmdeep, Med Associates Inc., USA), housed in light- and sound-attenuating boxes were used in the experiment. Each chamber contained a central, recessed magazine that provided access to food reinforcement (45 mg, Formula A/I; Noyes, USA) delivered by a pellet dispenser. An infrared emitter and receiver in the magazine allowed assessment of the number of magazine entries and the length of time spent in the magazine. A 10-Hz train of clicks (clicker) with a sound level of 75 dB was available as discrete auditory cue. It was produced by a heavy-duty relay mounted behind the wall opposite the food tray. The background noise in the chamber was 60 dB. A PC equipped with Med-PC software (Med Associates Inc, Vermont, USA) controlled the equipment and recorded the data.

86

Drugs and injection procedure DEX (Sigma, UK) at doses of 0.3, 0.6 or 1.2 mg/kg in a volume of 2.0 ml/kg or SAL was injected sc, immediately post-training. The doses were based on those used in the first fear-conditioning experiment (2.1). For preparation detail, refer to Experiment 2.2. Behavioural procedure Magazine training. On each of the three days prior to the experiment, rats were given a 30-min session of magazine training during which food pellet reinforcement was delivered according to a VT-60 schedule. Conditioning. On days 1 and 2, animals were given conditioning sessions with post-training treatment. Six conditioning trials occurred after 7, 9.5, 12, 14.5, 17 and 19.5 min during a session lasting 22 min. Each trial comprised a 10-s presentation of the CS, clicker, terminating with the delivery of the US, four food pellets. Magazine approach, defined as number of entries, during the CS presentations, and matched preCS periods, was recorded. Drug treatment occurred immediately after the session, approximately 3-6 min after the last stimulus presentation. Animals were returned to their home cage immediately after this treatment. Extinction. On days 3 and 4, extinction test sessions took place, comprising 10 trials per session and lasting 30 min. CS presentations were equivalent to those in the conditioning session, except that no reinforcement was given. Trials were equally spaced, starting 2 min into the session, with an ISI of 2 min 50 s. Again, magazine approach during the CS and preCS periods was recorded. See Table 2.2.1 for an outline of the experimental design. Statistical analysis Refer to Experiment 2.1.

87

Table 2.3.1. Design of Experiment 2.3 (Glucocorticoids in appetitive conditioning I). SAL = SAL-treated control group; DEX = DEX-treated group with dose in mg/kg; MT = magazine training; VT = variable time schedule. Treatment according to group label. PHASE GROUP SAL DEX 0.3 DEX 0.6 DEX 1.2

Magazine Training

Conditioning

Extinction

Pre-training

Day 1-2

Day 3-4

MT: VT-60

6x Clicker ? Food & Treatment

10 x Clicker ? 0

Results Figure 2.3.1 illustrates the effects of different doses of post-training administration of DEX on appetitive discrete-cue Pavlovian conditioning in different stages of the experiment. The highest dose of DEX significantly enhanced conditioned magazine approach on those days following treatment. Both for magazine entries during the preCS period, and for magazine entries during the CS minus the preCS period, a three-way, 4 x 2 x 2, mixed ANOVA was carried out with TREATMENT as the between-subject factor (SAL, DEX0.3, DEX0.6, DEX1.2), and PHASE (COND, EXT) and SESSION (1 and 2) as repeated measures. PreCS period. There was no significant main effect of TREATMENT, F(3, 20) = 1.24, p = .32 (Ms: SAL = 0.71, DEX0.3 = 0.61, DEX0.6 = 0.72, DEX1.2 = 0.38), no significant interactions of TREATMENT x PHASE or TREATMENT x SESSION, nor a significant three-way interaction (all Fs < 1). There was a marginally significant main effect of PHASE, F(1, 20) = 3.98, p = .06, but not SESSION, F(1, 20) = 1.56, p = .23, with less preCS responding during the extinction (M = 0.44) than the conditioning phase (M = 0.77).

88

Magazine Approach (CS-preCS)

Conditioning

Extinction

5

5

*

4

DEX 1.2mg/kg DEX 0.6mg/kg DEX 0.3mg/kg SA L

3

4 3

† 2

2

1

1

0

0

-1

-1 1

2

1

2

Session

Treatment

Figure 2.3.1. Magazine entries (CS-preCS) (±SEM) in different post-training treatment groups, according to dose of DEX or SAL, during sessions of conditioning and extinction. Arrows indicate time of treatment, immediately after conditioning sessions 1 (trials 1-6) and 2 (trials 7-12). * p < .02, compared to all other groups. † p < .04, compared to groups SAL and DEX 0.3

89

CS-preCS period. A significant main effect of TREATMENT was found, F(3, 20) = 3.67, p < .03. Newman-Keuls posthoc analysis revealed that conditioned magazine approach was significantly higher in the DEX 1.2 (M = 1.52) than in the SAL (M = 0.31) and DEX 0.3 (M = 0.44) groups (ps < .04), and this difference approached significance (p = .08) with respect to the DEX 0.6 group (M = 0.77).A PHASE x SESSION interaction, F(1, 20) = 30.10, p < .001, confirmed the occurrence of conditioning and extinction. Simple effects analysis revealed that responding was significantly higher during session 2 of conditioning (M = 1.81) than during session 1 (M = 0.15), p < .001, whereas responding during session 1 of extinction (M = 0.84) was significantly higher than during session 2 (M = 0.24), p < .01. Furthermore, there was a three-way interaction of TREATMENT x PHASE x SESSION, F(3, 20) = 3.55, p < .04. Individual one-way analyses of TREATMENT were carried out for each of the conditioning and extinction sessions. During conditioning session 1, which occurred before the first treatment, there was no significant effect of TREATMENT, F(3, 20) < 1 (Ms: SAL = -0.22, DEX 0.3 = 0.28, DEX 0.6 = 0.19, DEX 1.2 = 0.36), confirming homogenous pre-treatment responding. During conditioning session 2, which occurred after the first treatment, there was a significant main effect of TREATMENT, F(3, 20) = 5.44, p < .01. Newman-Keuls posthoc analysis revealed that conditioned magazine approach was significantly higher in the DEX 1.2 group (M = 3.64) than in any other, p < .02 (Ms: SAL = 1.00, DEX 0.3 = 1.11, DEX 0.6 = 1.47). A significant main effect of TREATMENT was also found in extinction session 1, F(3, 20) = 4.01, p < .03. Newman-Keuls posthoc analysis showed significantly higher responding in the DEX 1.2 (M = 1.65) than in the SAL (p < .03) and the DEX 0.3 (p < .04) group (Ms: SAL = 0.67, DEX 0.3 = 0.33, DEX 0.6 = 0.72).

90

Discussion Experiment 2.3 demonstrated memory modulation, i.e. enhancement, of posttraining glucocorticoids in an appetitive discrete-cue conditioning paradigm. It constitutes one of the very few studies (e.g. Micheau, 1981, 1985) on the role of glucocorticoids in memory consolidation employing appetitive designs, and, as far as I am aware, it is the first experimental demonstration of glucocorticoids modulating learning about discrete cues predicting an appetitive US. This is important for a number of reasons. First, if the effects of post-training glucocorticoids were limited to aversive conditioning, one might argue that the administration of DEX could be aversive in itself, adding to the magnitude of the reinforcer’s aversive properties, in other words making the US more salient, and enhancing fear conditioning in this manner. This argument could, for example, be applied to Experiment 2.2, or to some of the inhibitory avoidance experiments by the Roozendaal and McGaugh group. If this was the case, however, glucocorticoids would not enhance but impair appetitive learning, creating conflicting counterconditioning associations with the CS, i.e. both pleasant and unpleasant consequences. However, since appetitive conditioning was enhanced by post-training administration of DEX, Experiment 2.3 strongly suggests that DEX effects are on the modulation of memory consolidation. Second, the results suggest that the modulation of learning about both aversive and appetitive events is underpinned by a common system, suggesting that arousal itself, be it caused by an event of either appetitive or aversive valence, triggers a cascade of physiological events such as the release of endogenous hormones, and hence plays a key role in this modulation. Finally, as the procedure employed discrete cues, as in Experiments 2.1 and 2.2, the results give further support to the proposal that the role of glucocorticoids in memory is not limited to contextual conditioning and forming a representation of

91

the context (see Pugh et al., 1997a,b). See the general discussion of this chapter for a more detailed analysis of the results and its implications.

2.4 GLUCOCORTICOIDS IN APPETITIVE DISCRETE-CUE CONDITIONING II

Experiment 2.3 demonstrated that glucocorticoids could modulate memory in an appetitive conditioning paradigm. In order to confirm these results, and, as in the aversive domain, to obtain evidence from two independent studies, a further experiment was carried out. Again, rats underwent appetitive conditioning in which an auditory stimulus predicted a food reward. Training was followed by treatment with either DEX or SAL, and memory was tested on days subsequent to treatment. This time, no dose-response curve was established but only one dose of DEX (1.2 mg/kg), highly effective in Experiment 2.3, was applied.

Method Subjects Subjects were 32 male, Lister Hooded rats (Harlan, UK; n = 16/group), weighing between 282 g and 371 g (M = 320 g) at the start of the experiment. For housing and feeding information, refer to Experiment 2.1. All experiments were carried out between 09:00 and 12:30.

92

Apparatus The apparatus was the same as in Experiment 2.3. Clicker (10 Hz, 75 dB, 10 s) served as auditory cue with the background noise being 60 dB. Food pellets (45 mg, Formula A/I; Noyes, USA) provided the appetitive US. Drugs and injection procedure DEX (Sigma, UK), at a dose of 1.2 mg/kg in a volume of 6.0 ml/kg, or SAL was injected (sc) immediately post-training. For preparation details, refer to Experiment 2.2. Behavioural procedure The behavioural procedure was the same as in Experiment 2.3. See Table 2.4.1 for an outline of the experimental design. Statistical analysis Refer to Experiment 2.1.

Table 2.4.1. Design of Experiment 2.4 (Glucocorticoids in appetitive conditioning II). SAL = SAL-treated control group; DEX = DEX-treated group with dose in mg/kg; MT = magazine training; VT = variable time schedule. Treatment according to group label.

PHASE GROUP SAL DEX 1.2

Magazine Training

Conditioning

Extinction

Pre-training

Day 1-2

Day 3-4

MT: VT-60

6x Clicker ? Food & Treatment

10 x Clicker ? 0

93

Results Figure 2.4.1 illustrates the effects post-training administration of DEX on appetitive Pavlovian conditioning. Data are collapsed into blocks of trials that occurred before (left panel) and after (right panel) post-training treatment. DEX treatment significantly enhanced conditioned magazine approach. For both magazine entries during the preCS period, and magazine entries during the CS minus the preCS period, analyses were carried out for those trials occurring prior to treatment, i.e. conditioning trials 1-6, and for those trials occurring after treatment, i.e. conditioning trials 7 to 12 and all extinction trials. PreCS scores / pre-treatment. A two-way, 2 x 6, mixed ANOVA was carried out with TREATMENT (DEX, SAL) as the between-subject factor and TRIAL (conditioning trials 1-6) as repeated measure. There were no significant main effects of TREATMENT, F(1, 30) = 1.32, p = .26, (Ms: SAL = 1.10, DEX = 0.78), or TRIAL, F(5, 150) < 1, nor a significant interaction of the two, F(5, 150) < 1, confirming that preCS scores did not differ between groups prior to treatment. PreCS scores / post-treatment. A two-way, 2 x 26, mixed ANOVA was carried out with TREATMENT (DEX, SAL) as the between-subject factor and TRIAL (conditioning trials 7-12, extinction trials 1-20) as repeated measure. There were no significant main effects of TREATMENT, F(1, 30) = 1.66, p = .21, (Ms: SAL = 0.83, DEX = 0.61), or TRIAL, F(25, 750) = 1.38, p = .10, nor a significant interaction of TREATMENT x TRIAL, F(25, 750) < 1. These data confirm that there were no differences in preCS responding following treatment. CS-preCS / pre-treatment. A two-way, 2 x 6, mixed ANOVA was carried out with TREATMENT (DEX, SAL) as the between-subject factor and TRIAL (conditioning trials 1-6) as repeated measure. There was no significant main effect of

94

Magazine Approach (CS - preCS) preCS)

2.0

2.0

* 1.5

1.5

1.0

1.0

0.5

0.5

0.0

0.0

SAL

DEX

SAL

Pre-Treatment Trials

DEX

Post-Treatment Trials

Figure 2.4.1. Magazine approach (CS – preCS) during conditioning and extinction trials prior to (conditioning trials 1-6) and following (conditioning trials 7-12, extinction trials 1-20) post-training treatment with DEX or SAL. * p < .05 as compared to the SAL control group.

95

TREATMENT, F(1, 30) < 1, (Ms DEX = 0.57, SAL = 0.59) and no significant interaction of TREATMENT x TRIAL, F(5, 150) < 1, confirming that groups did not differ prior to treatment. A significant main effect of TRIAL, F(5, 150) = 4.97, p < .001, indicated that conditioning took place. Newman-Keuls posthoc analysis revealed significantly greater conditioned magazine approach on trial 4 than on trial 1, and on trial 6 than on trials 1 to 3 (ps < .03). CS-preCS / post-treatment. A two-way, 2 x 26, mixed ANOVA was carried out with TREATMENT (DEX, SAL) as the between-subject factor and TRIAL (conditioning trials 7-12, extinction trials 1-20) as repeated measure. There was a significant main effect of TREATMENT, F(1, 30) = 7.40, p < .02, with greater levels of magazine approach in the DEX (M = 1.52 ) than in the SAL group (M = 0.73), confirming that post-training DEX can enhance memory. A significant main effect of TRIAL, F(25, 750) = 5.15, p < .001, mainly indicated extinction. There was no significant interaction of TREATMENT x TRIAL, F(25, 750) < 1.

Discussion Experiment 2.4 confirmed the results of Experiment 2.3, that glucocorticoid memory modulation could be demonstrated in appetitive conditioning, and provided extra confidence in them. Only one dose of DEX was applied (1.2 mg/kg) and it led to a significant increase in conditioned responding in post-treatment trials.

2.5 GENERAL DISCUSSION Post-training administration of the GR agonist DEX enhances learning in both aversive (Experiments 2.1 and 2.2) and appetitive (Experiment 2.3 and 2.4) discrete-cue Pavlovian conditioning paradigms. For both motivational domains, a

96

dose of 1.2 mg/kg of DEX enhanced memory on those trials of conditioning and extinction that occurred on the days after treatment. In none of the experiments was a difference found between groups during the first conditioning session, confirming homogenous responding prior to treatment. Treatment only occurred after conditioning, so that subjects were drug-free at the time of exposure to the CS and US in acquisition, as well as during exposure to the CS in extinction. This suggests an effect on memory consolidation unconfounded by possible effects on attentional, motivational or sensory-perceptual mechanisms at the time of conditioning or test, and excludes the possibility of state dependent effects. Since DEX has a much higher affinity for GRs than for MRs, it is likely that this modulation is mediated through central activation of GRs. In natural circumstances, increased GR activation only occurs in situations of arousal and during the circadian peak. These findings are consistent with work from other laboratories (e.g. Flood et al., 1978; Micheau et al., 1981, 1985; Veldhuis et al., 1985; Oitzl & de Kloet, 1992; Sandi & Rose, 1994a,b; Roozendaal & McGaugh, 1996a,b, 1997a). However, the present results extend previous findings, providing the first demonstrations of memory modulation by systemic post-training glucocorticoid administration in discrete-cue Pavlovian conditioning paradigms, of both appetitive and aversive nature. As DEX does not enter the brain very readily (e.g. de Kloet, 1997) but suppresses the release of endogenous corticosterone by binding to pituitary GRs, one could alternatively argue (see Chapters 1 and Appendix) that its effects may come about peripherally, by creating a state of functional adrenalectomy and reducing the release of endogenous ACTH and corticosterone. Hence, the memory enhancing effects might have been due to lower levels of ACTH, or lower rather than higher levels of central corticosterone. Furthermore, as CRH release from the hypothalamus

97

is not inhibited and can even be increased given the lack of negative feedback from endogenous corticosterone, one might also argue that CRH mediated the memory modulatory effects of DEX. CRH has been shown to enhance learning (Radulovic et al., 1999). However this possibility seems highly unlikely in procedures using an acute manipulation for a number of reasons. Despite its poor ability to cross the blood brain-barrier (de Kloet, 1997), some DEX still enters the brain (de Kloet et al., 1975), allowing for central effects. This would explain why relatively high levels of DEX had to be used in this experiment in order to demonstrate an effect. Central DEX blocks ADX-induced CRH increases (Sawchenko, 1991), implying opposing peripheral and central effects on CRH. It seems unlikely that the results arose from delayed peripheral DEX effects on corticosterone levels at retrieval, given 24h (Experiments 2.1, 2.3, 2.4) or even 48h (Experiment 2.2) delays. Furthermore, similar effects of memory enhancement have been demonstrated with both corticosterone and intracerebral infusions of GR agonists (e.g. Cordero & Sandi, 1998; Roozendaal & McGaugh, 1997a,b), as well as with post-training psychosocial stress (Liu et al., 1999). Finally, intracerebral infusions of GR antagonists (Roozendaal & McGaugh, 1997a) and systemic injections of the corticosterone synthesis inhibitor metyrapone (Roozendaal et al., 1996a), both of which mimic a state of ADX, impair rather than enhance memory, with the latter effect being attenuated by simultaneous systemic administration of DEX. The fact that memory modulation could be demonstrated in an appetitively motivated paradigm has important implications. First, it argues strongly against any possible assertion that the effects of post-training DEX may be due to a potential aversive nature of DEX itself. If, as a consequence of an aversive effect of DEX as compared to SAL, the US becomes more aversive, now consisting of a shock and any

98

unpleasant experience of DEX, rather than shock alone, this in itself could explain enhanced levels of expression of fear conditioning. However, there is little evidence for any aversive properties of DEX itself, and the subjective effects of the adrenocortical hormone cortisol in humans, in contrast to those of the adrenomedullary hormone adrenaline, are essentially unnoticeable (D. J.-F. de Quervain, personal communication, April 2001). Furthermore, it is unlikely that possible aversive effects of DEX, following its administration at the end of the conditioning session, would be associated specifically with a discrete cue occurring during that session. Even though all this suggests that the assertion is not a strong one, it is useful to rule it out experimentally. If the assertion was true, glucocorticoids would not enhance but impair appetitive learning. The results of Experiments 2.3 and 2.4 suggest otherwise, confirming an effect of memory modulatory processes. Second, HPA activation and subsequent increases in glucocorticoids often reflect the presence of aversive stressors. However, salient appetitive stimuli can also trigger the release of glucocorticoids (Piazza & Le Moal, 1997). For example, Merali et al. (1998) found that a meal-elicited CRH-release in the CEA in rats was as great as that associated with 20 min of restraint, and that levels of circulating ACTH and corticosterone increased in response to both appetitive and aversive stimuli. They suggested that “...Rather than evoking feelings of fear and anxiety, these systems may serve to draw attention to events or cues of biological significance, such as those associated with food availability as well as those posing a threat to survival” (Merali et al., 1998, p. 4758). Hence, learning about appetitive and aversive motivationally significant stimuli in the environment may be served by a common system of modulation by glucocorticoids.

99

Glucocorticoid modulation may occur more readily following aversive rather than appetitive events, as the former may result in greater levels of arousal than the latter. In the appetitive experiments presented here, a high enough level of phasic arousal was simulated pharmacologically by administering a sufficient dose of DEX. This might explain why Roozendaal and McGaugh (1996a,b) found memoryenhancing effects of DEX in an aversive inhibitory avoidance paradigm with a dose as low as 0.3 mg/kg. The aversive nature of this inhibitory avoidance task is likely to generate higher arousal levels with a greater release of endogenous corticosterone than may be the case with a food pellet reward. It has previously been demonstrated that different levels of reinforcers, for example varying levels of footshock, lead to different increases in plasma hormone levels (McCarty & Gold, 1981; Gold & McCarty, 1981). However, the main point of this project is not to examine which precise doses of exogenous DEX in combination with levels of endogenous corticosterone triggered by different reinforcers lead to modulation of memory consolidation. Rather, the important message is that memory modulation by glucocorticoids can be shown in conditioning paradigms employing both appetitive and aversive reinforcers. Another important conclusion to be drawn from the current experiments is that the role of glucocorticoids in memory modulation does not seem to be restricted to contextual conditioning. All experiments presented here differ from most previous studies in that they employed a Pavlovian conditioning paradigm that relied on discrete rather than contextual cues. It has been suggested by Pugh et al. (1997a,b) that the role of glucocorticoids, at least in fear conditioning, is selective for forming a long-term representation of the context. In a series of experiments, Rudy and coworkers (Rudy & Pugh, 1996, 1998; Rudy et al., 1999; Pugh et al., 1997a,b)

100

investigated factors that affect contextual- but not discrete-cue auditory fear conditioning. For example, post-training isolation, time of day, ADX, and pre- and post-training treatment with a GR antagonist have been found to affect and impair retention of contextual fear conditioning, and all of these treatments could be blocked by preexposure to the context prior to conditioning. While Rudy et al. (1999) proposed that the effect of isolation stress is probably not mediated by glucocorticoids but opioids, they have linked the fact that conditioning is worse when taking place at 12:00 than at 08:00 and 16:00 to circadian patterns of plasma glucocorticoid levels (Rudy & Pugh, 1998). However, Rudy and Pugh acknowledge that this hypothesis is problematic, given that the behavioural results do not match the circadian rhythm of endogenous corticosterone release. The well-documented inverted U-shape function of glucocorticoids on memory consolidation would lead to a prediction opposite to the observation that medium levels of glucocorticoids at 12:00 result in lower levels of conditioning than both low and high levels at 08:00 and 16:00 respectively. The impairing effects of ADX (Pugh et al., 1997b), reversed by corticosterone supplement, and the effects of GR antagonists given just before or after conditioning (Pugh et al., 1997a), can be ascribed to glucocorticoid mediation more directly. As ADX and GR antagonists did not affect auditory-cue fear conditioning, a selective role of corticosterone in contextual conditioning was proposed. As Pugh and Rudy (1997b) point out, contextual fear conditioning relies on two processes: a) the formation of a representation of the context and b) the association between this representation and the aversive consequences of the US (e.g. shock) (Fanselow, 1990; Rudy & Pugh, 1996; Kiernan & Westbrook, 1993; Young et al., 1994). Since preexposure to the context (24h before ADX) reversed the impairing effects of ADX on contextual conditioning, Pugh et al. (1997b) furthermore argued that

101

“...corticosterone contributes to the processes that construct a memory representation of the context but not to the processes that associate this representation with shock.” (Pugh et al., 1997b, p.507). At first sight, these results seem to contradict those presented in this chapter. The claim that glucocorticoids do not play a role in discretecue conditioning should be viewed in the light of the second claim, that corticosterone does not modulate motivationally salient associative processes but rather the formation of the context representation. In fact, Pugh et al. (1997a,b) did not investigate the full memory modulatory potential of glucocorticoids as such. Comparing performance in ADX and ADX+CORT replacement groups does not investigate the effects of an acute release of glucocorticoids due to stress. Corticosterone replacement took place in the rats’ drinking water on a regular basis and merely resulted in the presence of tonic corticosterone during conditioning. The extremely low doses of corticosterone replacement (2 mg/kg; Flood et al. (1978) applied a dose of 30 mg/kg) that were effective in blocking the ADX-induced impairments when given directly after conditioning, suggest that the possible effects of acute stress on memory were not being investigated. Low shock levels and other experimental parameters may account for the lack of a difference between the control and the ADX+CORT replacement group. The same line of reasoning applies to the study employing the GR antagonist RU38486. Again, normal conditioning was impaired by GR blockade and the effect was selective for contextual conditioning. However, higher shock levels and a higher dose of RU38486 might have resulted in different results. All in all, it might be the case that glucocorticoids play an important role in the formation of a context representation, as suggested by Pugh et al. (1997a,b), and this theory might even be able to explain some other memory

102

modulation results, e.g. those employing the inhibitory avoidance paradigm (e.g. Roozendaal & McGaugh, 1996a, 1997a). However, the results presented in this chapter show that glucocorticoids can also modulate memory consolidation in discrete-cue Pavlovian conditioning paradigms. As mentioned above, some of the differences may be due to the paradigm and the specific parameters used. For example, it could be possible that glucocorticoid modulation of discrete-cue conditioning occurs only at a particular threshold of arousal, and that this threshold was not achieved by the aversive US in Pugh et al.’s (1997a,b) studies. The procedures used in this project, in contrast, involved an experimental increase of GR activation by administration of an exogenous glucocorticoid agonist. Further, in the current experiments, contextual conditioning, or modulation of contextual conditioning, was neither anticipated nor observed. The use of Pavlovian CSs in these procedures provided extremely effective, discrete cues for reinforcement, overshadowing the acquisition of context-reinforcer associations (Rescorla & Wagner, 1972). As both procedures were optimised for discrete-cue learning, it is scarcely surprising that no effects of contextual conditioning were observed. At the same time, Pugh et al.’s (1997a,b) procedures may have been sensitive to the role of glucocorticoids, possibly in tonic rather than stress-related levels, in the formation of a contextual representation, but less so to their acute role in modulating memory consolidation of an association between a stimulus and an aversive or appetitive event. Pugh et al. (1997a,b) argue from their findings that the hippocampus constitutes the neural locus at which glucocorticoids exert their role in fear conditioning. Most parts of the hippocampus, particularly the DG, are indeed rich in GRs (Reul & de Kloet, 1985). It should be noted however that other brain regions

103

such as the central amygdala, the lateral septum, and the NTS also have high levels of GRs, and possess a greater ratio of GRs to MRs relative to the hippocampus (Reul & de Kloet, 1985). Glucocorticoids have been shown to enhance memory when being administered directly into the hippocampus (e.g. Cottrell & Nakajima, 1977; Micheau et al., 1981, 1985; Roozendaal & McGaugh, 1997b; Roozendaal et al., 1999a), but this is also true for microinfusions of glucocorticoids into other brain regions such as the basolateral amygdala (Roozendaal & McGaugh, 1997a), the NTS (Roozendaal et al., 1999b), and the NAC (Roozendaal, 2000). Whilst the hippocampus may play an important role in glucocorticoid effects on memory, a model solely based on hippocampal functioning, proposing a selective role of glucocorticoids in modulating the consolidation of the representation of context, selectively acting in the hippocampus, is not sufficient to explain the current data. Rather, it seems that glucocorticoids modulate a wider range of types of memory representations, including discrete cues, and their associations with emotionally salient events, in multiple brain regions. This modulation may, at least in part, be mediated by the basolateral amygdala (Quirarte et al., 1997; Roozendaal & McGaugh, 1997b, Roozendaal et al., 1999a,b; Roozendaal et al., 2001a; Setlow et al., 2000). Finally, the results reported here are consistent with a number of experiments involving emotional learning in human subjects. For example, a study by Bradley et al. (1992) supports the notion that arousal plays an important part in memory. They assessed immediate and delayed free-recall of pictures varying both in arousal (lowhigh) and affective valence (unpleasant-pleasant), and found that only arousal had a stable effect on memory, with highly arousing pictures, both pleasant and unpleasant, leading to enhanced performance.

104

In summary, memory enhancement by systemic post-training glucocorticoid agonist administration was demonstrated in discrete Pavlovian cue conditioning paradigms of both appetitive and aversive nature. The results support previous claims proposing a memory modulatory role of adrenocortical stress hormones, and suggest that this modulation is not limited to particular aversive learning situations, like contextual fear conditioning. Glucocorticoids seem to modulate memory for a variety of emotionally significant events or cues. On the basis of lesion and microinfusion studies (e.g. Roozendaal, 2000; Chapter 1), one might speculate that this modulation occurs in multiple brain regions, with the amygdala possibly playing a central coordinating role.

105

CHAPTER 3 Do glucocorticoids modulate associations lacking a discrete US ?

106

In Chapter 2, it was shown that glucocorticoids modulate memory consolidation for discrete cues in both aversive and appetitive tasks. These forms of learning involved direct experience of an aversive or appetitive event in the form of a discrete US, for example an electric shock or food. However, there are other forms of learning in which no discrete US is experienced. Learning can occur about associations between merely sensory stimuli, or about the relationship between a stimulus and the absence of a reinforcer. Chapter 3 contains a series of four experiments that aimed to investigate the question of whether glucocorticoids can also modulate learning that occurs in the absence of a discrete US. Experiments 3.1 (latent inhibition) and 3.2 (extinction) employed paradigms with CS – no US relationships, whereas Experiments 3.3 (sensory preconditioning) and 3.4 (secondorder conditioning) examined CS1-CS2 associations.

3.1 GLUCOCORTICOIDS IN LATENT INHIBITION

Latent inhibition (LI) is the retardation of acquisition of conditioned responding to a stimulus produced by repeated non-reinforced preexposure to that stimulus (Lubow & Moore, 1959; Lubow, 1973). It is considered by some to be an index of an individual’s ability to learn to ignore irrelevant stimuli. The mechanisms by which LI operates are, however, a matter of debate. Theories of LI that regard the effect to be a product of attention and associability (Lubow, Weiner, & Schnur, 1981; Pearce & Hall, 1980) assume an impairment of conditioning resulting from a decline of attention to, or associability of, the preexposed stimulus, perhaps in part due to its status as a consistent predictor of a null outcome. Theories of LI based on habituation

107

(e.g. Wagner, 1978, 1981) draw a parallel between habituation and LI in the sense that during both, associations between the CS and the context are formed. As a consequence, the CS is entirely predicted by the context and is no longer surprising, or worthy of attention, at the start of subsequent conditioning. Hence the stimulus elicits neither unconditioned orienting responses nor processing beneficial to conditioning. The difference between habituation and LI lies merely in the stage during which this reduced unconditioned behavioural expression is measured. However, Hall and Channel (1985) reported data that did not support such theories showing that a shift from one context at preexposure to a different but familiar context at test abolished LI but not habituation, whereas a 16-day retention interval between preexposure and test restored the orienting response but did not abolish LI (Hall & Schachtman, 1987). LI theories based on associative interference (Revusky, 1977; Kalat, 1977) propose that some form of learning during preexposure, for example a CS-context / context-CS or a CS-nothing association, interferes with later formations of CS-US associations during the conditioning stage. All theories proposing a failure at the CS-US encoding stage are challenged by findings that post-acquisition exposure to the US alone (Kasprow, Catterson, Schachtman, & Miller, 1984), or a 21-day delay from preexposure to testing (Kraemer & Roberts, 1984), reduces the magnitude of LI, showing that the CS-US association must have been fully formed (but see Killcross, 2001). In line with this, LI theories of retrieval failure (e.g. Miller, Kasprow, & Schachtman, 1986; Bouton, 1991) assume that learning of CS-US associations proceeds normally but has to compete with earlier associations formed during preexposure for behavioural expression at test. Theories based on habituation, associative interference and retrieval failure agree that some form of learning occurs during the preexposure stage. Killcross and Balleine (1996)

108

investigated more closely what is actually learned during that stage, manipulating the motivational state of the animal during preexposure. They concluded that repeated non-reinforced preexposure of a stimulus results in the animal learning that the stimulus is unrelated to events that are relevant to their current motivational state. In the context of this PhD, the question arose whether the type of learning that occurs during preexposure, most likely CS - nothing of importance associations, could be modulated by post-training (or in this case post-preexposure) administration of glucocorticoids, in the same way that glucocorticoids modulate learning about discrete motivationally significant events (Chapter 2). The only previous studies to date investigating the effects of glucocorticoids on LI (Shalev, Feldon, & Weiner, 1998; Shalev & Weiner, 2001) employed corticosterone administration prior to preexposure and conditioning and reported a disruptive effect of corticosterone on LI. As LI may not operate by attentional or associability mechanisms, but might rely on learning during preexposure, the questions and hypotheses pursued in the current project were different. If learning of some kind occurs during preexposure, as suggested by substantial evidence, and if this learning is susceptible to glucocorticoid memory modulation, treatment with the GR agonist DEX after preexposure may result in greater interference during conditioning or greater competition for behavioural expression, and hence an enhancement of LI. Alternatively, glucocorticoid memory modulation may depend on the presence of a discrete motivationally relevant reinforcer, so that DEX treatment may not have an effect. The aim of Experiment 3.1 was to pursue this question.

109

Method Subjects Subjects were 32 male Lister Hooded rats (Harlan, UK; n = 8/group), weighing between 310 g and 425 g (M = 358 g) at the beginning of the experiment (preexposure stage). For housing and feeding information, refer to Experiment 2.1. All procedures were carried out between 8:00 and 12:00. Apparatus Refer to Experiment 2.2. A retractable lever was located to the right of the magazine. Food pellets (45 mg, Formula A/I; Noyes, USA) provided positive reinforcement. In addition to a tone (3 kHz, 75 dB, 30 s), a 10-Hz train of clicks (clicker) with a sound level of 75 dB was available as another discrete auditory cue. It was produced by a heavy-duty relay mounted behind the wall on the side of the food tray. The background noise in the chamber was 55 dB. Footshock (0.5 mA, 0.5 s) served as the aversive US. Drugs and injection procedure DEX (Sigma, UK), at doses of 0.3, 0.6 or 1.2 mg/kg in a volume of 4.0 ml/kg, or SAL was injected (sc) immediately post-training (post-preexposure). For preparation details, refer to Experiment 2.2. Behavioural procedure Magazine and lever training. Animals received two, 30-min magazine training sessions, with the levers removed from the chambers, during which food pellet reinforcement was delivered according to a variable time 30-s schedule. In the next 30-min session, the lever on the right of the magazine was introduced into the chamber and pressing was reinforced under a VI-2 schedule for 30 min. Animals received between one and six VI-2 sessions until increasing reinforcement schedules

110

of VI-15 (one session), VI-30 (two to three sessions) and finally VI-60 (one session) were introduced. The number of sessions required to achieve a stable response baseline varied between individual animals. Additional magazine training sessions were given when necessary. Baseline response rates did not differ between treatment groups at the end of lever training (data not shown). A VI-60 schedule of reinforcement was maintained throughout the remainder of the experiment. Preexposure. On day 1 after lever training, animals received a 40-min preexposure session during which 12 non-reinforced, equally spaced presentations of either tone or clicker (A) occurred. This level of preexposure was chosen to allow evidence of enhanced LI (Killcross, Dickinson, Robbins, 1994a,b). Immediately after the session, rats received pharmacological treatment of either SAL or one of three different doses of DEX, and were then returned to their home cage. Habituation. On day 2, the day prior to conditioning, rats received a 40-min habituation session in which two nonreinforced, equally spaced 30-s presentations of each, tone and clicker occurred in a pseudo-randomised order (A-B-B-A or B-A-A-B) after 8, 16, 24 and 32 min, in order to eliminate unconditioned suppression. The use of tone and clicker during preexposure and habituation was counterbalanced in all treatment groups. Conditioning. On each of days 3 to 5, rats received a 40-min conditioning session in which a 30-s tone and a 30-s clicker were presented equally spaced after 13 and 26 min. The final 0.5 s of each stimulus coincided with the presentation of a 0.5-s footshock, delivered through the grid floor of the chamber. The order in which the stimuli were presented across the three sessions was counterbalanced according to the role of the stimuli in preexposure. Extinction. On each of the next eight days, rats received a 40-min extinction session. Two non-reinforced presentations of both the 30-s tone and the 30-s clicker were presented equally spaced, after 8, 16, 24 and 32

111

min, in a pseudo-randomised, counterbalanced order. Lever pressing rates in conditioning and extinction were recorded during the 30 s both prior to (preCS period) and during (CS period) the stimulus presentation. A suppression ratio for each trial was calculated. Table 3.1.1 displays an outline of the experimental design. Statistical analysis Refer to Experiment 2.1.

Table 3.1.1. Experimental design of Experiment 3.1 (Glucocorticoids in LI). SAL = SAL-treated control group; DEX = DEX-treated group with dose in mg/kg; MT = magazine training; LT = lever training; A & B = tone or clicker; VT = variable time schedule; VI = variable interval schedule; RL = right lever. Treatment according to group label.

PHASE

Magazine- & Lever Training Pre-experiment

GROUP SAL DEX 0.3 DEX 0.6 DEX 1.2

Preexposure Habituation Day 1

MT: VT-30 12 x LT:VI-2, 15, 30, 60 CS A ? 0 RL ? Food + Treatment

Day 2 2x CS A ? 0 CS B ? 0

Conditioning Extinction Day 3-5

Day 6-13

1x 2x CS A ? Shock CS A ? 0 CS B ? Shock CS B ? 0

Results The data were analysed in two ways, using separate preCS and CS scores as dependent variables in one analysis, and a suppression ratio (CS / preCS+CS) in the other. Furthermore, habituation, conditioning and extinction trials were analysed separately. Figure 3.1.1 shows responding levels to novel and preexposed stimuli across conditioning irrespective of the treatment condition. Figure 3.1.2 shows responding levels to novel and preexposed stimuli across conditioning and extinction

112

Leverpressing during CS

5 novel pre-exposed

4

3

2

1

0 C1

C2

C3

E1

Trial

Figure 3.1.1. Responding levels to novel and preexposed stimuli across conditioning (C1-C3) and the first extinction trial (E1), irrespective of the treatment condition.

113

Conditioning

Extinction

Leverpressing during CS

8 7 6 5 4 3

SAL-N NOV SAL-P PXP 0.3DEX-N 0.3DEX-P 0.6DEX-N PXP 0.6DEX-P 1.2DEX-N NOV 1.2DEXP

2 1 0 1

2

1-4

3

5-8

9-12

13-16

Trial

Figure 3.1.2. Responding levels (+SEM) to novel and preexposed stimuli across conditioning (C1-C3) and extinction (presented in bins of 4 trials) in the different treatment conditions. N = novel; P = preexposed.

114

in the different treatment conditions. LI, as reflected by greater conditioning to novel stimuli, was observed in all treatment groups. It was not affected by DEX. PreCS scores. At first, preCS scores were collapsed over all habituation, conditioning and extinction trials and analysed in order to verify homogeneous preCS responding. A two-way, 4 x 2, mixed ANOVA was carried out, with TREATMENT (SAL, DEX 0.3, DEX 0.6, DEX 1.2) as the between-subject factor, and STIMULUS (preexposed, novel) as a repeated measure, referring to the type of stimulus following the preCS period. There was no significant main effect of TREATMENT, F(3, 28) < 1 (Ms: SAL = 5.59, DEX 0.3 = 4.83, DEX 0.6 = 5.60, DEX 1.2 = 5.35) or STIMULUS, F(1, 28) = 1.95, p = .17 (Ms: preexposed = 5.27, novel = 5.41), and no significant interaction, F(3, 28) < 1. This confirms homogenous preCS responding across the groups, justifying the use of CS scores and suppression ratios as the dependent variables in further analyses. CS scores. Habituation. The two habituation trials and the functionally equivalent first conditioning trial (data not shown) were analysed with a three-way, 4 x 2 x 3, mixed ANOVA with TREATMENT as the between-subject factor, and STIMULUS and TRIAL (HAB 1 + 2, COND 1) as repeated measures. There was no significant main effect of TREATMENT, F(3, 28) < 1, and no significant interactions of TREATMENT x STIMULUS, F(3, 28) < 1, TREATMENT x TRIAL, F(6, 56) < 1, or TREATMENT x STIMULUS x TRIAL, F(6, 56) = 1.27, p = .28. A significant interaction of STIMULUS x TRIAL, F(2, 56) = 6.65, p < .01, indicates habituation to the novel stimulus. Simple effect analysis confirmed that responding was greater during the preexposed than the novel stimulus on the two habituation trials but this was no longer so on the first conditioning trial, confirming homogenous responding prior to conditioning.

115

Conditioning. The three conditioning trials plus the functionally equivalent first extinction trial were considered in a three-way, 4 x 2 x 4, mixed ANOVA, with TREATMENT as the between-subject factor, and STIMULUS and TRIAL (COND 13, EXT 1) as repeated measures. It revealed no significant main effect of TREATMENT, F(3, 28) < 1 (Ms: SAL = 2.55, DEX 0.3 = 1.97, DEX 0.6 = 1.64, DEX 1.2 = 1.56) and no significant interactions of TREATMENT x STIMULUS, F(3, 28) = 1.56, p = .22, TREATMENT x TRIAL, F(9, 84) < 1, or TREATMENT x STIMULUS x TRIAL, F(9, 84) < 1. There was a significant main effect of STIMULUS, F(1, 28) = 13.79, p < .001, with greater responding, and therefore less learning, to preexposed stimuli (M = 2.69) than to novel stimuli (M = 1.17). This demonstrates the occurrence of LI. A significant main effect of TRIAL, F(3, 84) = 56.55, p < .001 confirmed the occurrence of conditioning. Newman-Keuls posthoc analysis revealed greater responding on trial 1 (M = 4.31) than on all other trials (Ms: trial 2 = 1.83, trial 3 = 1.47, trial 4 = 0.11), ps < .001, and less responding on trial 4 than on all other trials, ps < .001. The interaction of STIMULUS x TRIAL was also significant, F(3, 84) = 9.93, p < .001. Simple effects analysis found that responding to preexposed stimuli was greater than to novel stimuli on trial 2, F(1, 28) = 15.72, p < .001 (Ms: preexposed = 3.16, novel = 0.50), and on trial 3, F(1, 28) = 17.73, p < .001 (Ms: preexposed = 2.84, novel = 0.09), confirming LI. The lack of a difference on trial 4 may be due to a floor effect (Ms: preexposed = 0.03, novel = 0.19). Extinction. In order to assess possible differences in LI that the conditioning phase may not have been sensitive to, the extinction trials were also analysed. A three-way, 4 x 2 x 16, mixed ANOVA, with TREATMENT as the between-subject factor, and STIMULUS and TRIAL (EXT 1 - 16) as repeated measures, was carried out. There were no significant main effects of TREATMENT; F(3, 28) < 1 (Ms: SAL

116

= 3.16, DEX 0.3 = 2.81, DEX 0.6 = 2.41, DEX 1.2 = 2.30), or STIMULUS, F(1, 28) < 1 (Ms: preexposed = 2.69, novel = 2.66), nor any significant interactions involving either factor. There was a significant main effect of trial, F(15, 420) = 62.15, p < .001, with a progressive increase in responding over trials confirming the occurrence of extinction. Suppression ratios: Habituation, conditioning and extinction data were also analysed with suppression ratio as the dependent variable. It was calculated by dividing the number of responses during the CS period by the number of responses during the preCS and CS period. The same ANOVAs were carried out, revealing exactly the same patterns of results as with CS scores (data not shown), confirming the reported findings.

Discussion LI was observed in all treatment groups, but it was not affected by postpreexposure DEX. At first sight, there seems to have been a tendency for greater levels of unconditioned suppression in the DEX treated animals (of all doses), compared to SAL animals on trial 1 of conditioning for preexposed stimuli, as if they failed, compared to controls, to show the same level of post-preexposure habituation, treating those stimuli as novel despite being familiar. However, this was not confirmed statistically. Furthermore, LI occurred in all animals independently of treatment. No differences between groups in the magnitude of this effect were observed. Given that there were only 12 preexposure trials in order to allow for a possible increase in LI, the magnitude of the effect was quite surprising. In previous studies, Killcross (personal communication, January 2002; Killcross et al., 1994a,b) did not see such a clear LI effect with only 12 preexposures. The results differed from those of Shalev et al. (1998) who found an impairing effect of glucocorticoids. This is

117

not surprising however, since their study pursued a different research question. In contrast to the current study, Shalev et al. (1998) administered corticosterone prior to preexposure and conditioning. Their aim was to investigate the links between glucocorticoids and mesolimbic DA (see Chapter 5), and between enhanced DA levels and an abolition of LI. Since glucocorticoids are thought to interact with mesolimbic DA (e.g. Piazza & Le Moal, 1996), and since a DA increase is associated with deficits in LI, Shalev et al. (1998) hypothesized that corticosterone would lead to a disruption of LI as well. Indeed, this is what they found. It should be pointed out here that there is evidence that the nigrostriatal and not the mesolimbic DA system mediates effects on LI (Killcross & Robbins, 1993; Ellenbroek, Knobbout, & Cools, 1997, but see Joseph et al., 2000). Since chronic states of hypercortisolism, as in patients treated regularly with cortisol or in people suffering from Cushing’s syndrome, have sometimes led to the development of psychosis, Shalev et al. (1998) also investigated the effects of chronic corticosterone treatment on LI, again finding a disruption, an effect later found to be specific to male subjects (Shalev & Weiner, 2001). They argued that corticosterone sensitises the response of an organism to stimuli in the environment, so that these stimuli are treated as novel. Given the proposed role of MRs in response selection and sensory integration (Oitzl & de Kloet, 1992), one could speculate that the effects found by Shalev and colleagues are mediated by MRs. The associative interference or the retrieval failure hypotheses allow three ways by which LI could be affected pharmacologically (Ellenbroek et al., 1997): enhancing or impairing CS-nothing associations at preexposure, enhancing or impairing CS-US associations at conditioning, or altering behavioural switching (Weiner, 1990), i.e. the change of ongoing behaviour in a conflicting situation. This

118

depends on the stage at which pharmacological treatment is administered. Given that some type of CS-nothing learning is likely to appear during preexposure (Bouton, 1991; Killcross & Balleine, 1996) and that this is the stage after which treatment occurred, the current results suggest that this form of learning may not be susceptible to glucocorticoid memory modulation.

3.2 GLUCOCORTICOIDS IN EXTINCTION

Another situation during which non-reinforced presentations of a stimulus occur is extinction (EXT). In contrast to LI, non-reinforced stimulus presentations in EXT occur subsequent to prior CS-US formation in conditioning. Rescorla and Wagner (1972) regarded EXT as a form of unlearning process where a change of existing CS-US associations takes place. This view can be challenged by the phenomenon of spontaneous recovery (Pavlov, 1927) in which the extinguished response recovers if a certain amount of time elapses before testing the CS again. Other evidence against the unlearning view of EXT comes from its dependency on the modulatory influence of context (Bouton, 1994). In a phenomenon called reinstatement, contextual conditioning can reinstate responding to the extinguished CS (e.g. Bouton, 1984), while the renewal effect describes how responding to an extinguished stimulus re-occurs in the conditioning context when EXT occurred in a different one (e.g. Bouton & Bolles, 1979). As mentioned earlier, LI is also contextspecific (Hall & Channel, 1985; Hall & Honey, 1989), and as with LI, theories of competing associations have been formulated for EXT (Pearce & Hall, 1980; Bouton, 1991). New learning is assumed to occur that subsequently competes with the original

119

CS-US representation for behavioural expression. This new learning may either be in the form of excitatory CS-nothing associations (Pearce & Hall, 1980; Bouton, 1991), or constitute an inhibitory link between the CS and the US or the CR (Delamater, 1996; Rescorla, 1996). Kraemer and Spear (1992) have proposed a shared retrieval mechanism in LI and EXT, describing the release-from-LI effect, i.e. the reduction or abolition of LI, as analogous to spontaneous recovery from EXT. There have been a number of studies on the effects of glucocorticoids on Pavlovian EXT, in most instances reporting a facilitation of EXT by glucocorticoids. For example, van Wimersma Greidanus (1970) found that post-training administration of a number of steroids such as progesterone, pregnenolone, corticosterone and DEX facilitated EXT 4 hr later in an active avoidance (pole-jumping) task, extending earlier work that reported facilitative effects of intracerebral implantations of corticosterone and DEX (van Wimersma Greidanus & de Wied, 1969). Similar effects were reported by Bohus and Lissak (1968) using cortisone, while ADX prior to EXT led to an impairment (but see Micco et al., 1979). Kovacs et al. (1976) found a dosedependent bidirectional relationship between corticosterone dose and the rate of EXT. Low doses facilitated and high doses delayed EXT of active avoidance behaviour. However, in each of these cases, treatment occurred before EXT trials, raising the possibility that it may have had an effect on attentional, sensory-perceptual or motivational factors. The behavioural results were linked to changes in mesencephalic and hypothalamic 5-HT levels, which changed in the same fashion as behaviour. Other studies examined the effects of corticosterone in appetitive operant tasks. Hennessy et al. (1973), as well as Garrud, Gray, and de Wied (1974) found a facilitative effect of corticosterone on EXT. Similarly, Micheau et al. (1982) reported that post-EXT microinjections of corticosterone into the hippocampus in mice

120

enhanced EXT 24 hr later. Port et al. (1998) also reported facilitated EXT in an appetitive instrumental paradigm following corticosterone but not DEX treatment prior to EXT trials. However, as treatment occurred prior to EXT trials, factors other than memory modulation cannot be ruled out. On the contrary, Micco et al. (1979) did not see an effect of corticosterone on appetitive EXT, and ADX led to a facilitation which could be blocked by corticosterone. With respect to this project, it was aimed to extend the previous literature and explicitly test whether post-EXT treatments with the GR agonist DEX would have a modulatory effect on the further course of Pavlovian EXT on subsequent days. If CSnothing associations (or inhibitory CS-US or CS-CR links) compete with previously acquired CS-US association for behavioural expression, resulting in the decline of overall behavioural responding to the CS, such associations may be susceptible to glucocorticoid-induced modulation and enhancement, eventually resulting in more rapid EXT. The results of the LI study could not find any evidence for such modulation and given the theoretical and experimental analogies between the two phenomena mentioned above, one might expect to find support for the null hypothesis. However, there is a further important difference in the nature of nonreinforced trials between LI and EXT. Whereas in LI, non-reinforcement of the (novel) stimulus may not come as a surprise so that the CS-nothing association could be described as neutral or non-emotional (but see Killcross & Balleine, 1996), nonreinforcement in an EXT procedure may be regarded slightly differently. Given the previously reinforced nature of the CS, the null-outcome in the new CS-nothing association could be described as emotional in the sense that the previously learned motivationally significant outcome no longer occurs. Depending on whether conditioning was appetitive or aversive, non-reinforced stimulus presentations in EXT

121

may carry elements of frustration or relief, and hence constitute a more emotional form of CS-nothing learning than is the case in LI. This may make a difference to the susceptibility of EXT to post-trial glucocorticoid memory modulation. In order to test these possibilities, Experiment 3.2, investigating the effects of post-EXT DEX, was conducted. The experiment was very similar to the aversive conditioning study that examined conditioned suppression (Experiment 2.2, Chapter 2), with the main difference that DEX treatment occurred after EXT trials.

Method Subjects Subjects were 16 male, Lister Hooded rats (Harlan, UK; n = 8/group), weighing between 381 g and 450 g (M = 427 g) at the beginning of the experiment (EXT stage). For housing and feeding information, refer to Experiment 2.1. All procedures were carried out between 10:00 and 13:00. Apparatus Refer to Experiment 2.2. A retractable lever was located to the left of the magazine. Food pellets (45 mg, Rodent Grain-Base Formula; BIO-SERV, USA) provided positive reinforcement. A 3-kHz tone (75 dB, 30 s) served as a discrete auditory stimulus in each chamber with the background noise being 55 dB. Footshock (0.5 mA, 0.5 s) was available as US. Drugs and injection procedure DEX (Sigma, UK), at a dose of 1.2 mg/kg in a volume of 6.0 ml/kg, or SAL was injected (sc) immediately post-EXT. This dose was selected because it had a modulatory effect when given post-training in appetitive and aversive conditioning paradigms (Chapter 2). The aim was to directly compare possible effects to those

122

seen in Experiment 3.2 where the experimental parameters were almost identical with administration occurring after training rather than after EXT. For preparation details, refer to Experiment 2.2. Behavioural procedure The behavioural procedure was the same as in Experiment 2.2 with the following modifications: 1. There were two rather than one conditioning sessions. 2. There were ten rather than six EXT sessions. 3. Treatment did not take place after conditioning but rather immediately after EXT sessions 1 and 2, i.e. after EXT trials 2 and 4. Table 3.2.1 displays an outline of the experimental design. Statistical analysis Refer to Experiment 2.1.

Table 3.2.1. Experimental design of Experiment 3.2 (Glucocorticoids in EXT). SAL = SAL-treated control group; DEX = DEX-treated group with dose in mg/kg; MT = magazine training; LT = lever training; VT = variable time schedule; VI = variable interval schedule; LL = left lever. Treatment according to group label, after EXT sessions 1 and 2 on days 5 and 6 only.

PHASE

GROUP SAL DEX 1.2

Magazine- & Lever Training Pre-experiment, Day 4 MT: VT-30 LT: VI-2, 15, 30, 60 LL ? Food

Habituation

Conditioning

Extinction

Day 1

Day 2-3

Day 5-14

2x Tone ? 0

1x Tone ? Shock

2x Tone ? 0 (+ Treatment)

123

Results Figure 3.2.1 illustrates the effects of post-EXT administration of DEX on conditioned suppression of lever pressing over the further course of EXT. DEX had no effect on extinction of conditioned suppression. PreCS scores: At first, preCS scores were collapsed over all habituation, conditioning and EXT trials and analysed in order to verify homogeneous preCS responding. A t-test was carried out to compare lever press rates (LP/30 s) in the two treatment conditions and no difference was found between them, p = .56, (Ms: DEX = 5.88; SAL = 6.58). Suppression ratios and CS scores were therefore considered appropriate measures to assess conditioned suppression of lever pressing and constitute dependent variables for further analyses. Suppression ratios. Habituation and conditioning. Groups did not differ prior to treatment during habituation and conditioning. A two-way, 2 x 4, mixed ANOVA with one between groups factor, GROUP (pre-DEX, pre-SAL) and one repeated measure, TRIAL (HAB 1 + 2, COND 1 + 2) found no main effect of GROUP, F(1, 10) < 1, (Ms: pre-DEX = 0.22, pre-SAL = 0.33) nor an interaction of GROUP x TRIAL, F(3, 30) < 1. A significant main effect of TRIAL, F(3, 30) = 3.30, p < .04, reflects the occurrence of conditioning. Newman-Keuls posthoc analysis revealed a lower suppression ratio on conditioning trial 2 (M = 0.08) than on conditioning trial 1 (M = 0.42), p < .03. Extinction. A two-way, 2 x 20, mixed ANOVA with one between groups factor, TREATMENT (DEX, SAL) and one repeated measure, TRIAL (EXT trials 1-20), revealed no significant main effect of TREATMENT, F(1, 7) = 1.36, p = .28 (Ms: DEX = 0.26, SAL = 0.34). The TREATMENT x TRIAL interaction was also not significant, F(19, 133) < 1. There was a significant main effect of TRIAL, F(19, 133) = 10.61, p < .001, with suppression ratios increasing over trials, reflecting EXT.

124

Suppression Ratio (CS/preCS+CS)


Extinction

0.6

0.6

0.5

0.5

0.4

0.4

0.3

0.3

0.2

0.2

0.1

0.1 DEX SAL

DEX SAL

0.0

H1

H2

C1

C2

1

Trial

2

3

4

5

6

7

8

9 10

Session (blocks of 2 trials)

Treatment

Figure 3.2.1. Conditioned suppression of lever pressing (+SEM) over the course of EXT in animals that received either DEX or SAL immediately after EXT sessions 1 and 2.

125

0.0

CS scores: For best comparison to Experiment 2.2, the suppression ratio data are presented here. However, all data were also analysed with CS scores as the dependent variable. The same ANOVAs were carried out, revealing exactly the same patterns of results as with suppression ratios (data not shown), confirming that there were no differences between treatment groups (all Fs < 1).

Discussion No effect of post-EXT administration of DEX was found, with EXT proceeding at the same rate in both treatment groups. Both groups responded equally prior to treatment, and EXT took a total of 20 trials to reach a suppression ratio of 0.5, i.e. complete non-suppression. Hence, both floor and ceiling effects that might have masked possible effects of DEX can largely be ruled out. Given the results of the LI study, acceptance of the null hypothesis may not come as a surprise. Even though the nature and mechanisms of LI or EXT have not been completely identified, most studies point to the formation of CS-nothing association at preexposure and EXT stages respectively. Assuming that this is the case, the results provide further support for the possibility that glucocorticoid memory modulation may be selective for associations containing a discrete US, and therefore may not operate on CS-nothing associations. On the other hand, there are important differences between LI and EXT. In contrast to the preexposure stage during LI, non-reinforced stimulus presentations in EXT may contain elements of frustration or, in this case, relief, and hence be of a more emotional nature than in LI. Glucocorticoid levels have indeed been found to rise during appetitive EXT, an effect linked to both anticipation of food reward and frustration (Coe, Stanton, & Levine, 1983; Coover, Goldman, & Levine, 1971;

126

Coover, Murison, Sundberg, Jellestad, & Ursin, 1984; Dantzer, Arnone, & Mormede, 1980; de Boer, de Beun, Slangen, & van der Gugten, 1990; Goldmann, Coover, & Levine, 1973; Levine & Coover, 1976; Randich, Froehlich, Fraley, Fjermestad, & Brush, 1976), as well as in aversive EXT, where the increase has been attributed to the conditioned aversive properties of the CS (Coover, Sutton, Welle, & Hart, 1978). Since the human literature suggests a selective role for glucocorticoids in the modulation of emotional memory (Buchanan & Lovallo, 2001), one might have expected DEX to have an effect in the current study. Indeed, other authors have reported a glucocorticoid-induced facilitation of EXT (Bohus & Lissak, 1968; van Wimersma Greidanus, 1970; Garrud et al., 1974; Kovacs et al., 1976; Micheau et al., 1982; Port et al., 1998, but see Micco et al., 1979). However, a number of reasons might account for these apparent discrepancies. First of all, none of these studies employed an aversive discrete-cue Pavlovian conditioning procedure, and hence they may not be fully comparable to the current one. Secondly, most of them were not carried out from a perspective of memory modulation and EXT as a form of learning, but instead pursued other questions, e.g. the effects of increased glucocorticoid levels during an ongoing EXT process, so that treatment was administered before EXT trials (Bohus & Lissak, 1968; Kovacs et al., 1976; Port et al., 1998). In one study (van Wimersma Greidanus, 1970), treatment occurred after one EXT session but increased glucocorticoid levels are likely to have been present during the next one, 4 hr later, which served as test. A facilitation of EXT in these studies might hence constitute a performance deficit of retrieval, in line with other work reporting a detrimental effect of glucocorticoids on memory retrieval (de Quervain et al., 1998, 2000). However, this explanation does not hold for Micheau et al. (1982) who reported a facilitation of EXT in an appetitive instrumental procedure where treatment occurred after one EXT

127

session while the test took place 24 hr later, similar to the current study’s design. It is possible, however, that EXT of appetitive operant conditioning comprises a situation substantially different from aversive Pavlovian EXT, given that the omission of the US might constitute a frustrative non-reward. However, it is also possible that the current study was not sensitive enough to detect an effect of DEX. Kovacs et al. (1976) reported dose-dependent bidirectional effects of corticosterone on the rate of EXT. In the current study, only one dose of DEX was applied which has been chosen for its observed post-training effects on the acquisition of discrete-cue Pavlovian associations (Experiment 2.2). It is hence possible that this dose was ineffective whereas other, possibly higher, doses might have led to different results. To establish whether the lack of an effect of DEX in the current study was due to quantitative reasons, for example an insufficient dose, or qualitative reasons, for example a lack of emotional arousal in aversive EXT per se, further experiments need to be carried out. As glucocorticoid memory modulation has been suggested to depend on concurrent noradrenergic activity in the amygdala (e.g. Quirarte et al., 1997), possibly a neural reflection of emotional arousal, it would, for example, be interesting to investigate if the same dose of DEX administered in combination with a noradrenergic agonist would affect EXT.

Experiments 3.1 and 3.2 investigated the effects of post-trial glucocorticoids in paradigms that are likely to involve learning about CS-nothing associations. They stand in contrast to the more traditional CS-US associations of the experiments in Chapter 2, in the sense that they lack the occurrence of a discrete reinforcer. Glucocorticoids were not found to modulate these associations. However, there are

128

other forms of learning that lack the presence of a discrete US, namely stimulusstimulus (and stimulus-response) associations. The remainder of this chapter describes two experiments investigating the effects of post-trial glucocorticoids on two Pavlovian conditioning paradigms, sensory preconditioning and second-order conditioning, each of which are likely to rely on stimulus-stimulus (S-S) or stimulusresponse (S-R) associations.

3.3 GLUCOCORTICOIDS IN SENSORY PRECONDITIONING

In the sensory preconditioning (SPC) paradigm (Brodgen, 1939), two motivationally neutral stimuli, A and X, are presented together, either in a contingent sequence (e.g. Rizley & Rescorla, 1972) or as a simultaneous compound (e.g. Rescorla & Cunningham, 1978; Ward-Robinson, Symonds, & Hall, 1998; WardRobinson et al., 2001). In the next stage, one of these stimuli (X) is conditioned to a motivationally significant reinforcer. In a final test stage, responding to A is greater than to an appropriate control stimulus. SPC could either occur by means of an associative chain where presentation of A in test evokes a representation of X which in turn evokes a representation of the reinforcer. Hence, the retrospective motivational attribution of A may well rely on the formation of neutral, S-S (A-X) associations prior to conditioning of X. Alternatively, during conditioning, X may evoke a representation of A which itself gets conditioned (fantastic conditioning). In either case, an association between A and X needs to be formed in the compound exposure stage first. Like LI and EXT, SPC is context specific (Ward-Robinson et al., 1998).

129

As far as I am aware, the effects of glucocorticoids in emotionally neutral paradigms such as SPC have not yet been investigated. In humans, it has been found that glucocorticoids do not modulate consolidation of neutral material (de Quervain et al., 2000; Buchanan & Lovallo, 2001). Experiment 3.3 uses a taste aversion SPC paradigm, employing within-compound flavours in the A-X phase, in order to investigate if consolidation of the neutral S-S association could be modulated by the post-training administration of the GR agonist DEX.

Method Subjects Subjects were 32 male, Lister Hooded rats (Harlan, UK; n = 16/group), weighing between 408 g and 498 g (M = 450 g) at the beginning of the experiment. They were singly housed in a temperature-controlled colony room (21±2ºC) and maintained on a 12:12-hr-light-dark cycle (lights on at 08:00 am) with restricted access to water (see behavioural procedure). Prior and during the experiment, rats were fed approximately 15g of food pellets per day to maintain their body weights. All procedures were carried out at approximately 11:00 and 16:00. Apparatus All procedures were carried out in the rats’ home cages. The cages contained standard sawdust bedding, were constructed of opaque plastic and measured 24-cmwide x 20-cm-high x 41-cm-long. Measured amounts of tap water, 0.33 M sucrose, 0.16 M SAL, 60 µM quinine, and 0.01 M hydrochloric acid were presented with inverted 50-ml centrifuge tubes fitted with ball-bearing-tipped spouts. When the solutions were presented as compounds, they were made up so as to maintain the

130

molarity of the individual flavours. A weighing balance was used to determine fluid consumption to the nearest 0.1g. Drugs and injection procedure DEX (Sigma, UK), at a dose of 1.2 mg/kg in a volume of 6.0 ml/kg, or SAL was injected (sc) immediately after the flavour compound presentation period. For preparation details, refer to Experiment 2.2. Intraperitoneal (ip) administration of 0.30M Lithium Chloride (LiCl) at a dose of 10 mg/kg took place immediately after presentations of flavour X during conditioning. Rats were injected within 20 min of the removal of the centrifuge tubes. Behavioural procedure The procedure was modified from Rescorla and Cunningham (1978). Half of the subjects were assigned to the experimental (EXP) and the rest to the control group (CON). Table 3.3.1 displays an outline of the experimental design. Water Deprivation: A schedule of water deprivation was established in the home cages on the three days prior to the experiment. The standard water bottles were removed and access to fluid was restricted to a single 30-min presentation of water at 11:00. During all stages of the experiment, rats only received extra water access where indicated. Compound exposure: Rats received three 30-min presentations of each compound solution AX and BY respectively on alternate days (1-6) at 11:00. Flavours A and B were sucrose and SAL for half of the subjects in each group, with the flavours reversed for the remainder. Flavours X and Y were quinine and hydrochloric acid for EXP, whereas the reverse was the case for CON. For EXP, AX was immediately followed by DEX and BY by SAL treatment. For CON, AX was followed by SAL and BY by DEX. Conditioning: On days 7 and 10, rats received a conditioning trial comprising a 30-min presentation of X followed by an ip injection of the emetic LiCl

131

in order to induce nausea.. On days 8 and 9, rats received a non-reinforced 30-min presentation of flavour Y. Presentations occurred at 11:00 and 30-min access to water was granted at 16:00. Day 11 was a recovery day during which rats were given two sessions of 30-min access to water. SPC test: On day 12, rats were given free access to solutions A and B for 30 min each. One solution was presented at 11:00 and the other at 16:00. The presentation order was counterbalanced in both groups. Taste aversion test: On day 13, rats were given free access to solutions X and Y for 30 min each, in order to confirm that an aversion of X relative to Y had been established. One solution was presented at 11:00 and the other at 16:00. The presentation order was counterbalanced in both groups. Statistical analysis Refer to Experiment 2.1.

Table 3.3.1. Experimental design of Experiment 3.3 (Glucocorticoids in SPC). TA = taste aversion; CON = Control Group (DEX after BY); EXP = Experimental Group (DEX after AX); A & B = SAL or sucrose; X & Y = quinine or hydrochloric acid. + indicates aversive reinforcement by LiCl.

Phase

Group

Compound exposure Day 1-6 (alternately)

Conditioning

SPC Test

TA Test

Day 7 &10

Day 8-9

Day 12

Day 13

EXP

AX (DEX)

BY (SAL)

X+

Y-

A vs. B

X vs. Y

CON

AX (SAL)

BY (DEX)

X+

Y-

A vs. B

X vs. Y

132

Results SPC test. Figure 3.3.1 shows the results of the crucial SPC test and illustrates the effects of DEX administered after the S-S stage. SPC was established, as reflected by lower consumption of flavour A than B, but DEX was without effect. A two-way, 2 x 2, mixed ANOVA with GROUP (CON, EXP) as the between subject factor and FLAVOUR (A, B) as a repeated measure was carried out. There was a significant main effect of FLAVOUR, F(1, 30) = 11.16, p < .01, with A (M = 8.81) being less consumed than B (M = 14.09). This demonstrates a sensory preconditioning effect. A had previously been paired with X which went on to be aversively reinforced by LiCl. There was also a significant main effect of GROUP, F(1, 30) = 40.10, p < .001, with the CON group (M = 15.22) consuming more solution overall than the EXP group (M = 7.68). There was no significant interaction of GROUP x FLAVOUR, F(1, 30) < 1, indicating no selectivity or difference in the magnitude of sensory preconditioning across the groups. To further assess possible differences in SPC magnitude between groups, consumption of A was expressed relative to consumption of B by using difference scores (A-B), and a t-test was conducted to compare the two groups (data not shown). No difference between groups was found, p = .85. Sensory Preconditioning (compound exposure) and conditioning. Table 3.3.2 displays the overall consumptions of compounds and flavours in the two groups collapsed over trials during compound exposure and conditioning. Compound exposure: A two-way, 2 x 2, mixed ANOVA was carried out, with GROUP (CON, EXP) as the between-subject factor and COMPOUND (AX and BY) as repeated measure. There was a preference for AX over BY in CON animals whereas the opposite was the case for EXP. This was confirmed by a significant interaction of GROUP x COMPOUND, F(1, 30) = 52.11, p < .001 and subsequent

133

CON

EXP

Mean consumption in g

20 18 16 14 12 10 8 6 4 2 0 A

B

A

B

Flavour

Figure 3.3.1. Mean consumptions of flavours A and B in the two groups in the final sensory preconditioning test. CON = Control Group (SAL after AX), EXP = Experimental Group (DEX after AX).

134

Table 3.3.2. Mean consumptions (g) of compounds AX and BY during compound exposure, and flavours X and Y during conditioning, in the two groups. CON = control group (SAL after AX); EXP = experimental group (DEX after AX).

GROUP PHASE

EXP

CON

AX

6.24

9.28

BY

8.80

6.96

X

2.94

7.06

Y

9.59

7.37

STIMULI Compound exposure Conditioning

135

simple-effect analysis. There were no main effects of GROUP, although significance was approached, F(1, 30) = 3.15, p = .09, or COMPOUND, F(1, 30) < 1. Conditioning. A three-way, 2 x 2, mixed ANOVA was carried out with GROUP (CON and EXP) as the between-subject factor, and FLAVOUR (X and Y) as repeated measure. There was a significant main effect of FLAVOUR, F(1, 30) = 82.60, p < .001, with greater consumption of Y (M = 8.48) than of X (M = 5.00), suggesting an effect of conditioning. The main effect of GROUP approached significance, F(1, 30) = 3.72, p = .06, whereas the interaction of GROUP x FLAVOUR, F(1, 30) = 68.56, p < .001, was significant. A better assessment of conditioning can be found in the analysis of the taste aversion test (below). The fact that the compounds AX and BY and the flavours X and Y were differentially consumed by the two groups is most likely due to the fact that flavour X was always quinine for EXP and hydrochloric acid for CON. Quinine may be a less popular flavour for rats than hydrochloric acid (J. Ward-Robinson, personal communication, January 2001) and may even have mildly aversive properties (Cahill & McGaugh, 1990). However, it is unlikely that this influenced the result of the SPC in any systematic way, and could not mask possible effects of DEX. Taste aversion test. The results of the taste aversion test are displayed in Figure 3.3.3. For the test aversion test, a two-way, 2 x 2, mixed ANOVA with GROUP (CON, EXP) as the between subject factor and FLAVOUR (X, Y) as repeated measure was carried out. There was a significant main effect of FLAVOUR, F(1, 30) = 359.05, p < .001, with X (M = 0.56) being consumed less than Y (M = 9.27). There was also a significant main effect of GROUP, F(1, 30) = 4.93, p < .04, with the CON group (M = 4.38) consuming less solution overall than the EXP group (M = 5.44), and a significant interaction of GROUP x FLAVOUR, F(1, 30) = 7.10,

136

EXP

CON Mean consumption in g

12 10 8 6 4 2 0 X

Y

X

Y

Flavour

Figure 3.3.3. Mean consumptions of flavours X and Y in the two groups in the taste aversion test. CON = Control Group (SAL after AX), EXP = Experimental Group (DEX after AX).

137

p < .02. Simple effect analysis revealed greater consumption of flavour Y by the EXP group (M = 10.41) than the CON group (M = 8.13), p < .03, while there was no difference in flavour X, p = .18. These data suggest that a taste aversion had been established to X. The difference of overall consumption between groups was probably due to the lower consumption of Y in CON (where it was quinine) than in EXP (where it was hydrochloric acid).

Discussion This experiment aimed

to investigate the effects of post-training

glucocorticoids on the consolidation of memory for a neutral stimulus-stimulus association. A taste aversion SPC paradigm was employed where two neutral flavour stimuli, A and X, were paired with each other in compound, followed by aversive conditioning to one of these stimuli, X, by inducing sickness through LiCl. An SPC effect was successfully demonstrated, in form of a lower consumption of flavour A, as compared to a control flavour B, in a later test. Irrespective of whether SPC arises from the formation of an associative chain or from fantastic conditioning, it relies on a learning process during the S-S compound exposure stage. Administration of DEX after the S-S compound exposure stage did not affect SPC. As far as I am aware, the current study constitutes the first investigation of the effects of glucocorticoids on SPC. In the human literature and in line with the current results, glucocorticoids have been reported to not affect the consolidation of memory for neutral information (de Quervain et al., 2000; Buchanan & Lovallo, 2001). There were some differences between groups in consumption of flavours and compounds in the compound exposure and conditioning stages. This could be explained by the fact that, due to an error in counterbalancing, X was always quinine

138

for EXP and hydrochloric acid for CON animals. Quinine is probably a less-preferred flavour in rats (J. Ward-Robinson, personal communication, January 2001) and may even be regarded as slightly aversive (Cahill & McGaugh, 1990). However, it is unlikely that these differences had a systematic effect on the SPC test data. Results of the taste aversion test confirmed successful aversive conditioning to X in both groups. The lack of an interaction in the SPC test suggests that SPC took place equally in both groups. The additional analysis of difference scores, finding no difference between groups, confirmed an equal magnitude of the SPC effect in both groups. Finally, the results are in line with those of the other experiments presented in this chapter where DEX did not affect paradigms that a lack a discrete US.

3.4 GLUCOCORTICOIDS IN SECOND-ORDER CONDITIONING

Another experimental paradigm containing stimulus-stimulus associations is that of second-order conditioning (SOC). SOC paradigms contain exactly the same stages as SPC procedures, except that the order of the stimulus-stimulus (CS1-CS2) and stimulus-reinforcer (CS2-US) phases is reversed. In SOC paradigms, conditioning between a CS (X) and a US (1st order conditioning) occurs before paired presentations of A and X. In a later test, A triggers greater responding than a control stimulus, B. As for SPC preparations, A-X presentations can occur either in a contingent sequence or as a simultaneous compound. The crucial difference between SPC and SOC is that A and X are emotionally neutral when paired in SPC, whereas X will already have acquired some emotional significance through previous first-order conditioning in SOC. There are three mechanisms that could explain SOC. According to the

139

associative chain account, presentation of A during test evokes a representation of X, activating a representation of the US. In another account, fantastic conditioning during the second-order stage causes A to become conditioned to a representation of the US, evoked by the presentation of X. Finally, SOC may rely on S-R rather than SS associations, whereby in the second-order stage A gets paired with the conditioned response evoked by X. In a way, the relationship between SPC and SOC, due to the temporal arrangement of the design, is reminiscent of that between LI and EXT. CS-nothing and S-S presentations, the phases under investigation in this project and after which treatment is given, are emotionally neutral and occur before the CS-US stage in LI and SPC. In EXT and SOC, these phases are identical apart from their temporal arrangement, occurring after the CS-US stage, and by virtue of which they may gain emotional significance. The effects of glucocorticoids on SOC have not been investigated before. Given that DEX did not have an effect on the other paradigms described in this chapter, i.e. LI, EXT and SPC, one might expect the study to generate a further null result. On the other hand, the possibly emotional rather than neutral nature of one of the stimuli in the S-S phase, as well as the possible involvement of S-R rather S-S learning, might make SOC more susceptible to glucocorticoid memory modulation than the other paradigms. Indeed, SPC and SOC have been dissociated pharmacologically before in a study by Nader & LeDoux (1999). The authors reported that pre-treatment with the D2 dopamine receptor agonist quipirole blocked the acquisition of SOC but not SPC. The aim of Experiment 3.4 was to investigate the influence of post-trial DEX, in the S-S phase, on second-order conditioning. In contrast to the SPC taste aversion experiment, this study employed a fear-conditioning version of SOC, as it was

140

believed to be difficult to achieve sufficient amounts of S-S exposure, i.e. consumption of a flavour compound, if one of its flavour elements had previously been paired with LiCl.

Method Subjects Subjects were 16 male, Lister Hooded rats (Harlan, UK; n = 8/group), weighing between 266 g and 296 g (M = 277 g) at the beginning of the experiment. For housing and feeding information, refer to Experiment 2.1. All procedures were carried out between 13:00 and 17:00. Apparatus Refer to Experiments 2.2 and 3.1. A retractable lever was located to right of the magazine. Food pellets (45 mg, Rodent Grain-Base Formula; BIO-SERV, USA) provided positive reinforcement. A 3-kHz tone (75 dB, 10 s) and clicker (10 Hz, 75 dB, 10 s) served as discrete auditory cues. The background noise in the chamber was 55 dB. The chamber was illuminated by a house light located in a central position of the ceiling. The light was activated before rats were placed into the chamber and remained so until after they were taken out again. Two further panel lights located on the left and the right of the food tray, activated in parallel, served as discrete visual cue (light); deactivation of the central house light provided another one (dark). Footshock (0.5 mA, 0.5 s) constituted the aversive US. Drugs and injection procedure DEX (Sigma, UK), at a dose of 1.2 mg/kg in a volume of 6.0 ml/kg, or SAL was injected (sc) immediately post-training. For preparation details, refer to Experiment 2.2.

141

Behavioural procedure The procedure was modified from Rizley & Rescorla (1972). Magazine and lever training: Animals received one or two, 30-min magazine training sessions, with the levers removed from the chambers, during which food pellet reinforcement was delivered according to a VT-30 schedule. In the next 30-min session, the lever on the left of the magazine was introduced into the chamber and pressing was reinforced under a VI-2 for 30 min. Animals received between one and five VI-2 sessions before increasing reinforcement schedules of VI-15 (one or two sessions), VI-30 (one session) and finally VI-60 (two sessions) were introduced. The number of sessions required to achieve a stable response baseline varied between individual animals. Baseline response rates did not differ between treatment groups at the end of lever training (data not shown). A VI-60 schedule of reinforcement was maintained throughout the remainder of the experiment. First-order conditioning: On day 1, rats received a 40-min conditioning session with four, equally spaced, 10-s presentations (X+) of either dark (house light off) or light (panel lights on), counterbalanced in both groups. The final 0.5 s of the stimulus coincided with the presentation of a 0.5-s footshock, delivered through the grid floor of the chamber. The ISI was 8 min long. Second-order conditioning: On day 2, rats first received a 30-min VI-60 lever pressing reminder session to reinstate any baseline responding that might have been disrupted due to shock presentations in the conditioning session on the previous day. Later, they received a 40-min stimulus-stimulus conditioning session with four, equally spaced, 10-s presentations of an auditory stimulus A (tone or clicker) and visual stimulus X (dark or light), presented as a simultaneous compound, AX. The ISI was 8 min long. At the end of this session, rats immediately received treatment of either DEX (experimental group) or SAL (control group) and were then returned to

142

their home cage. On day 3, rats received a 40-min stimulus-stimulus conditioning session with four, equally spaced, 10-s presentations an auditory stimulus B (clicker or tone) and Y (light or dark), presented in a compound BY. The ISI was 8 min long. At the end of the session, all rats immediately received SAL treatment and were then returned to their home cages. Designation of visual and auditory stimuli to X or Y and A or B, and the order of compound presentations were counterbalanced in both treatment groups. The same cycle of first and second-order conditioning was repeated once more from day 4 to 6. Second-order conditioning test: On day 7, rats received a 40-min test session with two, equally spaced, 10-s presentations, each of A and B, in the order A-B-B-A or B-A-A-B. In all sessions, suppression ratios for all trials were calculated by dividing the number of LP during the CS period by the total number of LP during the preCS and CS periods. Table 3.4.1 displays an outline of the experimental design. Statistical analysis Refer to Experiment 2.1.

Table 3.4.1. Experimental design of Experiment 3.4 (Glucocorticoids in SOC). SAL = SAL-treated control group; DEX = DEX-treated group with dose in mg/kg; MT = magazine training; LT = lever training; VT = variable time schedule; VI = variable interval schedule; LL = left lever. A and B = tone or clicker; X and Y = dark or light. Treatment according to group label.

PHASE

GROUP SAL DEX 1.2

Magazine- & Lever Training

First-order conditioning

Pre-experiment

Day 1 & 4

Day 2 & 5

Day 3 & 6

Day 7

MT: VT-30 LT: VI-2, 15, 30, 60 LL ? Food

4x X ? Shock

4x AX + Treatment

4x BY + SAL

2 x A? 0 2 x B? 0

143

Second-order conditioning

Test

Results PreCS scores: At first, preCS scores were collapsed over all first-, secondorder conditioning and test trials and analysed in order to verify homogeneous preCS responding (LP/30 s). A t-test was carried out to compare the two treatment conditions and no difference was found between them (Ms: DEX = 4.13; SAL = 4.05), p = .90. Suppression ratios were therefore considered appropriate as the dependent variable for further analyses. Suppression ratios. First-order conditioning. Figure 3.4.1 shows suppression ratios to X on conditioning days 1 and 2. Suppression was greater on day 2. Scores of the four trials of conditioning days 1 and 2 were collapsed, and a two-way, 2 x 2, mixed ANOVA with TREATMENT (DEX, SAL) as the between-subject factor and DAY (1 and 2) as repeated measure was carried out. There was no significant main effect of TREATMENT, F(1, 14) < 1 (Ms: DEX = 0.19 , SAL = 0.17), and no significant interaction of TREATMENT x DAY, F(1, 14) < 1. There was a significant main effect of DAY, F(1, 14), p < .001, with lower suppression ratios on day 2 (M = 0.09) than on day 1 (M = 0.27), confirming aversive conditioning to X. Second-order conditioning test. Figure 3.4.2 compares the two treatment groups in their mean suppression ratios of lever pressing to stimuli A and B, collapsed over two trials, during the test session. Second-order conditioning was established, as reflected by greater suppression to A than B, but DEX was without effect. A threeway, 2 x 2 x 2, mixed ANOVA with TREATMENT (DEX, SAL) as the betweensubject factor and STIMULUS (A and B) and TRIAL (1 and 2) as repeated measures was carried out. There was no significant main effect of TREATMENT, F(1, 14) = 1.08, p = .32. The two-way and three-way interactions TREATMENT x STIMULUS, TREATMENT x TRIAL, STIMULUS x TRIAL, and TREATMENT x STIMULUS x

144

DEX

Suppression Ratio (CS / preCS+CS)

SAL 0.5 0.4 0.3 0.2 0.1 0.0 1

2

1

2

Day

Figure 3.4.1. Suppression ratios to X on first-order conditioning days 1 and 2. SAL = SAL-treated control group, DEX = DEX-treated group. Note that these treatment labels refer to the AX second-order conditioning stage.

145

Suppression Ratio (CS / preCS+CS)

SAL

DEX

0.6 0.5 0.4 0.3 0.2 0.1 0.0

A

B

A

B

Stimulus

Figure 3.4.2. Suppression ratios to A and B (collapsed over two trials) in the secondorder conditioning extinction test. SAL = SAL-treated control group, DEX = DEXtreated group. Note that these treatments refer to the AX second-order conditioning stage.

146

TRIAL were also not significant, all F(1, 14) < 1. There was a significant main effect of STIMULUS, F(1, 14) = 4.92, p < .05, with a lower suppression ratio to stimulus A (M = 0.31) than to stimulus B (M = 0.41). This indicates a successful demonstration of second-order conditioning. There was also a significant main effect of trial, F(1, 14) = 23.46, p < .001, with an increase in suppression ratios from trial 1 (M = 0.26) to trial 2 (M = 0.46), indicating extinction.

Discussion The effects of post-trial glucocorticoids on SOC were investigated. Using visual and auditory cues, and footshock as an aversive US, first- and second-order conditioning was established in both treatment groups. However, post-training DEX treatment did not have any effect on the acquisition of SOC. The study constitutes the first assessment of glucocorticoids on SOC. The outcome may not be entirely surprising given the series of null results from the LI, EXT and SPC studies. However, SOC differs from the other three paradigms in a number of ways. A pharmacological dissociation between SPC and SOC, for example, has been shown by Nader and LeDoux (1999). An important difference is that during the learning phase of interest (A-X) in SOC, the predicted stimulus, X, possesses (second-order) emotional properties, acquired in first-order conditioning. This allows for the possibility of an S-R learning process to take place between A and the conditioned response to X. It could have been possible that in the absence of a discrete US, glucocorticoids would modulate the formation of S-R but not S-S associations. If one then assumes that S-R plays a role in SOC, an effect of DEX could have been anticipated. Secondly, the experience of the second-order emotional properties of X may result in some concurrent physiological process, for

147

example the release of NA (Quirarte et al., 1997), that may be necessary for glucocorticoids to exert an effect on memory consolidation. However, as DEX did not have an effect on SOC, one might conclude that glucocorticoids rely on the presence of a discrete US in order to modulate memory. Alternatively, as previously mentioned in the EXT study, the dose of DEX may not have been sufficient to modulate memory. Hence, an account of the null results that emphasises the lack of a discrete US may actually be of a quantitative rather than qualitative nature. The second-order experience of X may result in lower levels of endogenous glucocorticoids than a firstorder US, for example, footshock, so that a greater dose of DEX than the one used here might have exerted an effect.


In Chapter 2, it was demonstrated that post-training glucocorticoids can modulate memory of discrete-cue Pavlovian CS-US associations, in both aversive and appetitive domains. However, there are other forms of Pavlovian learning. This chapter presented a series of experiments examining glucocorticoids’ ability to modulate memory for associations that lack a discrete US. Instead, learning occurred about the relationship between a stimulus and the absence of any event (LI and EXT) or about the relationship between two stimuli, neither of which constitute a primary reinforcer (SPC and SOC). In LI and SPC, the predicted absence of any event and the predicted second stimulus respectively do not bare emotional significance. In EXT, the absence of any event implies that the CS is no longer reinforced so that it might carry elements of frustration or relief, depending on the affective valence of the US in

148

the prior conditioning stage. In SOC, the predicted stimulus carries an indirect or secondary emotional value as it previously predicted a reinforcer itself. The four paradigms have in common that in none of them (i.e. during the phase of interest) does learning involve the presence of a discrete US. All experiments demonstrated the basic effects under investigation but modulation by post-training glucocorticoids did not occur in any of the studies. Table 3.5.1 gives an overview of the different paradigms employed in this chapter and their different experimental stages.

Table 3.5.1. Overview of the different paradigms employed in the studies of Chapter 2 and Chapter 3. The different phases are described in their temporal relation to the CS-US conditioning stage. Light-grey shadings indicate the learning stage of interest after which treatment (DEX vs. SAL) is administered.

Chapter 2

Phase Paradigm Discrete-cue Pavlovian Conditioning LI

pre CS-US

A? 0

EXT

CS-US

post CS-US

Test

A+

A

A+

A

A+

A? 0

A

3 SPC

X? A

SOC

A+ A+

X vs. Y X? A

X vs. Y

Experiment 3.1 investigated the effects of post-preexposure administration of DEX on LI, the retardation of acquisition of conditioned responding to a stimulus by repeated non-reinforced preexposure to that stimulus (Lubow & Moore, 1959; Lubow, 1973). The preexposure stage in LI is likely to comprise a learning process during which a stimulus is unrelated to any emotionally significant event. LI was established in all treatment groups but post-preexposure DEX treatment did not have an effect on 149

LI. Although the effects of glucocorticoids on LI have been investigated before (Shalev et al., 1998; Shalev & Weiner, 2001), reporting a glucocorticoid-induced decrease, these studies did not pursue questions related to memory modulation. Experiment 3.1 constitutes the first investigation of the effects of glucocorticoid administration after the preexposure stage on later LI. Experiment 3.2 examined the effects of glucocorticoids on EXT, a procedure in which stimuli previously associated with an emotional event now occur in its absence. However, due to this previous contingency and in contrast to LI, the absence of the emotional event in EXT could be regarded as emotional itself, possibly generating relief or frustration. EXT is likely to form another learning paradigm in which a CS is paired with the absence of a US. Post-extinction administration of DEX did not have an effect on the subsequent course of EXT. This stands in contrast to a number of studies that reported a facilitative effect of glucocorticoids on EXT (Bohus & Lissak, 1968; van Wimersma Greidanus, 1970; Garrud et al., 1974; Kovacs et al., 1976; Micheau et al., 1982; Port et al., 1998). However, the majority of these studies examined the effect of enhanced levels of glucocorticoids induced prior to EXT on the ongoing process. Their results are thus in line with studies suggesting a detrimental effect of glucocorticoids on memory retrieval (de Quervain et al., 1998, 2000). The only previous study directly comparable to Experiment 3.2 was that of Micheau et al. (1982) who found a facilitative effect of glucocorticoids on the extinction of appetitive instrumental responding. However, there may be a difference between appetitive and aversive EXT. From an evolutionary point of view, it might well prove adaptive to readily learn that a particular environmental constellation no longer predicts the occurrence of food. In contrast, premature EXT of an aversive

150

contingency might prove lethal, a theory that would support the null results reported here. Experiment 3.3 employed a taste aversion SPC paradigm to assess the effects of DEX on the formation of neutral S-S associations. In SPC, two neutral stimuli are presented together, followed by conditioning of one of them, resulting in increased responding to the other relative to a control stimulus in a later test. SPC probably presents the most apt paradigm to assess neutral learning in animals. SPC was demonstrated in both treatment groups but DEX administration immediately after the S-S phase did not have an effect on SPC. Even though the effects of glucocorticoids on SPC have not been investigated before, the results are in line with some of the human literature, suggesting that memory modulation by glucocorticoids is selective for emotional learning (Buchanan & Lovallo, 2001; de Quervain et al., 2000). The last study of the chapter, Experiment 3.4, investigated the effects of DEX on SOC. In SOC, a neutral stimulus, A, is paired with another stimulus, X, that has gained emotional significance through previous (first-order) conditioning. In a later test, A generates greater responding than a control stimulus. Again, basic SOC was established but administration of DEX immediately after the second-order conditioning phase did not exert an effect. As for SPC, the effects of glucocorticoids on SOC have not been tested before. As mentioned above, the basic effects of all paradigms were established but no modulation was achieved by DEX treatment immediately after the phase of interest, at a dose that has been effective in modulating discrete-cue Pavlovian conditioning (Chapter 2). One might conclude from these results that glucocorticoids do not modulate memory in learning paradigms that lack the occurrence of a discrete US (as it is the case in the paradigms studied here). In line with the findings of this chapter

151

are reports from the human literature, according to which glucocorticoids only enhance memory consolidation for emotionally arousing material. Memory for neutral material is not modulated by glucocorticoids (Buchanan & Lovallo, 2001; de Quervain et al., 2000). The absence of any effects of DEX could be due to a number of reasons. It may be the case, for example, that non-emotional learning, comprising CS-nothing or neutral S-S associations, is mediated by different neural systems than standard discrete-cue Pavlovian conditioning. Whereas structures like the NAC and the CEA, for example, have been implicated in reward-related and aversive conditioning (e.g. Everitt & Robbins, 1992; Killcross et al., 1997; Parkinson, Cardinal, & Everitt, 2001), a putative non-emotional-learning system may involve other structures and may not be susceptible to glucocorticoid memory modulation. It might furthermore be possible that those paradigms comprising CS-nothing relationships, i.e. LI and EXT, do not rely on associative CS-nothing learning processes in the first place. Instead they might depend on attentional processes that are not affected by post-trial glucocorticoids. This explanation does not apply to SPC and SOC, however, and the substantial evidence supporting a role of associative processes in LI and EXT make this account an unlikely one. Since none of the paradigms in this chapter, unlike those of Chapter 2, involves the presence of a discrete US, one might assume that the experience of a discrete US triggers a physiological process that glucocorticoids may depend on, in order to modulate memory. One such process could be the release of NA. Emotionally significant events usually result in activation of both the adrenal cortex and the adrenal medulla. It may be the case that both the adrenergic and the adrenocortical system need to be active in order to achieve memory modulation. Most studies

152

investigating memory modulatory effects of one of these systems by experimental manipulation rely on activity of the other being induced by the (emotionally salient) training condition. Indeed, it has been proposed that memory modulation by glucocorticoids depends on central noradrenergic activity (e.g. Quirarte et al., 1997; for reviews see Cahill & McGaugh, 1996, 1998; Cahill et al., 1997; Roozendaal, 2000). In line with this idea as well as with the results reported here, plasma and central NA levels have been reported to rise in response to footshock (Galvez et al., 1996; Gold & McCarty, 1981; McCarty & Gold, 1981; Quirarte et al., 1998), and de Boer et al. (1990) found enhanced plasma NA in response to food-reinforced operant behaviour, while levels were decreased in EXT. Further support comes from the work of Cahill and colleagues who, among others, not only found that memory is better for emotionally arousing than neutral material (Bradley et al., 1992; Christianson & Loftus, 1991; Cahill et al., 1994, 1995, 1996; van Stegeren et al., 1998; Ikeda et al., 1998; Moayeri et al., 2000; Kazui et al., 2000), but also that this effect is mediated by ß-adrenergic mechanisms (Cahill et al., 1994; O’Carroll et al., 1999b; van Stegeren et al., 1998; but see O’Carroll et al., 1999a). It could therefore be the case that presentations of a discrete US trigger the release of NA, resulting in levels sufficient for DEX to have an effect. In other words, if the memory modulatory effects of glucocorticoids rely on ß-adrenergic activity, one could speculate that the reason why DEX did not enhance memory in any of the experimental paradigms in this chapter might be due to a lack of emotional arousal. Learning about the absence of an event or about the relationship between neutral stimuli, even if an indirect or secondary emotional value is present, i.e. associative learning in the absence of a discrete reinforcer, may hence not trigger any or sufficient levels of emotional arousal and subsequent central noradrenergic activity, in order for glucocorticoids to exert any

153

enhancing effects on memory consolidation. In contrast, other, more emotional, tasks (Chapter 2) may be susceptible to such a modulation due to sufficient endogenous NA levels as a consequence of the training reinforcer (US). An alternative possibility is that the dose of DEX was either too low or, given the proposed inverted U-shape function between glucocorticoids and memory consolidation, too high, to have had an effect. It is unlikely that it was too high, given that the same dose resulted in successful modulation of memory in the paradigms of Chapter 2, where additional endogenous corticosterone is likely to have aided the process. Further studies with higher doses of DEX, or concurrent administration of noradrenergic drugs, might prove helpful in the future to clarify these matters.

154

CHAPTER 4 Differential effects of glucocorticoids in the memory modulation of sensory and motivational reinforcer properties

155

The previous chapters demonstrated that glucocorticoids can modulate memory in discrete-cue Pavlovian paradigms of both appetitive and aversive nature (Chapter 2), and that a discrete US may well have to be present for such a modulation to occur (Chapter 3). This penultimate experimental chapter aims to investigate whether glucocorticoid memory modulation targets specific properties of the US. Two experiments, 4.1 and 4.2, employed a Pavlovian-instrumental transfer task to dissociate glucocorticoids’ role in learning about the sensory and motivational properties of a reinforcer. Results suggest that glucocorticoids might enhance learning about the latter, perhaps at a cost to the former.

4.1 GLUCOCORTICOIDS IN PAVLOVIAN-INSTRUMENTAL TRANSFER I (PILOT)

Glucocorticoids play an important role in the modulation of emotional memories. This has been shown a number of aversive (e.g. Flood et al., 1978; Pugh et al., 1997; Roozendaal & McGaugh, 1996a,b; Sandi & Rose, 1997) and more recently also in appetitive paradigms (this thesis, Chapter 2). However, very little is known about whether such a modulation targets specific aspects of the CS-US association. When an animal learns about the predictive relationship between a previously neutral stimulus in the environment (CS) and an emotionally significant stimulus (US), for example a food reward or a footshock, it forms association with representations of different aspects of the reinforcer. This idea goes back at least to Konorski (1967; see Dickinson & Dearing, 1978) who proposed that different types of conditioned behaviour are controlled by different properties of the reinforcer. Preparatory

156

conditioned responses, such as a general increase in locomotor activity in anticipation of reward, or fear-potentiated startle, are generalised responses controlled by primary affective, or protopathic, properties of the US which I shall refer to from now as motivational properties. In contrast, consummatory conditioned responses frequently mimic those elicited by the US and are reinforcer-specific. For example, Moore (1973) found that pigeons peck differently to solid food than to liquid water reinforcement. Consummatory responses are controlled by the idiopathic, or, as I shall refer to them, the sensory properties of the US. In other words, the motivational properties refer to the general arousing, affective nature of a reinforcer, in some sense its appetitive- or aversiveness, whereas the sensory properties reflect the unique properties of a reinforcer, for example the precise taste and texture of a food reward. In order to investigate if glucocorticoids modulate associations between the CS and selectively the sensory or the motivational properties of the US, a PavlovianInstrumental transfer paradigm (PIT) was employed (Kruse, Overmier, Konz, & Rokke, 1983). Reinforcer-specific PIT allows a separate analysis of the sensory and motivational properties of a US by making use of the fact that Pavlovian stimuli can exert control over instrumental action by an outcome-specific as well as a general process. The general process refers to the finding that instrumental responding during any reward signal may be energised due to a non-specific arousal effect that is related to the valence, but not identity, of the signalled reward. The outcome-specific process constitutes a selective elevation in a specific instrumental response during a reward signal that shares the same outcome with that instrumental action. For example, if, during presentation of a tone that predicts delivery of food, a rat is pressing on one lever (A) it has been trained on for food (X) and another lever (B) that it has been trained on for sucrose (Y), an elevation of lever pressing is observed on both levers as

157

compared to ISI (general process), but lever pressing is greater on lever A than on lever B (outcome-specific process). Only the latter provides information about the sensory properties of a US. The aim of the current study was to make use of these findings and employ a PIT design in order to investigate whether memory modulation of glucocorticoids is selective for specific properties of the US. Experiment 4.1 is a pilot study, its primary role to get an indication if such selectivity exists, and to optimise parameters for a possible further study.

Method Subjects Subjects were 8 naive male, Lister Hooded rats (Harlan, Bicester, UK), weighing between 360 g and 435 g (M = 391 g) at the start of the experiment. For housing and feeding information, refer to Experiment 2.1. The experiment was carried out between 08:00 and 12:00. Apparatus Refer to Experiment 2.3. Each chamber contained a central, recessed magazine that provided access to food (45 mg, Formula A/I; Noyes, USA) or sucrose reinforcement (20% solution) delivered by a pellet dispenser and a liquid dipper respectively, and two retractable levers that were located on each side of the magazine. Apart from clicker (10 Hz, 75 dB, 2 min), a 3-kHz tone (75 dB, 2 min), produced by lab-manufactured tone generator and delivered through a speaker mounted in the side of the chamber, was available as further discrete auditory cue. The background noise in the chamber was 60 dB.

158

Drugs and injection procedure DEX (Sigma, UK), at a dose of 1.2 mg/kg in a volume of 4.0 ml/kg, or SAL was injected (sc) immediately post-training. For preparation details, refer to Experiment 2.2. Behavioural procedure Magazine and lever training: All animals received two, 30-min magazine training sessions, with the levers removed from the chambers, one during which food pellets, and one during which sucrose, were delivered according to a VT-30 schedule. In the next 30-min session, one lever, either on the left or the right of the magazine was introduced into the chamber and pressing was reinforced with either food or sucrose under a VI-7 schedule. This pattern was repeated over the next two sessions with increasing reinforcement schedules of VI-15 and VI-30. In further sessions, the opposite lever was reinforced with the other reward on a VI-30 schedule. Finally, VI30 and VI-60 sessions were given during which both levers and both rewards were present. When an animal did not respond satisfactorily to a new schedule, it received further training with that schedule until all animals responded equally well. Rats were trained either on the left lever for food and the right lever for sucrose, or vice versa. The order of levers trained and lever-reward relation were counterbalanced. Magazine and lever training occurred over a period of 24 days. Conditioning: Over a period of six days, animals were given conditioning sessions with post-training treatment. Sessions lasted 31 min and comprised six conditioning trials, each of which constituted a 2-min presentation of either clicker or tone with food pellets or sucrose being delivered during the second minute of that stimulus, according to a random interval (RI)-30 schedule. ISI were 3 min long. On odd days, rats received one possible stimulus-reward relation (tone-pellet, tone-sucrose, clicker-pellet or clicker-

159

sucrose) followed by administration of either DEX or SAL, and on even days, they received the alternative combination of stimulus, reward and treatment. These parameters were counterbalanced across the conditioning sessions and also with respect to the different instrumental contingencies learned previously. Magazine approach was recorded during the CS presentations and matched ISI periods. Drug treatment occurred immediately after the session, approximately 3-6 min after the last stimulus presentation. Animals were returned to their home cage immediately after this treatment. Transfer-of-Control Test: All rats received two reminder sessions of lever-pressing, and then, on each of the next two days, a test-session during which both levers were made available in extinction. Three non-reinforced presentations of both the tone and the clicker occurred in a pseudo-randomised order. Stimuli were 2 min long with a 3-min ISI. Responses on both levers were recorded during the stimuli presentations and during the ISI. For the purposes of the transfer-test, each lever was assigned the label SAME during one type of stimulus and DIFF (‘different’) during the other, depending on whether responding to that lever during instrumental training led to the same or to a different reward as predicted by the current Pavlovian stimulus. See Table 4.1.1 for an outline of the experimental design. Statistical analysis Refer to Experiment 2.1.

160

Table 4.1.1. Experimental design of Experiment 4.1 (Glucocorticoids in PIT I). The Instrumental Training phase also contained magazine training. SAL = saline control treatment; DEX = dexamethasone treatment; RL = right lever; LL = left lever; A & B = tone or clicker; X & Y = food pellets or sucrose. All actions, stimuli, reinforcers and treatments were counterbalanced. Instrumental Training

Pavlovian Conditioning

Transfer Test

Pre-experiment, Day 7 & 8

Day 1, 3, 5

Day 2, 4, 6

Day 9 & 10

RL ? X LL ? Y

6xA? X +DEX

6xB? Y + SAL

3xA? 0 3xB? 0 Levers available

Results Pavlovian Conditioning. Figure 4.1.1 compares magazine approach during stimuli that were followed by either SAL or DEX treatment after each conditioning session, across all 18 trials, displayed in bins of two trials. DEX did not affect the rate of conditioning. Magazine approach, defined as the number of entries into the magazine during the 2-min CS minus the adjusted ISI period, transformed into 1-min scores, served as the dependent variable. Three analyses were carried out, one comprising only the pre-treatment trials of the first conditioning session, one comprising the post-treatment trials of the second and third conditioning session, and a final analysis directly comparing pre- and post-treatment trials. For pre-treatment trials, a two-way, 2 x 6, within-subject ANOVA with TREATMENT (SAL, DEX) and TRIAL (conditioning trials 1-6) as repeated measures found no significant main effect of TREATMENT, F(1, 7) < 1, (Ms: SAL = 21.18, DEX = 18.81), and no significant interaction TREATMENT x TRIAL, F(5, 35) < 1, confirming homogeneous pre-treatment responding. There was a significant main effect of TRIAL, F(5, 35) = 4.81, p < .01.

161

Magazine Entries / min (CS - adjusted ISI)

16 14 12 10 8 6 4 SAL DEX

2 0 1

2

3

4

5

6

7

8

9

Bins of 2 Trials

Treatment

Figure 4.1.1. Pavlovian Conditioning with two different stimuli (tone and clicker) predicting two different reinforcers (food and sucrose) where sessions were followed either by SAL or DEX treatment. Sessions comprised three blocks of two trials. Arrows indicate treatment.

162

Newman-Keuls posthoc analysis showed that responding on trial 1 was less than on all other trials (all ps < .02), indicating acquisition. For post-treatment trials, a twoway, 2 x 12, within-subject ANOVA with TREATMENT (SAL, DEX) and TRIAL (conditioning trials 7-18) as repeated measures again found no significant main effect of TREATMENT, F(1, 7) < 1, giving no indication of glucocorticoid memory modulation (Ms: SAL = 21.26, DEX = 24.08). There was no significant main effect of TRIAL, F(11, 77) < 1, indicating that response levels were at ceiling. The interaction of TREATMENT x TRIAL was not significant, F(11, 77) = 1.06, p = .41. Finally, pre-treatment trials (1-6) and post-treatment trials (7-12) were collapsed into single data sets, and a two-way, 2 x 2, within-subject ANOVA was carried out with TREATMENT (SAL, DEX) and PHASE (pre-, post-treatment) as repeated measures. There were no significant main effects of TREATMENT, F(1, 7) < 1, or TRIAL, F(1, 7) = 1.76, p = .23, and no significant interaction, F(1, 7) = 1.88, p = .21. Results show that response levels reached ceiling early on and hence do not provide evidence for memory enhancement by post-training DEX. Pavlovian-Instrumental transfer. General transfer effect. Figure 4.1.2 displays the general transfer effect which was found with both treatments. It was investigated by comparing baseline lever pressing during the ISI-period to lever pressing during those stimuli that were followed by SAL during the Pavlovian stage and to those that were followed by DEX, irrespective of lever or reward. For ISI responding, lever pressing on each lever during the ISI periods was expressed in LP/min rates and the mean was calculated from the two levers. For lever pressing during the stimuli, an average was calculated for LP during Dex-Same and DexDifferent, as well as for LP during SAL-Same and SAL-Different. These scores were also adjusted for an LP/min rate.

163

Lever Presses / min

6

*

*

5 4 3 2 1 0 CS

ISI

CS

DEX

SAL Condition

Figure 4.1.2. General transfer effect. Overall lever pressing (irrespective of lever) is greater during any reward-signalling stimulus as compared to ISI, irrespective of posttraining treatment of that stimulus in the Pavlovian conditioning stage (DEX or SAL).

* indicates statistical significance (p = .05).

164

A one-way ANOVA was calculated for ISI, SAL and DEX as the three levels of the STIMULUS factor, and a significant main effect was found, F(2, 14) = 3.96, p < .05. Newman-Keuls posthoc analysis revealed that there was significantly less responding during the ISI (M = 3.47) than during SAL-stimuli (M = 4.74), p < .05, and than during DEX-stimuli (M = 4.46), p = .05, while responding did not differ between DEX- and SAL-stimuli (p = .57). Outcome-specific transfer effect. Figure 4.1.3 demonstrates the presence and absence of the outcome-specific transfer effect with SAL and DEX treatment respectively by comparing lever pressing rates during stimuli that share the same or a different outcome with the instrumental response on the given lever. Trials were collapsed and a two-way, 2 x 2, within-subject ANOVA was carried out with LEVER (SAME, DIFF) and TREATMENT (DEX, SAL) as the two factors. LEVER refers to the relationship of a lever and a stimulus. TREATMENT refers to the drug administered after Pavlovian conditioning to the stimuli. There was no main effect of LEVER, F(1, 7) < 1 (Ms: SAME = 4.73, DIFF = 4.47) and no main effect of TREATMENT, F(1, 7) < 1 (Ms: DEX = 4.46 , SAL = 4.74). The interaction of LEVER x TREATMENT also failed to reach significance, F(1,7) = 2.26, p = .18. However, given that this was a pilot-study, simple-effect analysis was carried out nevertheless and revealed that for stimuli followed by SAL treatment in Pavlovian conditioning, lever pressing was greater on the lever that shared an outcome (SAME, M = 5.31) than on the lever that did not share an outcome (DIFF, M = 4.17) with that stimulus, p < .03. This constitutes a specific transfer effect. In contrast, for stimuli that were followed by DEX treatment in Pavlovian conditioning, lever pressing on the SAME (M = 4.15) and on the DIFF lever (M = 4.78) did not differ, p = .46.

165

Lever Presses / min

6

*

5 4 3 2 1 0 SAME

DIFF

SAME

SAL

DIFF

DEX Condition

Figure 4.1.3. Outcome-specific transfer effect. For SAL-paired but not DEX-paired stimuli, lever pressing is greater on the lever that shares an outcome with that stimulus (SAME) than on the other lever (DIFF). * indicates statistical significance (p < .05)

166

Discussion DEX given after Pavlovian conditioning trials to one stimulus abolished the outcome-specific transfer effect. This effect was seen during presentations of the other, saline-control, stimulus. The general transfer effect was not affected by DEX. This suggests that administration of DEX after Pavlovian conditioning trials compromised information about the sensory properties of the US. This might seem counterintuitive at first since DEX has been shown to enhance conditioning in chapter 3, as well as in inhibitory avoidance studies by others (e.g. Roozendaal & McGaugh, 1996a). However, these studies did not dissociate the sensory and motivational properties of the reinforcer. It is possible that enhancing information about the motivational properties might come at the cost of reducing information about the sensory properties. In fact, freezing and conditioned suppression, the measures of aversive Pavlovian conditioning enhanced by DEX treatment in Chapter 2, represent classic examples of preparatory behaviour, thought to be controlled by the motivational properties of the US. However, in the current study there was no increase in the general transfer effect for stimuli followed by DEX treatment. Furthermore, there was no apparent enhancement of Pavlovian conditioning by DEX treatment either. The parameters used may not have been sensitive enough to reveal such effects. However, the compromised outcome-specific transfer effect was sufficiently encouraging to suggest I carry out a further PIT study. A similar pattern of results in the presence of evidence for DEX-induced enhancements of Pavlovian conditioning would strengthen an argument that glucocorticoid memory modulation might be selective for the motivational properties of the reinforcer, and might come at the cost of losing sensory information.

167

4.2 GLUCOCORTICOIDS IN PAVLOVIAN-INSTRUMENTAL TRANSFER II

The aim of the study was to replicate the findings of Experiment 4.1, hopefully in combination with a DEX-induced enhancement of Pavlovian conditioning. The method was identical to that of the pilot study with a few exceptions. The number of subjects was increased to 16 to enhance statistical power and reduce variability. The lever training phase was completed much more rapidly. Finally, a Pavlovian extinction test was carried out after the transfer test to obtain additional measures of Pavlovian conditioning.

Method Subjects Subjects were 16 naive male, Lister Hooded rats (Harlan, Bicester, UK), weighing between 257 g and 290 g (M = 273 g) at the start of the experiment. For housing and feeding information, refer to Experiment 2.1. The experiment was carried out in the afternoons between 14:00 and 18:00. Apparatus The apparatus was the same as in Experiment 4.1. Drugs and injection procedure DEX (Sigma, UK), at a dose of 1.2 mg/kg in a volume of 6.0 ml/kg, or SAL was injected (sc) immediately post-training. For preparation details, refer to Experiment 2.2.

168

Behavioural procedure The behavioural procedure was identical to that of Experiment 4.1, except the following differences: Magazine and lever training was completed within one week. Reinforcement during Pavlovian conditioning followed an RI-15 rather than an RI-30 schedule. Finally, two Pavlovian extinction test sessions were carried out at the end of the experiment on days 10 and 11. These mimicked the transfer-of-control sessions with rats receiving three non-reinforced presentations of both tone and clicker, except that the levers were not available. Magazine approach was recorded during the CS presentations and matched ISI periods. See Table 4.2.1 for an outline of the experimental design. Statistical analysis Refer to Experiment 3.2.

Table 4.2.1. Experimental design of experiment 4.2. (Glucocorticoids in PIT II). The Instrumental Training phase also contained magazine training. SAL = saline control treatment; DEX = dexamethasone treatment; RL = right lever; LL = left lever; A & B = tone or clicker; X & Y = food pellets or sucrose. All actions, stimuli, reinforcers and treatments were counterbalanced.

Instrumental Training Pre-experiment, Day 7 RL ? X LL ? Y

Pavlovian Conditioning Day 1, 3, 5 6xA? X + DEX

Transfer Test

Pavlovian Extinction

Day 2, 4, 6

Day 8 & 9

Day 10 & 11

6xB? Y + SAL

3xA? 0 3xB? 0 Levers available

3xA? 0 3xB? 0 Levers not available

169

Results Pavlovian Conditioning. Figure 4.2.1 compares magazine approach during stimuli that were followed by either SAL or DEX treatment during conditioning, both in pre-treatment (conditioning session (C) 1) and post-treatment (C 2,3; extinction sessions (E) 1,2) phases. Magazine approach, defined as time spent in the magazine during the CS minus the ISI period served as the dependent variable. Trials were collapsed to form a mean score for each conditioning or extinction session. In order to normalise the distribution, these mean scores underwent a square-root transformation. For conditioning session 1, prior to any treatment, a t-test was carried out to compare pre-SAL and pre-DEX responding. Equivalent pre-treatment responding was confirmed as the two samples did not differ (Ms: pre-SAL = 2.58, pre-DEX = 3.34), p = .43. The remaining sessions were classed as post-treatment, and a two-way, 2 x 4, within-subject ANOVA with TREATMENT (SAL, DEX) and SESSION (C 2,3; E 1,2) as repeated measures was carried out. There was a significant main effect of TREATMENT, F(1, 15) = 5.50, p < .03, revealing greater magazine approach with DEX (M = 2.87) than with SAL treatment (M = 1.67). A significant main effect of SESSION, F(3, 45) = 30.03, p < .001, showed that magazine approach differed between sessions. Newman-Keuls post-hoc analysis revealed that magazine approach in both extinction sessions (Ms: E1 = 0.55, E2 = 0.22) was lower than either conditioning session (Ms: C2 = 4.36, C3 = 3.93), all ps < .001, confirming the occurrence of extinction. The interaction of TREATMENT x SESSION was not significant, F(3, 45) = 1.25, p = .30. As a second measure of Pavlovian conditioning, CS scores of extinction trials were analysed with a two-way, 2 x 6, within-subject ANOVA, with TREATMENT (SAL, DEX) and TRIAL (extinction trials 1-6) as repeated measures. A significant main effect of TREATMENT, F(1, 15) = 7.57,

170

Magazine Approach (CS-ISI, time (s), sqrt)

5 SAL DEX

4 3 2 1 0 -1 C1

C2

C3

E1

E2

Session Treatment Figure 4.2.1. Pavlovian Conditioning and extinction with two different stimuli (tone and clicker) predicting two different reinforcers (food and sucrose) where sessions were followed either by SAL or DEX treatment. Conditioning (C) and extinction (E) sessions comprised six and three trials respectively. Arrows indicate treatment.

171

p < .02, with greater responding on DEX-paired (M = 11.08) than SAL-paired (M = 6.11) stimuli is shown by Figure 4.2.2. There was no main effect of TRIAL, F(5, 75) < 1, and no significant interaction, F(5, 75) = 1.19, p = .32. Pavlovian-Instrumental transfer. General transfer effect. Figure 4.2.3 displays the general transfer effect. This compares baseline lever pressing during the ISI-period to lever pressing during those stimuli that were followed by SAL or DEX during Pavlovian conditioning, irrespective of lever. For ISI responding, lever pressing on each lever during the ISI periods was expressed as LP/min rates and the mean was calculated from the two levers. For lever pressing during the stimuli, an average was calculated for LP during DEX-SAME and DEX-DIFF, as well as for LP during SAL-SAME and SAL-DIFF. These scores were also adjusted for an LP/min rate. A one-way ANOVA was calculated for ISI, SAL and DEX as the three levels of the STIMULUS factor, and a significant main effect was found, F(2,30) = 6.32, p < .01. Newman-Keuls posthoc analysis revealed that there was less responding during the ISI (M = 2.97) than during DEX-stimuli (M = 3.69), p < .03, and also than during SAL-stimuli (M = 4.01), p < .01, while responding did not differ between DEX- and SAL-stimuli (p = .30). Outcome-specific transfer effect: Figure 4.2.4 demonstrates the presence and absence of the outcome-specific transfer effect with SAL and DEX treatment respectively by comparing lever pressing rates during stimuli that share the same or a different outcome with the instrumental response on the given lever. Trials were collapsed and a two-way, 2 x 2, within-subject ANOVA was carried out with LEVER (SAME, DIFF) and TREATMENT (DEX, SAL) as the two factors. LEVER refers to the relationship of a lever and a stimulus. There was no main effect of LEVER, F(1, 15) = 3.16, p = .10 (Ms: SAME = 4.06, DIFF = 3.64) and no main effect of

172

Magazine Approach during CS (s)

14

* 12 10 8 6 4 2 0 SAL

DEX

Treatment

Figure 4.2.2. Magazine approach during extinction of Pavlovian Conditioning of two different stimuli (tone and clicker) predicting two different reinforcers (food and sucrose) where sessions were followed either by SAL or DEX treatment. Data are collapsed over 6 extinction trials. * indicates statistical significance (p < .05).

173

5

*

Lever Presses / min

* 4 3 2 1 0 CS

ISI

CS

DEX

SAL Condition

Figure 4.2.3. General transfer effect. Overall lever pressing (irrespective of lever) is greater during any reward-signalling stimulus as compared to ISI, irrespective of posttraining treatment of that stimulus in the Pavlovian conditioning stage (DEX or SAL). * indicates statistical significance (p < .03).

174

6

Lever Presses / min

* 5 4 3 2 1 0 Same

Diff.

Same

SAL

Diff.

DEX Condition

Figure 4.2.4 Outcome-specific transfer effect. For SAL-paired but not DEX-paired stimuli, lever pressing is greater on the lever that shares an outcome with a stimulus (SAME) than on the other lever (DIFF). * indicates statistical significance (p < .04).

175

TREATMENT, F(1,15) < 1 (Ms: DEX = 3.69 , SAL = 4.01). The interaction of LEVER x TREATMENT was marginally significant, F(1,15) = 4.01, p = .06. Simpleeffect analysis revealed that for stimuli followed by SAL treatment in Pavlovian conditioning, lever pressing was greater on the lever that shared an outcome (SAME, M = 4.73) than on the lever that did not share an outcome (DIFF, M = 3.28) with that stimulus, p < .04. This constitutes a specific transfer effect. In contrast, for stimuli that were followed by DEX treatment in Pavlovian conditioning, lever pressing on the SAME (M = 4.00) and on the DIFF lever (M = 3.38) did not differ, p = .25.

Discussion The discussion of Experiment 4.2 is included in the General Discussion (4.3) below.


Experiment 4.2 investigated the question whether glucocorticoid memory modulation was selective for associations between the CS and certain properties of the US. It was found that memory modulation was bidirectional, selectively enhancing and impairing associations with different properties of the reinforcer. While learning about the sensory properties was impaired by post-training glucocorticoids, as it was the case in (pilot-) Experiment 4.1, their memory enhancing effect seen in Pavlovian conditioning is likely to be due to strengthened associations with the motivational properties of the US. Many, if not all, reinforcers possess sensory properties, reflecting how they are perceived by the senses and what makes them distinguishable from other reinforcers.

176

Reinforcers also possess motivational properties, revealing the direction of their emotional valence. A selective PIT design allows a separate analysis of these properties by making use of the fact that Pavlovian stimuli can exert control over instrumental responding by two processes (Kruse et al., 1983; Trapold & Overmier, 1972). For example, if a tone has reliably signalled food, instrumental responding for any appetitive reinforcer is increased during the occurrence of that tone as compared to the ISI. This is due to a general arousal process as both the Pavlovian and the instrumental outcome share their motivational properties. If lever pressing for a reinforcer during the occurrence of a stimulus that predicts the same reinforcer is compared to lever pressing for another reinforcer, responding is greater for the former than the latter. In other words, responding is enhanced further when the reward-signal and the reward-response share not only the motivational but also the sensory properties. Hence, this effect can only be due to learning about the sensory properties of the reinforcer. The current experiment employed this design in order to specifically analyse glucocorticoid memory modulation of Pavlovian conditioning, by administering post-training glucocorticoids and saline control during the Pavlovian stage of the design. Appetitive

Pavlovian

conditioning

was

enhanced

by

post-training

administration of the glucocorticoid receptor agonist DEX. This is consistent with previous findings reported in this thesis (Chapter 3) and from other laboratories (e.g. Flood et al., 1978; Roozendaal & McGaugh, 1996a,b; Sandi & Rose, 1994a,b; Veldhuis et al., 1985) and supplies further evidence that glucocorticoids modulate memory. The fact that animals were drug-free before and during conditioning, and received treatment after the training events, suggests an effect on memory consolidation unconfounded by possible effects on attentional, motivational or

177

sensory-perceptual mechanisms. It furthermore supports the idea that glucocorticoid memory modulation is not selective for learning about the context, as it has been suggested to be the case at least in fear conditioning (Pugh et al., 1997), but instead that it can be shown in appetitive, discrete-cue Pavlovian conditioning paradigms (this thesis, Chapter 3). Finally, it also constitutes a rare demonstration of glucocorticoid memory modulation using a within-subject design where all subjects were exposed to all treatments at some point during the experiment, while most previous studies employed between-subject designs. A general enhancement of lever pressing during the reward-signalling stimuli as compared to ISI was observed, both with DEX and with SAL treatment given after Pavlovian conditioning of those stimuli. This constitutes a typical example of the general transfer effect. However, there was no difference between the two treatment conditions. The outcome-specific transfer effect is reflected by greater responding on the lever that shares exactly the same outcome with the activated Pavlovian stimulus than on the other lever. This was only the case with stimuli followed by SAL during Pavlovian conditioning. The fact that DEX enhanced Pavlovian conditioning but, at the same time, abolished the outcome-specific, sensory-dependent transfer effect, constitutes the main finding of this study. The results suggest that post-training glucocorticoids selectively enhance associations between predictive stimuli in the environment and the representation of the motivational properties of a reinforcer. In this context, the term motivational could be exchanged with emotional, affective or protopathic – referring to those properties of the reinforcer that make it a biologically significant event for the organism. The sensory, or hedonic or idiopathic, properties of the reinforcer, in contrast, determine its specific qualities and reflect its particular experience. The impairment in the representation of the sensory properties of the US

178

following DEX treatment suggests that glucocorticoids may trigger processes by which the sensory detail of the US is overshadowed by its motivational characteristics. Given the DEX-induced impairment of associations with the sensory properties of the US, the memory-enhancing effects seen in Pavlovian conditioning can only be due to strengthened connections with the motivational properties of the US. However, the question remains why there was no enhanced general transfer effect with DEX. It is possible that general responding during a reward signal was at ceiling and that the test was not sensitive enough to detect a DEX-induced facilitation of the general transfer effect. Another explanation could be that strengthened associations with the motivational US properties may not have access to instrumental performance. In contrast, a deficit in sensory processing, as it is expressed as a bias, must influence instrumental performance. The findings also have some interesting implications for general models of amygdala functioning in learning and memory. The BLA is thought to play a key role in glucocorticoid memory modulation (e.g. Roozendaal & McGaugh, 1997a; Roozendaal, 2000), integrating hormonal and neuromodulatory factors on memory consolidation. Glucocorticoid effects on memory require intact ß-adrenergic activity within the BLA (Quirarte et al., 1997). Furthermore, Killcross and Blundell (in press) have proposed that the BLA plays an important role in processing the sensory properties of the reinforcer and integrating their representation alongside the motivational properties. Interestingly, lesions of the BLA have been found to impair the outcome-specific transfer effect (Blundell, Hall, & Killcross, 2001), in a manner similar to the impairment seen by DEX in the current experiment, while leaving the general transfer effect intact. It is possible that processing of sensory and motivational properties of a reinforcer are in competition with each other and that memory

179

modulatory glucocorticoid effects in the BLA occur at the expense of its resources involved in generating a mental representation of the sensory properties of a reward. In other words, there may be a functional shift in the BLA with increasing levels of stress, with the motivational properties of the US taking priority in processing (possibly involving the CEA), while processing of the sensory properties is reduced. This functional shift may constitute the neural basis of the overshadowing mechanism mentioned above. The results imply that glucocorticoid memory enhancement comes at a price. Processing of detail may be lost for the greater good of enhanced motivational evaluation. All demonstrations of glucocorticoid memory modulations, both in this thesis and elsewhere, involved relatively simple, contextual or discretecue, first-order Pavlovian conditioning paradigms. In fact, as it was shown in Chapter 3, DEX failed to modulate memory in non-US learning paradigms. The glucocorticoid-modulated memories seem to be of a simpler type, i.e. restricted to the motivational aspects of an outcome. Some support for the view that emotional arousal leads to differential processing and memory consolidation comes from a recent study by Adolphs et al. (2001) of amygdala functions in humans. They found that patients with unilateral damage of the amygdala, as well as brain-damaged and normal controls, remembered the gist of visual stimuli better than the detail when the stimuli were (aversively) emotional rather than neutral. In other words, emotional arousal enhanced memory for the gist, and impaired memory for the detail of visual stimuli. Furthermore, Adolphs and colleagues found that this pattern was completely reversed in a patient with bilateral amygdala damage. The subject remembered the gist of emotional stimuli worse, and detail better, than controls. Similarly, Christianson and Loftus (1991) found that the central aspect of a visual slide was better remembered if it was of

180

emotional rather than neutral content. Conversely, peripheral aspects were remembered better in neutral slides. However, these results could be explained entirely by attentional processes (attentional-narrowing hypothesis). Yuille and Cutshall (1986) conducted a case study of eyewitness memory of a crime, and found that some sensory aspects of the event tended to be forgotten, while other, more emotionally significant (motivational) aspects were well retained: For example, one witness graphically described the wounds on the body…Whereas this information was highly accurate, she erroneously described him as wearing a T-shirt and a red and black jacket. He actually wore a dark blue sweater and a blue jean jacket. (Yuille & Cutshall, 1986, p.296)

Wessel and Merckelbach (1998) found enhanced free recall of central, threatrelevant, and impaired recall of peripheral, threat-irrelevant, information in spider phobics. While many theories of memory enhancement for the emotional aspects of an event concentrate on increased attention as a cause, the reported studies here suggest that differential consolidation processes may also play a role.

181

CHAPTER 5 Individual differences in learning due to differences in HPA-reactivity ?

182

The last empirical chapter does not involve any exogenous glucocorticoid administrations. Instead, it aimed to set up an endogenous manipulation by making use of known individual differences in HPA-reactivity which, so far, have not been studied in the light of glucocorticoid modulation of learning and memory. Typically, a population of outbred rats may be divided, by median-split, into high (HR) and low responders (LR) on the basis of their locomotor response to novelty. HR are more HPA-reactive in the sense that, as compared to LR, they show a higher and prolonged corticosterone response to a stressor such as novelty (Piazza et al., 1991a). HR have also been found to self-administer amphetamine more readily than LR (Deminiere, Piazza, Le Moal, & Simon, 1989; Piazza et al., 1989; Piazza et al., 1990). Furthermore, HR were found to show a higher and longer stress-induced increase in DA in the nucleus accumbens (Rouge-Pont et al., 1993) and a reduction of DA in the PFC (Piazza et al., 1991b). This difference between HR and LR in mesolimbic DA in response to stress depends on corticosterone release and can be blocked by ADX (Rouge-Pont, Deroche, Le Moal, & Piazza, 1998; Barrot et al., 2000; but see Imperato, Puglisi-Allegra, Casolini, & Angelucci, 1991). Differences in locomotor activity to novelty seen between 4-months-old HR and LR were no longer apparent after 16 and 21 months but the increased corticosterone response to novelty remained. High responsiveness to novelty has been linked to the sensation-seeking personality-trait in humans and HR were thought to actually seek novelty (Dellu et al., 1993, 1996a). Kabbaj et al. (2000) characterised HR and LR further, both on a behavioural level in tests of anxiety, and on a neurobiological level, looking at expression of stress-related hormones. They found that HR explored anxiogenic environments more readily than LR, despite, or possible because of, greater corticosterone responses, confirming the proposal of increased novelty-seeking.

183

Furthermore, HR and LR had distinct expression patterns of certain stress-related molecules. HR had lower levels of hippocampal GR mRNA while MRs were not expressed differently. These levels of hippocampal GRs were linked to the noveltyseeking behaviour, since LR receiving a GR antagonist behave indistinguishably from HR. Greater levels of stress-induced corticosterone might down-regulate hippocampal GRs by a negative-feedback mechanism. Finally, Cools and colleagues (Cools et al., 1993; Cools and Gingras, 1998) found a bimodal shape of individual variation in Wistar rats, with HR and LR showing distinct genetic, neuroendocrinological, immunological, pharmacological and behavioural patterns. In addition to the openfield test on locomotor reactivity to novelty, similar to that of Piazza’s groups, “Nijmegen high and low responders” can also be classified by an intruder test, resulting either in a fleeing or non-fleeing response, and by an apomorphine test, measuring gnawing spells after apomorphine administration. In line with Piazza et al.’s proposals, Cools and Gingras (1998) outlined a relationship between responsiveness, the neurochemical state of the nucleus accumbens and drug addiction. They argued that a predisposition to abuse drugs depends on an interaction between an individual’s genotype, i.e. their predisposition to become a HR or LR, early postnatal factors that may determine the phenotype, and the degree of stress experienced when experiencing the addictive drug. Piazza and colleagues proposed a role of glucocorticoids as a substrate of reward (Piazza & Le Moal, 1997), based on a number of observations. First of all, glucocorticoids are secreted in response to rewarding stimuli such as food, sex or drugs of abuse (Caggiula et al., 1991; Fuller & Snoddy, 1981; Honma et al., 1984; Merali et al., 1998; Orchinik et al., 1988). Secondly, they influence reward-related behaviour, e.g. the self-administration of amphetamine (Piazza et al., 1991a) and

184

cocaine (Mantsch, Saphier, & Goeders, 1998). Finally, Piazza and Le Moal (1997) also claim that glucocorticoids themselves have positive reinforcing effects, pointing to the self-administration and abuse of corticosterone by rats and humans respectively (Deroche, Piazza, Deminiere, Le Moal, & Simon, 1993; Dixon & Christy, 1980; Piazza et al., 1993; but see Broadbear, Winger, & Woods, 1993), and to psychopathological conditions like Addison’s disease that are characterised by both anhedonia and hypocortisolism (although the same argument with opposite conclusions could be made for many cases of major depression). These reward-related effects and observations, which have led Piazza and Le Moal (1997) to describe glucocorticoids as “endogenous state-dependent psychostimulants” (p.364), are thought to result from glucocorticoids’ above mentioned modulatory influence on the mesolimbic DA system (Piazza & Le Moal, 1997). Based on these assumption and studies, Piazza and Le Moal (1996, 1997) suggested that there may be a relationship between HPA-reactivity and propensity to drug addiction. According to Piazza and colleagues, a greater and prolonged corticosterone-release in response to novelty or to a reinforcer such as amphetamine, as displayed by HR, enhances the phasic release of DA in the shell of the nucleus accumbens, in the case of amphetamine making the drug more reinforcing, and predisposing HR individuals to develop drug addiction (Piazza & Le Moal, 1996, 1997; Piazza et al., 1996). One should note though that the mesolimbic DA system is also activated in response to aversive (e.g. Abercrombie, Keefe, DiFrischia, & Zigmond, 1989; Imperato et al., 1992) and neutral events (Young, Ahier, Upton, Joseph, & Gray, 1998). While such proposals about a relationship between stress, novelty, glucocorticoids, DA and drug addiction sound compelling and may well explain the behavioural data, it is interesting, and perhaps surprising, that no attempts have been

185

made to consider the memory modulatory effects of glucocorticoids in this context. The fields of drug addiction and learning and memory have converged progressively over the last few years. During a recent conference (Irvine, USA) in a symposium termed “Addiction and memory: common mechanism ?”, Robbins and Everitt (2001) described drug addiction as “a pathological subversion of normal brain learning and memory processes strengthened by the motivational impact of drug-associated stimuli, leading to the establishment of compulsive drug-seeking habits.” (abstract 14, no page number). Excessive glucocorticoid release in response to stress (including appetitive reinforcers such as drugs) as might be seen in HR, may constitute one way by which normal learning and memory processes such as the formation of associations between stimuli in the environment and the rewarding properties of the drug experience, might be strengthened in a pathological and non-adaptive way. For example, drug self-administration in Piazza et al’s experiments is basically a discrimination learning task as animals have to discriminate between two holes in order to self-administer the drug. Nose-pokes to one but not the other hole lead to drug administration. Hence, it seems possible that differences in learning, rather than, or in addition to, differences in reinforcer value, might contribute to the different levels of amphetamine self-administration in HR and LR. In order to test this hypothesis, and to attempt to bring together the strands of research by Piazza and colleagues, and the glucocorticoid memory modulation literature, a population of rats was screened for their locomotor response to novelty, and divided into HR and LR on the basis of this response. Subsequently, HR and LR were tested in a number of behavioural learning paradigms, with the general hypothesis that HR would learn more rapidly than LR. In other words, high, as compared to low, responsiveness,

186

could be seen as an additional endogenous glucocorticoid treatment, similar to the exogenous DEX-treatment in the studies reported in previous chapters. The locomotor activity to novelty task is documented in section 5.1. Experiments 5.2 and 5.3 compare HR and LR in simple Pavlovian conditioning paradigms of appetitive and aversive nature respectively. Experiment 5.4 investigates possible individual differences in instrumental learning and the formation of habits.

5.1 LOCOMOTOR ACTIVITY TO NOVELTY IN A POPULATION OF LISTER-HOODED RATS

In order to compare HR and LR in a number of learning tasks, a population of 64 outbred Lister Hooded rats was screened for individuals’ locomotor activity in a novel environment over a 2-hr period. This procedure was kept as similar as possible to that of Piazza and colleagues (e.g. Piazza et al., 1989, 1990, 1991a). According to the results of this screening procedure, rats were grouped into HR, LR and medium responders (MeR). The former two groups were then compared behaviourally in the experiments reported in this chapter.

Method Subjects Subjects were 64 naive male, Lister Hooded rats (Harlan, Bicester, UK), weighing between 194 g and 258 g (M = 222 g) shortly before the screening procedure. Apart from a single weighing process, rats were left unhandled during their first two weeks in the laboratory. They were housed in pairs in a temperature-

187

controlled colony room (21±2°C) and maintained on a 12-h light/12-h dark cycle (lights on at 08:00 am) with free access to water. During the course of all experiments, rats were fed approximately 15 g of food pellets per day to maintain their body weights. Feeding occurred in the late afternoon or evening, after experimental sessions. Screening was carried out in the afternoons between 16:00 and 18:00, as in Piazza et al.’s procedure, over a period of eight days. Apparatus Screening in this study took place in a set of eight rectangular boxes that each contained two infrared beams. Each time a rat crossed one of the lines between the infrared emitters and receivers, a recording took place. Despite their rectangular shape, the boxes differed in a number of ways to the rats’ home cage environment. They were larger (33.0-cm-wide x 15.5-cm-high x 39.0-cm-deep), had a grid floor without sawdust bedding, and were located in a different, much smaller, room. An Acorn A500 microcomputer (Acorn, UK) controlled the equipment and recorded the data. Procedure Rats were screened in sets of eight over the course of eight successive days. On each day at approximately 16:00, a group of rats was brought to the locomotor screening room, and rats were placed into individual cages. The recording software was started once all rats were placed in their cage, and the experimenter quietly left the room. Recording took place over the course of 2 hr. Shortly after 18:00, rats were returned to their home cages.

188

Results and Discussion A population of 64 rats was screened for their locomotor activity to novelty, and those individuals with the highest and lowest eight scores constituted the HR and LR groups. Following locomotor screening, eight rats whose locomotor scores were very close to the median of the overall population were excluded and passed on to another experiment. The screening procedure was very similar to that of Piazza and colleagues (e.g. Piazza et al., 1989, 1990, 1991a). However, a different strain, ListerHooded rats, was used. While some strains, e.g. Wistar rats, show a bimodal shaped variation of behaviour (Cools et al., 1993), the distinction between HR and LR is less clear in other strains. Norwegian Browns, for example, are marked by a unilateral Gaussian distribution, with the peak being found just between LR and HR. Lewis and Fisher rats are thought to be more homogenous. (Cools, personal communication, February 2001). Since the distribution of locomotor activity scores in the current screening procedure followed a normal distribution, it was decided, rather than dividing the population by a median-split, to polarise the groups further and to only label the top and bottom eight responders as HR and LR. The highest overall responder subsequently turned out to show extremely deviant behaviour in the learning experiments and was excluded from the group of HR and replaced by the 9th highest responder. The rest of the population was considered MeR. Hence, 56 rats (8 HR and 8 LR which were the main focus of subsequent analysis, 39 MeR and one extreme outlier) participated in all of the behavioural experiments to follow. They were run in the order of their cage labels and no reallocation of cages took place after the screening procedure. All subsequent reference to HR, LR and MeR is made according to this distinction unless otherwise indicated. In Experiment 5.4, HR and LR groups were extended to the top and bottom 10 responders for one analysis, in

189

order to increase statistical power. The method subsection Subjects is reduced in the following reports of Experiments 5.2 to 5.4, and reference is made to this section. A one-way ANOVA comparing responding in the different eight boxes was carried out in order to test for possible box effects. There was no box effect, F(7, 56) = 1.24, p = .30 (data not shown). Figure 5.1.1 displays an overall histogram of the entire population. Figure 5.1.2 shows the development of locomotor activity across 2 hr in HR, LR, MeR, the extreme outlier and in the overall population. Figure 5.1.3 compares overall responding of HR and LR. A two-sample t-test confirmed that HR were more active than LR, p < .001.

190

8

LR

MR

HR

7

Frequency

6 5 4 3 outlier 2 1

10 00 11 00 12 00 13 00 14 00 15 00 16 00 17 00 18 00

90 0

80 0

70 0

60 0

50 0

40 0

30 0

20 0

10 0

0

0

number of total infrared beam breaks (in bins of 50)

Figure 5.1.1. Histogram of the original population of 64 Lister Hooded rats according to their locomotor activity to novelty, measured as infrared beam breaks over a 2-hr period.

191

HR LR overall mean

250

MeR

beam breaks / 10 min

extreme outlier

200 150 100 50 0 10 20 30 40 50 60 70 80 90 100 110 120 time (min)

Figure 5.1.2. Locomotor activity, measured as beam breaks per 10-min (+SEM), in HR (n = 8) and LR (n = 8), as well as in MeR (n = 39), the extreme outlier and the overall population (N = 64), over a period of 120 min divided into 10-min bins.

192

*

total number of beam breaks

1600 1400 1200 1000 800 600 400 200 0 HR

LR

overall population

group

Figure 5.1.3. Locomotor activity, measured as total beam breaks over 2 hr (+SEM) in HR (n = 8) and LR (n = 8), as compared to the overall population (N = 64). * indicates statistically significant difference (p < .001).

193

5.2 HR VS. LR IN APPETITVE DISCRETE-CUE PAVLOVIAN CONDITIONING

The first paradigm in which HR and LR were compared was appetitive discrete-cue Pavlovian conditioning. A clicker stimulus predicted the delivery of a food pellet. The design was very similar to those of the appetitive conditioning experiments in Chapter 2. As there were no constraints due to drug treatment in the current experiment, the salience of the appetitive reinforcer was reduced from four food pellets in Experiments 2.3 and 2.4 to a single food pellet. This way a slower conditioning progress was anticipated, enhancing the chance to detect a group difference. It was hypothesized that HR may show greater conditioned responding than LR, possibly due to greater HPA-reactivity in response to the food reward, and hence greater memory modulatory influences on learning.

Method Subjects Refer to section 5.1, method and results/discussion. All experiments were carried out between 14:00 and 20:00. Apparatus Refer to Experiment 2.3. A 10-Hz train of clicks (clicker) with a sound level of 75 dB was available as discrete auditory cue. The background noise in the chamber was 60 dB. Food pellets (45 mg, Rodent Grain-Base Formula; BIO-SERV, USA) served as an appetitive reinforcer.

194

Behavioural procedure Magazine training: On each of the two days prior to the experiment, rats were given a 30-min session of magazine training during which food pellet reinforcement was delivered according to a VT-60 schedule. Animals were returned to their home cages immediately after each session. Conditioning: On days 1 to 5, animals were given conditioning sessions. Six conditioning trials occurred after 3, 8, 13, 18, 23, 28 min during a session lasting 30 min. Each trial comprised a 15-s presentation of the CS, clicker, terminating with the delivery of the food-pellet US. Magazine approach during the CS presentations, and matched preCS periods was recorded, both in form of the number of magazine entries (magazine entries) and the time spent in the magazine (magazine time). Extinction: On days 6 to 8, extinction test sessions took place, comprising 6 trials per session and lasting 30 min. CS presentations were equivalent to those in the conditioning session, except that no reinforcement was given. Trials were equally spaced, starting 3 min into the session, with an inter-trialinterval of 4 min 45 s. Again, magazine approach during the CS and preCS periods was recorded. See Table 5.2.1 for an outline of the experimental design. Statistical analysis Refer to Experiment 2.1. Table 5.2.1. Experimental design of Experiment 5.2. (HR vs. LR in appetitive conditioning). HR = high responders; LR = low responders; MT = magazine training. VT = variable time schedule.

PHASE GROUP HR LR

Magazine Training

Conditioning

Extinction

Pre-training

Day 1-5

Day 6-8

MT: VT-60

6x Clicker ? Food

6x Clicker ? 0

195

Results Figure 5.2.1 illustrates the course of appetitive Pavlovian conditioning and extinction in HR and LR, operationalised as magazine entries during the CS. Figure 5.2.2 compares the difference in mean extinction scores between HR and LR, using magazine time, collapsed across trials, as the dependent variable. HR showed greater levels of conditioned magazine approach than LR. For both pre-CS and CS data of both measures, a two-way, 2 x 8, mixed ANOVA was carried out with GROUP as the between-subject factor (HR, LR), and SESSION (conditioning sessions 1-5, extinction sessions 1-3) as a repeated measure. Mean scores were calculated for each session of conditioning or extinction. Magazine entries. For preCS data, there was no significant main effect of GROUP, F(1, 14) = 2.23, p = .16 (Ms: HR = 1.61; LR = 1.13), and no significant interaction of GROUP x SESSION, F(7, 98) < 1. CS scores were therefore regarded as an appropriate measure of conditioning. For the CS data, a significant main effect of GROUP was found, F(1, 14) = 4.66, p < .05, revealing greater magazine approach in HR than in LR (Ms: HR = 6.15 , LR = 4.53). The interaction of GROUP x SESSION was not significant, F(7, 98) < 1. There was a significant main effect of SESSION, F(7, 98) = 23.21, p < .001, indicating the occurrence of conditioning and extinction. Magazine time. For preCS data, there was no significant main effect of GROUP, F(1, 14) < 1 (Ms: HR = 0.95; LR = 0.83), and no significant interaction of GROUP x SESSION, F(7, 98) < 1. CS scores were therefore regarded as an appropriate measure of conditioning. For the CS data, no significant main effect of GROUP, F(1, 14) = 2.18, p = .16 (Ms: HR = 3.85, LR = 2.78), and no significant interaction of GROUP x SESSION, F(7, 98) = 1.32, p = .25, were found. There was a

196

12 HR LR

Magazine entries (CS)

10 8 6 4 2 0 -2 C1 C2 C3 C4

C5

E1

E2

E3

Session

Figure 5.2.1. Appetitive Pavlovian conditioning and extinction in HR and LR, measured in magazine entries during CS presentations (±SEM). Session means are presented. C = conditioning session, E = extinction session.

197

Magazine time (s) during CS

*

5

4

3

2

1

0 HR

LR

Group

Figure 5.2.4. Difference in mean extinction scores (collapsed over 18 trials), between HR and LR, measured in magazine time (s) during CS presentations (±SEM). * significant, p < .04

198

significant main effect of SESSION, F(7, 98) = 20.54, p < .001, indicating changes of conditioned responding as conditioning and extinction proceeded. Magazine time. Extinction alone. The main effect comparison of GROUP was not far from a lenient alpha-level of .1, and such a main effect was found in the magazine entry data (p < .05). Furthermore, there may have been a ceiling effect at the peak of conditioned responding that masked possible differences. Hence, the extinction data was analysed again separately. All extinction trials were collapsed into a single mean extinction score and a two-sample t-test was performed on these data. This confirmed a significant difference between HR and LR during extinction, p < .04 (Ms: HR = 3.96, LR = 2.07).

Discussion In this experiment, two subgroups of a population of rats, HR and LR, were compared in an appetitive Pavlovian conditioning paradigm. It was hypothesized that HR would show greater levels of conditioned responding and thus learning. HR have previously been shown to have a greater HPA response, i.e. greater and prolonged release of glucocorticoids in response to stress with no difference in basal levels (Piazza et al., 1991a), and glucocorticoids are also released in response to “positive stressors” such as food reward (Caggiula et al., 1991; Fuller & Snoddy, 1981; Honma et al., 1984; Merali et al., 1998; Orchinik et al., 1988; Piazza & Le Moal, 1997). Since glucocorticoids can enhance memory consolidation in a retrograde fashion (Chapter 2), different HPA-reactivity levels between HR and LR may result in differences in learning. A difference between the two groups was indeed found. HR displayed greater levels of conditioned responding, i.e. responding was greater during the CS whereas there was no difference in baseline responding. This difference was observed

199

using either magazine entries or magazine time as dependent variable. In the latter case, the difference was restricted to the extinction period. It might be possible that a ceiling effect masked differences between the groups during conditioning which only became apparent during extinction. One might argue that HR just show slower extinction, in the sense that they learn the new CS–nothing contingency slower than LR. Following this logic, one might assume that the omission of the expected reward constitutes a stressor, in which case HR should actually show greater levels of extinction (Port et al., 1998). However, Chapter 3 failed to find support for glucocorticoids modulating extinction learning, and it seems more likely that greater levels of conditioned responding during extinction in HR constitute greater savings from conditioning and hence reflect better levels of initial learning. One could also argue that HR show better levels of performance or retrieval rather than learning. This however seems unlikely, as glucocorticoids are thought to have impairing effects on memory retrieval (e.g. de Quervain et al., 1998), and there are no theoretical reasons to believe that greater HPA-reactivity in HR enhanced performance, i.e. conditioned responding, during extinction trials. The distinction between memory retrieval and consolidation in such a paradigm is, of course, an indistinct one, as they co-occur in many situations. Finally, it should be noted that the observed differences cannot be due to differences in baseline activity, as there were no group differences in the preCS data. Given that learning was superior in HR, the results are still in line with Piazza et al’s theory. Piazza and colleagues argue that glucocorticoids play a sensitising role, making appetitive rewards such amphetamine, and possibly food as well, more reinforcing, by enhancing phasic mesolimbic DA activity. Rather than being a result of greater memory modulatory influences, enhanced learning in HR may have arisen

200

because the reinforcer, i.e. the food pellet, is experienced as more salient. In the Rescorla-Wagner model (Rescorla & Wagner, 1972), this would reflect an increase of ß or ?, and also predict a faster acquisition or higher amplitude of learning. Hence, the present experiment cannot rule this interpretation out. An explicit test that would differentiate between a memory modulation and a sensitisation account of enhanced learning would be an aversive Pavlovian conditioning paradigm. Here, the sensitisation account (as it appeals only to effects on appetitive rewards) would predict lower levels of conditioned responding, whereas the memory modulation account would predict higher ones. Such an experiment is described in the next section (5.3).

5.3 HR VS. LR IN AVERSIVE DISCRETE-CUE PAVLOVIAN CONDITIONING

Whereas Experiment 5.2 compared HR and LR in appetitive Pavlovian conditioning, this study constitutes the aversive counterpart, employing a conditionedsuppression fear-learning paradigm. The difference between HR and LR in appetitive conditioning could be interpreted to be due to differences in memory consolidation, in line with the memory modulation literature (e.g. Roozendaal, 2000; this thesis), or due to differences in magnitude of reinforcement experience, in line with claims that glucocorticoids boost the reinforcing properties of a reinforcer and possess reinforcing properties themselves (Piazza & Le Moal, 1997). Both accounts predict increased conditioned responding in HR. However, in an aversive paradigm predictions would differ. If glucocorticoids, which are thought to be released to a greater extent after

201

stress in HR, have rewarding properties (Piazza & Le Moal, 1997), one would expect that HR show lower levels of conditioned responding in an aversive learning procedure. The reinforcing properties of glucocorticoids would compete with the aversive properties of the footshock US and potentially reduce the salience of the aversive experience. According to the memory modulation account, however, conditioned responding should be greater in HR, in line with the results of exogenous post-training administration of glucocorticoids in aversive conditioning (Chapter 2). Therefore, the current study not only compares HR and LR in aversive Pavlovian conditioning, but also disambiguates the findings of Experiment 5.2.

Method Subjects Refer to section 5.1, method and results/discussion. All experiments were carried out between 14:00 and 20:00. Apparatus Refer to Experiment 2.2. A retractable lever was located to right of the magazine. Food pellets (45 mg, Rodent Grain-Base Formula; BIO-SERV, USA) provided positive reinforcement. Three lights, one on each side of the magazine and one on the centre of the ceiling, provided a discrete visual stimulus in each chamber. The background noise of the chamber was 55 dB. Footshock (0.5 mA, 0.5 s) served as the aversive US. Behavioural procedure Lever press training. Over the course of three days, all animals received one FI-20 and two VI-30 lever training sessions, during which pressing on the left lever led to food pellet reinforcement. A VI-60 schedule of reinforcement was imposed on

202

all further sessions, and maintained throughout the remainder of the experiment. On the day following conditioning, all rats received a VI-60 reminder session. Habituation. On each of the next three days, rats received a 20-min habituation session in which two, non-reinforced, 30-s presentations of the light occurred after 7 and 17 min. The session served to eliminate unconditioned suppression of lever pressing during the presentation of the light. Conditioning. On the next day, rats received a 10-min conditioning session in which a 30-s light was presented after 7 min. The final 0.5 s of the light coincided with the delivery of a 0.5-s foot shock, delivered through the grid floor of the chamber. Extinction. Two days after conditioning, rats received one 20-min extinction session. Non-reinforced presentations of a 30-s light were given after 7 and 17 min. Lever pressing rates were recorded during the 30 s prior to the light presentations (preCS period) and during the 30-s light presentations (CS period) in all sessions. A suppression ratio for each trial was calculated by dividing the number of LP during the CS period by the total number of LP during the preCS and CS periods. Table 5.3.1 displays an outline of the experimental design. Statistical analysis Refer to Experiment 2.1.

203

Table 5.3.1. Experimental design of Experiment 5.3 (HR vs. LR in fear conditioning). HR = high responders; LR = low responders; MT = magazine training, LT = lever training; VI = variable interval schedule; LL = left lever.

PHASE

GROUP HR LR

Magazine- & Lever Training Pre-experiment Day 5 MT: VT-30 LT: VI-2, 15, 30, 60 LL ? food

Habituation

Condiioning

Extinction

Day 1-3

Day 4

Day 6

2x light ? 0

1x light ? shock

2x light ? 0

Results Figure 5.3.1 illustrates mean levels of unconditioned suppression in HR and LR in the habituation and conditioning sessions. There was no difference between the groups. Figure 5.3.2 shows levels of conditioned suppression of lever pressing in the two groups on the first extinction trial where HR displayed significantly greater conditioned suppression than LR. PreCS scores. There was no difference between the two treatment groups in baseline lever pressing during the preCS period. A two-way, 2 x 9, mixed ANOVA with the between-group factor GROUP (HR, LR) and the repeated measure TRIAL (preCS period of all 9 trials) revealed that the groups did not differ, F(1, 14) < 1, (Ms (LP/30 s): HR = 6.08; LR = 6.53). Suppression ratios were therefore considered an appropriate measure to assess conditioned suppression of lever pressing and constitute the dependent variable of further analyses. In one case, in a single data cell where the absence of any lever pressing during either preCS and CS led to a division-by-zero error, the suppression ratio was calculated on the basis of the session’s overall response rate.

204

0.5

Suppression Ratio

0.4

0.3

0.2

0.1

0.0 HR

LR

Group

Figure 5.3.1. Unconditioned suppression of lever pressing, totalled across habituation and single conditioning trial, in HR and LR.

205

0.5

*

Suppression Ratio

0.4

0.3

0.2

0.1

0.0 HR

LR

Group

Figure 5.3.2. Conditioned suppression of lever pressing in HR and LR on extinction trial 1. * statistically significant, p < .05.

206

Habituation & Conditioning. Groups did not differ during habituation and conditioning (see Figure 5.3.1). A two-way, 2 x 7, mixed ANOVA with one between groups factor, GROUP (HR, LR) and one repeated measure, TRIAL (habituation trials 1-6, and the single conditioning trial) found no main effect of GROUP, F(1, 14) < 1, (Ms: HR = 0.24, LR = 0.27), nor an interaction GROUP x TRIAL, F(6, 84) = 1.38, p = .23. There was a significant main effect of TRIAL, F(6, 84) = 5.49, p < .001, reflecting a reduction of unconditioned suppression over trials. Newman-Keuls posthoc analysis revealed a higher suppression ratio on the conditioning (M = 0.30) than on the first habituation trial (M = 0.07), p < .02. Extinction. The extinction trials were analysed together with the last habituation trial and the conditioning trial. Extinction trials constituted the postlearning phase, whereas the habituation and conditioning trials belonged to the prelearning phase. A three-way, 2 x 2 x 2 mixed ANOVA with one between groups factor, GROUP (HR, LR) and two repeated measure, PHASE (pre-learning, postlearning) and TRIAL (trials 1 and 2), did not reveal a significant main effect of GROUP or TRIAL, both F(1, 13) < 1. A significant main effect of PHASE, F(1, 14) = 18.61, p < .001, indicated greater suppression post-learning (Ms: pre-learning = 0.37 , post-learning = 0.17), indicating conditioning. The interaction of PHASE x TRIAL, F(1, 14) = 9.19, p < .01, reflected post-learning extinction. The interactions of GROUP x PHASE, F(1, 14) = 1.08, p = .32, and GROUP x TRIAL, F(1, 14) < 1, were not significant. However, there was a significant three-way interaction of GROUP x PHASE x TREATMENT, F(1, 14) = 8.39, p < .02. Subsequent simplesimple-effect analysis revealed that HR and LR did not differ on either pre-learning trial nor on the second extinction trial, all p = .17. However, there was a significant difference on the first extinction trial, F(1,14) = 4.75, p < .05, with HR showing

207

greater conditioned suppression reflected by lower suppression ratios than LR (Ms: HR = 0.04, LR = 0.23).

Discussion In this study, the two subgroups of a population of rats, HR and LR, were compared in an aversive Pavlovian conditioning paradigm, employing a conditioned suppression procedure. In Experiment 5.2, it was found that HR showed more rapid learning than LR in an appetitive Pavlovian conditioning experiment. It was hypothesized that HR would show better learning and greater conditioned responding in an aversive Pavlovian conditioning experiment as well. This was, indeed, found. While the groups did not differ in the degree of unconditioned suppression, i.e. during presentations of the lights before it was paired with footshock, there was a difference during extinction. HR showed greater suppression than LR during the first trial. However, the difference was no longer apparent in the second extinction trial. Since only one light-footshock pairing was given, conditioned suppression extinguished quite rapidly so that the procedure may no longer have been sensitive to group differences following the first trial. The fact that HR showed greater conditioned suppression than LR on trial 1 of extinction, likely to reflect better learning of the aversive contingency, has important implications. In line with the memory modulation literature (e.g. Roozendaal, 2000; this thesis), HR, which have a greater glucocorticoid release in response to stress, show better levels of learning. The results are reminiscent of the reported effects of exogenous glucocorticoid administration in Chapter 2, except that the manipulation in this experiment was achieved endogenously, by having HR and LR mimic DEX and SAL administration. The results are in contrast to predictions of the model of

208

Piazza and colleagues who proposed that glucocorticoids themselves have reinforcing properties via modulation of the mesolimbic DA system. While the results of the appetitive procedure (Experiment 5.2) could be explained by both accounts, the aversive procedure in this experiment allows a dissociation of the two. According to Piazza and Le Moal (1997), “...the activation of biological substrates of reward [glucocorticoids] by threatening situations, could play the role of counteracting threatmediated aversion, ensuring a better coping and adaptation of the individual to external aggressions.” (p. 367). Hence, one might expect that HR would show lower levels of conditioned suppression, as the aversive properties of the footshock would be compromised by the appetitive properties of glucocorticoids. Further, greater baseline levels of locomotor activity in HR, according to which groups were defined in the first place, would also have predicted lower levels of conditioned suppression in HR. That this was not the case suggests that the results of Experiment 5.2 were likely due to differences in memory consolidation, rather than to an enhancement of the reinforcing properties of the food US. Similarly, the greater levels of amphetamine self-administration reported in HR (Piazza et al., 1990) may, at least to some degree, be due to the same reasons. HR may learn the conditional discrimination that allows self-administration better, rather than just merely finding the amphetamine more reinforcing. Furthermore, amphetamine itself, like glucocorticoids, has been shown to act as a retrograde memory modulator (McGaugh, 1966).

209

5.4 HR VS. LR IN INSTRUMENTAL LEARNING

Experiments 5.2 and 5.3 showed some evidence of better learning in HR than LR in an appetitive Pavlovian conditioning paradigm. In Experiment 5.4, these two groups were compared in an appetitive instrumental procedure in which lever pressing led to a sucrose reward. In addition to looking at instrumental acquisition rates, the present study also investigated shifts in the nature of associations governing instrumental performance. In order to do so, the experiment included a reversible reinforcer-specific devaluation treatment that was applied repeatedly, after different levels of instrumental training. Instrumental behaviour is thought to be based on two types of associations: action-outcome (A-O) associations allow an organism to encode the specific consequences of their actions, whereas stimulus-response (S-R) associations do not contain any specific information about the reinforcer. According to some researchers, during the course of training, the relative contribution of these associations to behaviour are thought to shift from predominantly A-O responding to S-R habits (Adams & Dickinson, 1981; Adams, 1982; Dickinson, 1985). Whereas AO responding is goal-directed and sensitive to devaluation of the reinforcer, S-R habits are not (see general discussion). It has been argued that drug addiction can be understood as the product of aberrant Pavlovian and instrumental learning processes, and more specifically, although not yet experimentally tested, that drug-seeking behaviour is initiated under the control of an A-O process but becomes an S-R process-governed compulsive habit, as addiction sets in (Everitt, Dickinson, Robbins, 2001). In the light of Piazza and colleagues’ proposal that HR are prone to develop drug addiction, comparing HR

210

and LR in a paradigm investigating the rate of formation of habits, seemed of particular interest. Experiment 5.4 was conducted prior to (fear-conditioning) Experiment 5.3. The altered order of presentation was chosen for clarity. However, it is important to note that animals had not yet experienced aversive experimental procedures at the time of this instrumental learning study.

Method Subjects See 5.1, method and results/discussion. In one analysis, the top and bottom 10 rather than 8 responders were utilised, in order to increase statistical power. All experiments were carried out between 14:00 and 20:00.

Apparatus Twelve identical operant chambers as described in Experiment 2.3 were used in the experiment. Each chamber contained a central, recessed magazine that provided access to sucrose reinforcement (20% solution) delivered by a liquid dipper. A retractable lever was located on the left side of the magazine. The background noise in the chamber was 60 dB . Pre-feeding took place with either sucrose (20% solution) or food pellets (45 mg, Rodent Grain-Base Formula; BIO-SERV, USA). Behavioural procedure Rats received sessions of lever training interspersed with test sessions carried out in extinction. They received one session per day. Magazine and lever training. On the first day, rats were given a 20-min session of magazine training during which sucrose was delivered according to a VT-30 schedule. In the next session, the lever on

211

the left of the magazine was introduced into the chamber and pressing was reinforced with sucrose under a fixed interval (FI)-20 schedule (i.e. reinforcers were made available following the first response after intervals of 20 s). In all further sessions of lever training, a VI-30 schedule of reinforcement was applied. Lever training sessions lasted up to 20 min but were terminated once the maximum number of reinforcers, 31, was reached. Therefore each animal received equal exposure to the reinforcers across training. Animals received one VI-30 session before the first test phase, a further five before the second test phase, and a further ten before the final test phase. Test: Three test phases took place at different stages of the experiment, following a low (initial sessions of FI-20 and VI-30), medium (a further five VI-30 sessions) and high (a further ten VI-30 sessions) level of lever training. Each test phase took place over three days and comprised two actual test sessions, on the first and third day, and a VI30 reminder session of lever training in between. Test sessions were 15 min long, and animals were given the opportunity to press the lever and to enter the magazine, but no reinforcement occurred, i.e. tests were in extinction. Within each test phase, the two sessions were preceded by a 30-min pre-feeding session during which animals were given free access to either the same reinforcer as that earned by lever training, sucrose (devalution), or a different reinforcer, food pellets (non-devaluation). The order in which these two test sessions occurred was counterbalanced across animals and test phases. Following test sessions in the first test phase, animals were given a choice-consumption task with 15-min free access to both food pellets and sucrose, in order to determine if the pre-fed reinforcer was devalued, i.e. if rats consumed less of it than of the non-pre-fed reinforcer. See Table 5.4.1 for an outline of the experimental design.

212

Statistical analysis Refer to Experiment 2.1. Table 5.4.1. Experimental design of Experiment 5.4 (HR vs. LR in appetitive instrumental learning). HR = high responders; LR = low responders; MT = magazine training; LT = lever training; FI = fixed interval schedule; VI = variable interval schedule. PHASE

Test I Training I

GROUP

(low level of training)

Test II

(medium Training II level of training)

devaluation & test

HR LR

MT LT FI-20 LT VI-30

reminder LT

Test III Training III

devaluation & test

5 x LT VI-30

non-devaluation & test

reminder LT


(high level of training) devaluation & test

10 x LT VI-30

reminder LT


Results Lever training: Figure 5.4.1 illustrates lever pressing (LP/min) and magazine entry (ME/min) rates in HR and LR across the first, FI-20, lever training session. Magazine entry rates were greater in HR than in LR during this session. Figure 5.4.2 compares HR and LR across lever training sessions, with respect to LP/min, ME/min and reinforcement(RNF)/min rates respectively. Groups did not differ from each other. FI-20 session: A minute-by-minute response analysis for the first 14 min (during which none of the animals had finished the session) of the first, FI-20, lever training session was performed for both lever pressing and magazine entry, utilising a two-way, 2 x 14, mixed ANOVA, with GROUP (HR, LR) as the between-subject factor and MINUTE (1-14) as a repeated measure. Lever pressing: There was no

213

14

(A)

Lever presses / min

12 10 8 6 4 2 0

(B)

Magazine entries / min

18 16 14 12 10 8 6 HR LR

4 2 0 1

2

3

4

5

6

7

8

9

10

11

12

13

14

minute

Figure 5.4.1 Minute-by-minute illustration of mean (±SEM) LP/min (panel A) and ME/min (panel B) rates in HR and LR in the first, FI-20, lever training session.

214

18

(A)

Lever presses / min

16 14 12 10 8 6 4

HR L

2

R

0

(B)

Magazine entries / min

14 12 10 8 6 4 2

(C)

Reinforcements / min

0

2.0

1.5

1.0

0.5

low-

medium-

high levels of training

0.0 1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

Training session Figure 5.4.2. HR and LR are compared in their mean (±SEM) LP/min (panel A), ME/min (panel B), and RNF/min (panel C) rates across VI-30 lever training sessions, i.e. with a low, medium and high levels of training.

215

significant main effect of GROUP, F(1, 14) = 1.71, p = .21, (Ms: HR = 6.39; LR = 4.90), nor a significant interaction with MINUTE, F(13, 182) = 1.05, p = .44. A significant main effect of MINUTE, F(13, 182) = 8.30, p < .001, reflected an increase of lever responses across the session. Magazine entries. There was a significant main effect of GROUP, F(1, 14) = 5.09, p < .05, with HR (M = 15.54) approaching the magazine more than LR (M = 12.53) (see Figure 5.4.1). This effect may be due to differences in Experiment 5.2, however, which took place before. There was no significant effect of MINUTE, F(13, 182) < 1, nor an interaction of GROUP x MINUTE, F(13, 182) < 1. VI-30 sessions. For each dependent variable (LP/min, ME/min and RNF/min rates), a two-way, 2 x 16, mixed ANOVA was carried out, with GROUP (HR, LR) as the between-subject factor and SESSION (1-16) as a repeated measure. Lever pressing: There was no main effect of GROUP, F(1, 14) < 1 (Ms: HR = 13.18 , LR = 12.23), and no significant interaction of GROUP x SESSION, F(15, 210) < 1. A significant main effect of SESSION, F(15, 210) = 5.53, p < .001, reflected a general increase in lever pressing rates as training progressed. Newman Keuls posthocanalysis revealed, for example, that responding in the last training session was significantly greater than in the first four sessions, all ps < .04 (see Figure 5.4.2, panel A). In a further analysis, only those training sessions that directly preceded a test phase were utilised as levels of the factor SESSION in a two-way, 2 x 3, mixed ANOVA. Again, no effect of GROUP, F(1, 14) < 1, nor a GROUP x SESSION interaction, F(2, 28) < 1, was found. Magazine entries: There was no main effect of GROUP, F(1, 14) < 1 (Ms: HR = 10.07 , LR = 9.80), and no significant interaction of GROUP x SESSION, F(15, 210) < 1. A significant main effect of SESSION, F(15, 210) = 4.78, p < .001, reflected a general decrease in magazine entries rates as

216

training progressed. Newman Keuls posthoc-analysis revealed, for example, that responding in the first training session was significantly greater than in sessions 7-16 onwards, all ps = .05 (see Figure 5.4.2, panel B). Reinforcement rate. With a VI30 schedule, RNF-rates are limited to an average of 2 reinforcers per minute. There was no main effect of GROUP, F(1, 14) < 1 (Ms: HR = 1.85, LR = 1.86), and no significant interaction GROUP x SESSION, F(15, 210) = 1.30, p = .20. There was a significant main effect of SESSION, F(15, 210) = 2.61, p < .01. Newman Keuls posthoc-analysis revealed that reinforcement rates in the first training session were lower than in all other sessions, except 9, 13 and 14, all ps < .03. Hence, rats reached ceiling after session 1 (see Figure 5.4.2, panel C).

Test phases. Figure 5.4.3 shows lever pressing rates in HR and LR, following reinforcer devaluation and non-devaluation, with a low, medium and high level of training. A three-way, 2 x 2 x 3, mixed ANOVA with GROUP (HR, LR) as the between-subject factor, and DEVALUATION (devalued, non-devalued) and TRAINING (low, medium, high) as repeated measures, was carried out. There was no significant main effect of GROUP, F(1, 14) < 1, (Ms: HR = 2.39, LR = 2.14). The main effect of DEVALUATION was significant, F(1, 14) = 4.77, p < .05, with lower responding in the devalued (M = 2.03) than in the non-devalued (M = 2.49) condition. A main effect of TRAINING, F(2, 28) = 31.64, p < .001 indicated a general reduction of responding (Ms: low = 2.93; medium = 2.32; high = 1.54). The interaction of DEVALUATION x TRAINING was also significant, F(2, 28) = 3.98, p < .04, whereas the interactions of GROUP x DEVALUATION and GROUP x TRAINING

217

non-devalued devalued

Lever presses / min.

4 3 2 1 0

HR

LR

HR

LR

HR

LR

Test I

Test II

Test III


(medium level of training)

(high level of training)

Figure 5.4.3. Lever pressing during test sessions following different levels of training in HR and LR.

218

were not, both Fs (2, 28) < 1. However, there was a marginally significant three-way interaction of GROUP x DEVALUATION x TRAINING, F(3, 28) = 2.97, p = .07.

Expansion of groups to top and bottom 10 responders. Since there was a marginally significant three-way interaction in the analysis above, it was decided to raise statistical power by increasing the number of subjects to n = 10/group, comprising of the top and bottom 10 rather than 8 responders. Figure 5.4.4 shows lever pressing rates in HR and LR, following reinforcer devaluation and nondevaluation, after low, medium and high levels of training. The same analysis as previously was carried out, employing a three-way, 2 x 2 x 3, mixed ANOVA with GROUP (HR, LR) as the between-subject factor, and DEVALUATION (devalued, non-devalued) and TRAINING (low, medium, high) as repeated measures. The dependent variable, LP/min, was made subject to a square-root transformation, in order to normalise the distribution. There was no significant main effect of GROUP, F(1, 18) < 1, (Ms: HR = 1.50, LR = 1.43). The main effect of DEVALUATION was marginally significant, F(1, 18) = 4.19, p = .06. There was a trend for lower responding in the devalued (M = 1.40) than in the non-devalued (M = 1.52) condition. A main effect of TRAINING, F(2, 36) = 41.62, p < .001 indicated a general reduction of responding (Ms: low = 1.67; medium = 1.52; high = 1.20). The interaction of DEVALUATION x TRAINING was also significant, F(2, 36) = 3.31, p < .05, whereas the interactions of GROUP x DEVALUATION and GROUP x TRAINING were not, both Fs (2, 36) < 1. However, the three-way interaction of GROUP x DEVALUATION x TRAINING was significant, F(3, 28) = 3.23, p = .05. In order to break down this interaction, individual analyses were carried out for each level of TRAINING, in form of two-way, 2 x 2, mixed ANOVA with GROUP (HR, LR) as

219

non-devalued devalued

Lever presses / min

4 3 2 1 0 HR

LR

HR

LR

HR

LR

Test I

Test II

Test III


(medium level of training)

(high level of training)

Figure 5.4.4. Lever pressing during test sessions following different levels of training in HR and LR (n = 10/group).

220

the between-subject factor and DEVALUATION as a repeated measure.

Low

level

of

training:

There

was

a

significant

main

effect

of

DEVALUATION, F(1, 18) = 8.02, p = .01, with lower responding in the devalued (M = 1.55) than in the non-devalued condition (M = 1.79). The main effect of GROUP, F(1, 18) < 1, (Ms: HR = 1.67, LR = 1.67), and the interaction of GROUP x DEVALUATION, F(1,18) = 1.84, p = .19, were not significant. This indicates that both groups were equally sensitive to the devaluation with low levels of training. Medium level of training: There was no significant main effect of GROUP, F(1, 18) < 1, (Ms: HR = 1.55, LR = 1.48). The main effect of DEVALUATION, F(1, 18) = 3.62, p = .07, approached significance with marginally lower responding in the devalued (M = 1.44) than in the non-devalued condition (M = 1.59). There was a significant interaction of GROUP x DEVALUATION, F(1,18) = 4.43, p < .05. Simple effect analysis revealed less responding in the devalued than in the nondevalued condition in LR, p = .01, (Ms: non-devalued = 1.63, devalued = 1.32), but not in HR, p = .89, (Ms: non-devalued = 1.54, devalued = 1.56), indicating that with medium levels of training only LR are sensitive to the devaluation. High level of training: There were no significant main effects of GROUP, F(1, 18) = 1.45, p = .25, (Ms: HR = 1.55, LR = 1.48), or DEVALUATION, F(1, 18) < 1, (Ms: non-devalued = 1.18, devalued = 1.22). The interaction of GROUP x DEVALUATION, F(1, 18) < 1, was also not significant. These results suggest that neither group is sensitive to devaluation with high levels of training.

Consumption test: Figure 5.4.5 displays the results of the consumption test in both HR and LR. Reinforcer consumption in the consumption test was differentially

221

pre-fed non-pre-fed

Consumption (g)

6 5

*

*

HR

LR

4 3 2 1 0

Group

Figure 5.4.5. Consumption of pre-fed and non-pre-fed reinforcers in HR and LR in the consumption test. * statistically significant, p = .01

222

calculated for the reinforcer that was pre-fed (SAME) and for the one that was not (DIFF), independent of the actual reinforcer types. A two-way, 2 x 2, mixed ANOVA was carried out with GROUP (HR, LR) as the between-subject factor and PREFEEDING (SAME, DIFF) as a repeated measure. There was a significant main effect of PRE-FEEDING, F(1, 14) = 23.58, p < .001, with non-pre-fed reinforcers being consumed more (M = 4.27g) than pre-fed reinforcers (M = 0.95g). This confirms that the devaluation procedure was successful. There was no significant main effect of GROUP, F(1, 14) < 1, and no significant interaction GROUP x DEVALUATION, F(1, 14) < 1. This demonstrates that devaluation occurred equally in both groups, irrespective of the effect of devaluation on lever pressing performance.

Discussion Experiment 5.4 investigated the course of extended instrumental training in HR and LR, and if these groups differed in their sensitivity to reinforcer-specific devaluation procedures. At the start of lever training, during the FI-20 session, there were no differences between groups on lever pressing. However, greater magazine approach in HR perhaps indicated a more rapidly acquired Pavlovian association between food and magazine in these animals, resulting from the magazine training session, and supporting the results of Experiment 5.2. In fact, given that the current experiment was directly preceded by the appetitive Pavlovian conditioning study, enhanced levels of magazine entry in HR might have been a direct result from that experiment (5.2). There were no differences between groups in lever pressing, magazine approach and reinforcement rates across the 16 VI-30 sessions, suggesting equivalent rates of instrumental acquisition between groups. Reinforcer exposure was kept constant for all animals across all sessions.

223

The more important results of the study were obtained in the extinction test sessions that followed devaluation and non-devaluation pre-feeding procedures. Reinforcer-specific satiety devalues the reinforcer that was pre-fed but not other reinforcers (in fact, as compared to a no pre-feeding condition, it may devalue other reinforcers to some extent, but it is the difference in degree of devaluation that is critical in this context). In this study, where rats had learnt to press a lever for a sucrose reward, pre-feeding with sucrose, but not with pellets, resulted in devaluation. The consumption test after the first test-period confirmed that devaluation was successful in both groups of animals. As long as instrumental responding is, at least to some extent, governed by goal-directed A-O processes, it should be sensitive to a prefeeding procedure, and lever pressing rates in the devalued condition should be lower than in the non-devalued one. However, as instrumental responding gradually becomes more S-R driven with extended training, sensitivity to the pre-feeding procedure should decline, eventually resulting in equal response rates in the devalued and non-devalued condition. With a low level of training (after one VI-30 session), rats in both HR and LR groups displayed a goal-directed pattern, reflecting sensitivity to devaluation and suggesting instrumental behaviour was governed by A-O associations. However, with a medium level of lever training, i.e. after a further five sessions, only LR displayed sensitivity to devaluation. HR in contrast, seemed to demonstrate behaviour governed by S-R associations. Finally, with a high level of lever training following a further ten sessions of training, both groups failed to show sensitivity to the devaluation procedure, responding equally in both conditions. The results suggest that HR might shift to habit responding more readily than LR. This is interesting, given that HR have been suggested to be more prone to drug addiction (e.g. Piazza et al., 1990). The current study supports these claims, but for reasons

224

other than those previously advocated. As Everitt et al. (2001) have pointed out, drugseeking, like food-seeking, may, after extended training at the onset of addiction, become a compulsive habit governed by S-R processes. As HR seem to form habits more readily than LR, it would indeed make them more prone to develop drug addiction but perhaps because of a change in mechanisms controlling the shift from A-O to S-R rather than (or in addition to) changes in the reinforcing impact of rewards and drugs of abuse. One might argue that the reason animals lose their sensitivity to the devaluation procedure is not because they form S-R habits, but rather that it stems from differences in outcome (US) processing. Lever pressing might become associated with the sensory and the motivational properties of the US. Pre-feeding of either reinforcer would devalue the motivational properties of the US to the same extent. However, the sensory properties, making sucrose distinct from food pellets, are selectively devalued in the current reinforcer specific devaluation procedure. If an animal then does not respond differentially in the two pre-feeding conditions, this may be due to the sensory properties of the US (or outcome) being overshadowed by its motivational properties. With extended training, rats might predominantly form associations between responses (lever pressing) and the motivational properties of the food reward outcome, at the cost of its sensory properties. Given the results of Chapter 4, where glucocorticoids were theorised to exert such an effect, based on the results of the Pavlovian-Instrumental transfer studies, this explanation would also fit quite well. The more stress-reactive HR might secrete greater glucocorticoid levels during instrumental lever training, and, as a consequence, undergo the shift towards forming associations predominantly with motivational properties of the outcome more readily than their LR counterparts. Behaviourally, this would be reflected by a lack of

225

sensitivity to a reinforcer specific devaluation procedure, just as observed in this study (seen General Discussion; Chapter 6). Regardless of possible mechanisms, the results reported here suggest a differential pattern of the development of control of instrumental behaviour in HR and LR. HR either form habits more readily, or they more readily undergo an overshadowing process in which associations between response and outcome become increasingly controlled by the motivational properties of the reward. Given the greater glucocorticoid response of HR, the results are in line with those of Chapter 4 where exogenous glucocorticoid administration led to an enhancement of motivational and a decline of sensory processing of the reinforcer. In instrumental behaviour, such a shift towards exclusively encoding the motivational outcome properties may well constitute an intermediate state during the development from goal-directed A-O responding to pure S-R habit responding.


While the previous experiments investigated the effects of post-training administrations of glucocorticoids on a number of learning tasks, the aim of this chapter was to examine the effects of endogenous manipulation of glucocorticoid levels, and compare individuals that are thought to differ in their HPA-reactivity to stress in a number of learning procedures. Piazza and colleagues investigated the relationship between stress-reactivity, glucocorticoids, mesolimbic DA and propensity for drug addiction (for a review see Piazza & Le Moal, 1996). They divided populations of rats into HR and LR, and

226

found that HR displayed greater levels of glucocorticoids and mesolimbic DA in response to stress, and self-administered amphetamines more readily. Piazza and colleagues argued that glucocorticoids modulate reward-related behaviour by increasing the reinforcing properties of appetitive reinforcers via interactions with mesolimbic DA. Hence, HR would have a greater propensity for sensation-seeking and drug addiction. However, the effects of glucocorticoids on memory consolidation (e.g. Roozendaal, 2000; this thesis) were not taken into consideration. The aim of the experiments reported in this chapter was to bring together these strands of research. Given that HR show greater levels of glucocorticoid release in response to stress, while baseline levels remain equal, straightforward predictions can be made from the results of the previous chapters. HR should show better memory consolidation in tasks that were influenced by post-training dexamethasone, for example discrete-cue Pavlovian conditioning of both appetitive and aversive nature (Chapter 2). In these instances, memory modulatory effects of glucocorticoids may constitute an additional or alternative explanation to the sensitisation account of HR’s propensity for drug addiction. A population of 64 rats was screened for their locomotor activity to novelty, and those individuals with the highest and lowest eight (or ten) scores constituted the HR and LR groups. HR were indeed found to show more rapid learning in both appetitive and aversive conditioning paradigms. Given the hypotheses outlined above, the results provide tentative support tothose of Chapter 2, and furthermore suggest that HR may experience greater memory modulatory effects by glucocorticoids. This could explain, at least to some extent, why HR have been found to self-administer amphetamine more readily than LR. The self-administration procedure itself is

227

embedded in a discriminational learning paradigm, where rats have to distinguish between nose-pokes into a reinforced and a non-reinforced hole. HR and LR were also compared in an appetitive instrumental procedure to investigate possible differences in their acquisition rates of operant contingencies and, more importantly, in the way their sensitivity to reinforcer-specific devaluation developed across extended training. In this goal revaluation procedure, rats are prefed with a reinforcer that is either the same or different from that which they have learned to respond for. More vigorous responding in the non-devalued condition where pre-feeding occurred with a different reinforcer, has been interpreted as evidence for purposive goal-directed action (Dickinson, 1985). In other words one can assume that the rats’ behaviour is governed by A-O associations, likely to involve a full representation of the outcome’s sensory properties. However, with extended training, the difference between conditions has been found to decline and to finally vanish. One can infer from this that rats no longer have a representation of the outcome that would allow them to determine whether or not this outcome differs from the reinforcer that was pre-fed. According to one theory, a behavioural autonomy of instrumental behaviour might have developed instead, with animals experiencing a shift from A-O governed behaviour to mechanistic S-R habits (Adams & Dickinson, 1981; Adams, 1982; Dickinson, 1985). Experiment 5.4 found that HR lost their sensitivity to the devaluation procedure more readily than LR, i.e. with a medium rather than high level of training. HR might have either formed habits more readily, or they might have more readily undergone an overshadowing process so that associations between response and outcome got increasingly controlled by the motivational reinforcer properties. Since HR are thought to secrete greater levels of glucocorticoids in response to stress (Piazza et al., 1991a), and since such “stressors”

228

may also be of a appetitive nature (e.g. Merali et al., 1998; Piazza & Le Moal, 1997), the differences in HPA-reactivity may have caused this pattern of differential behaviour across training. Given that drug addiction has been linked to S-R habits (Everitt et al., 2001), these results furthermore support the idea that HR may have a higher propensity to become addicted to drugs of abuse. Further, it may be the case that in drugs of abuse, in contrast to other appetitive reinforcers such as food, the difference between sensory and motivational properties is less distinct. Hence, either account, more rapid habit formation or enhanced motivational encoding, would support a greater proneness of HR to develop drug addiction. The reason for this, however, seems at least partly to be an aberrant learning and memory system, rather than merely a greater degree of experienced reinforcement.

229

CHAPTER 6 Overall summary, limitations and conclusions

230

This thesis focused on the acute role of glucocorticoids in associative learning and memory consolidation. By conducting behavioural experiments in rats and largely by applying pharmacological post-training treatment in form of the GR agonist DEX, the aim was to shed light on a number of related questions that have not been given much consideration yet. The topic of glucocorticoid memory modulation is part of the larger, fascinating question of how emotion can influence learning. Even though an extensive literature already deals with this topic, it has not been explored very thoroughly from a standpoint of behavioural analysis. Furthermore, the picture sometimes appears confusing due to a lack of differentiation between studies on the acute and chronic effects of glucocorticoids, as well as between studies investigating memory acquisition or consolidation and studies concerned with memory retrieval. This project deals exclusively with the acute effects of glucocorticoids on memory consolidation. Even though it has been generally accepted that acute glucocorticoid administration can enhance memory consolidation, a number of questions have remained either controversial or unexplored. In this projects some of these issues were addressed and investigated empirically. This final chapter constitutes a brief summary of the results reported in Chapters 2 to 5, embedded in a general discussion, attempting to bring these findings together. The implications deriving from the results are analysed, including both local inferences regarding the specific question under investigation, as well as more global ones. Some of these are merely speculative, for example ideas regarding the possible clinical relevance of some the findings or views from the standpoint of evolutionary psychology. A repetition of the detailed general discussions is avoided however, and the reader is referred to the individual chapters. Following this, the limitations of the project are addressed, pointing out some possible future lines of research that may

231

overcome these limitations. The chapter ends with some concluding remarks about the findings as a whole. This is accompanied by the proposal of an extended overall model of glucocorticoids in associative learning and memory consolidation.

6.1 OVERALL SUMMARY

6.1.1 Discrete-cue vs. context learning modulation The first empirical chapter of this thesis dealt with two major questions regarding the generality of glucocorticoid memory modulation; whether it can be shown in discrete-cue Pavlovian conditioning, and whether it also applies to appetitive procedures. As mentioned above, it has been generally accepted, mainly from animal studies, that glucocorticoids at the right dose exert an enhancing effect on memory consolidation. This has been shown in a number of experiments in which glucocorticoids were administered either prior to, or shortly after, the learning experience, and retention was tested at a later time. Post-training administrations constitutes the more elegant technique, as sensory-perceptual, attentional or motivational factors can be ruled out. Any effects would have to be on consolidation processes. However, it should be noted that the range of behavioural paradigms that have been used to demonstrate glucocorticoid memory modulation is rather narrow. Most studies have employed either a passive (inhibitory) avoidance procedure or a spatial water-maze task. In contrast, glucocorticoid-induced memory enhancement had not been demonstrated in Pavlovian (classical) discrete-cue conditioning. Instead, it has been suggested that the role of glucocorticoids in fear conditioning might be selective to the formation and consolidation of context representations (Pugh et al.,

232

1997a,b). These authors found a memory modulatory effect on contextual but not auditory discrete-cue conditioning. However, other studies have hinted at the possibility that learning about discrete cues may be affected by glucocorticoids (Sandi & Rose, 1994a,b; Micheau et al., 1981, 1985). Further, the proposed involvement of the amygdala in these processes (Roozendaal, 2000), a structure that is believed to modulate memory in multiple brain regions (Packard et al., 1994), does not suggest a selectivity for context-dependent learning. Therefore, a first aim of the current project was to investigate if glucocorticoids would modulate memory in a Pavlovian discrete-cue conditioning procedure. The GR agonist DEX, which has reliably shown to enhance memory in the inhibitory avoidance paradigm (e.g. Roozendaal & McGaugh, 1996a, 1997b), was administered immediately after training in a classic discrete auditory-cue fearconditioning procedure. The measures of conditioned fear were either freezing (Experiment 2.1) or suppression of an ongoing instrumental response (Experiment 2.2). In both experiments, post-training DEX enhanced memory consolidation as tested in a later retention test. As far as I am aware, these studies constitute the first demonstrations of glucocorticoid memory modulation in discrete-cue Pavlovian conditioning. Experiment 2.1 also constitutes the first demonstration of glucocorticoid memory modulation in a within-subject procedure, exposing all subjects to all treatments and all training experiences. It is not clear why Pugh et al. (1997a,b), in contrast to the results of this thesis, failed to find an influence of glucocorticoids on discrete-cue conditioning. The studies may not have been comparable to each other for a number of reasons (see Chapter 2 for a detailed discussion).

233

The current results suggest that glucocorticoids’ role in memory is not selective to context-dependent learning. From an evolutionary standpoint this appears plausible. Many events of biological significance in an organism’s environment, such as the presence of a threat, are, at least to some extent and in some occasions, signalled by discrete cues. The smell of burning trees, the distinct noise produced by an approaching predator or prey, the taste of a poisonous plant or the sight of circling vultures all constitute examples for discrete cues signalling potential emotionally significant events, typically resulting in HPA activation. Being able to predict such significant events well from stimuli in the environment would allow a quick, appropriate response, and hence appear to be of great value for survival. Along similar lines, there are psychopathological conditions such as PTSD and simple phobia, in which associations between discrete stimuli and aversive events might have become excessively strengthened and thereby maladaptive, for example the smell of petrol or the sound of a plane in some war veterans. Further indirect support for the results observed here is found in lesion- and local infusion studies. Pugh et al. (1997a,b) have argued from their results that the hippocampus is likely to constitute the key region in glucocorticoid memory modulation, since it contains a large population of GR and has generally been implicated to be involved in contextual processing (Honey & Good, 1993; Phillips & LeDoux, 1992, 1994; Selden et al., 1991). However, it has been shown by Roozendaal and colleagues that other brain structures, particularly the BLA, also play an important role in glucocorticoid memory modulation (Quirarte et al., 1997; Roozendaal & McGaugh, 1997b, Roozendaal et al., 1999a,b; Roozendaal et al., 2001a; Setlow et al., 2000). Given that the amygdala modulates learning in a variety of brain regions (Packard et al., 1994), including but not limited to the hippocampus,

234

it seems likely, and is in agreement with the current data, that glucocorticoids modulate a wide range of types of memory representations, including discrete cues, and their associations with emotionally salient events, in multiple brain regions.

6.1.2 Modulation of appetitive learning The second question raised in Chapter 2 was concerned with the affective valence of stimuli that glucocorticoids aid learning about. The great majority of animal studies that have demonstrated glucocorticoid memory modulation employed aversive learning procedure. The only exception comes from Micheau et al. (1981, 1985) who showed an enhancement of learning by post-training corticosterone in an operant successive-discrimination task. There are no reports of glucocorticoid memory modulation in appetitive Pavlovian procedures. This is surprising given that glucocorticoids are also secreted in response to rewarding situations, such as food, sex or drugs of abuse (Caggiula et al., 1991; Fuller & Snoddy, 1981; Honma et al., 1984; Merali et al., 1998; Orchinik et al., 1988; Piazza & Le Moal, 1997). Hence, a further aim of this project was to investigate if glucocorticoid memory modulation could be shown in appetitive designs. DEX was administered immediately after training in appetitive discrete-cue Pavlovian conditioning procedures, in which auditory stimuli signalled the occurrence of a food US. Like in the aversive studies, post-training glucocorticoids were found to enhance memory. As far as I am aware, the results constitute the first demonstration of glucocorticoid memory modulation in appetitive Pavlovian procedures. This has a number of interesting implications. First, if the effects of post-training glucocorticoids were limited to aversive conditioning, one might argue that the administration of DEX could be aversive in itself, adding to the magnitude of the reinforcer’s aversive properties and making it more salient. Hence,

235

the memory enhancing effect might not have been due to effects on the consolidation process. The fact that glucocorticoids modulate appetitive learning allows to rule this possibility out. Second, the results are in line with the fact that glucocorticoids are also released in response to pleasant stimuli (see above). Again this appears plausible from an evolutionary perspective. Strengthened associations between stimuli in the environment and pleasant events such as the occurrence of food might be of great survival value. Learning about appetitive and aversive motivationally significant stimuli in the environment may therefore be served by a common system of modulation by glucocorticoids. Such a general motivational system has been implicated by Konorski (1967). Support comes from human studies suggesting that enhanced memory for emotional material is mediated primarily by general arousal rather than the valence of the material (Bradley et al., 1992). Finally, as the procedures involved discrete cues, the results give further support to the claim (see above) that the role of glucocorticoids in memory modulation is not limited to contextual learning. Indirectly, the findings may be of interest to the study of amygdala function. The amygdala has been suggested as a site that is important for this memory modulation (e.g. Roozendaal & McGaugh, 1996b, 1997a,b; Roozendaal, 2000). The results reported here may throw further light on the finding that complete amygdala lesions impair learning in a highly aversive, but not a mildly aversive, nor an appetitive, task (Cahill & McGaugh, 1990, but see Parkinson, Robbins, & Everitt, 2000). Glucocorticoid modulation may occur more readily following aversive rather than appetitive events, as the former may result in greater levels of arousal than the latter, perhaps reaching a necessary threshold of amygdala activation. The interaction of glucocorticoids with endogenous levels of central NA may also be of importance in

236

memory modulation (see Chapter 3). Given that the amygdala is thought to play an important role in glucocorticoid memory modulation, the current results are interesting in the light of studies investigating the role of the amygdala in emotional learning in humans. Hamann et al. (1999), measuring regional cerebral blood flow with PET, found that bilateral amygdala activity during memory encoding correlated with enhanced recognition memory for both pleasant and aversive visual stimuli, relative to neutral ones, and proposed that this enhancement was partly due to modulation of hippocampal activity by the amygdala. Previously, Cahill et al. (1996) reported a correlation between relative glucose metabolism rate in the amygdala and recall memory for aversive emotionally arousing films, but not for neutral films. More recently, Hamann and Mao (2000) found greater amygdala activation to both pleasant and unpleasant than to neutral verbal stimuli in a functional-magnetic-resonanceimaging study.

6.1.3 Glucocorticoids in non-US learning All demonstrations of glucocorticoid memory modulation involved emotional learning procedures. It is difficult to assess non-motivated learning in animals. In human subjects though, neutral learning can be assessed more straight-forwardly. Indeed, it has been found that glucocorticoid memory modulation in humans is selective for emotional material (Buchanan & Lovallo, 2001). The aim of Chapter 3 was to assess if glucocorticoids would modulate learning tasks lacking a discrete emotional event (US) in rats. Paradigms likely to constitute Pavlovian conditioning in the absence of a discrete US include LI, EXT, SPC and SOC. The degree to which these procedures are of emotional nature is debatable, but the absence of a discrete US clearly differentiates them from the aversive and appetitve discrete-cue conditioning

237

procedures employed in Chapter 2. LI is defined as the retardation of acquisition of conditioned responding to a stimulus by repeated non-reinforced preexposure to that stimulus. According to some theories, the preexposure phase of LI constitutes conditioning between a stimulus and the absence of a reinforcer (CS-nothing learning). Administration of DEX immediately after the preexposure phase was therefore regarded as a way to test if glucocorticoids would modulate memory of CS-nothing associations. The effects of post-preexposure administration of glucocorticoids had not been investigated before. No effects of DEX on LI were found, giving preliminary support for the idea that glucocorticoid memory modulation is selective for emotional material. EXT constitutes another procedure during which non-reinforced presentations of a stimulus occurs. EXT is believed by some researchers, like LI, to be a form of CS-nothing learning. In contrast to LI however, non-reinforced stimuli occur subsequent to prior reinforced CS presentations, implying that the omission of the US contains elements of frustration or relief. The effects of post-extinction administration of DEX on EXT constituted the second experiment of Chapter 3. There had been a number of previous studies investigating glucocorticoids in EXT procedures, generally reporting an enhancement of EXT (e.g. Bohus & Lissak, 1968; Kovacs et al., 1976; Micheau et al., 1982; Port et al., 1998; van Wimersma Greidanus, 1970; but see Micco et al., 1979). However, no effects of DEX were found in this project. One should note that most previous studies employed pre-extinction treatment. The only comparable study (Micheau et al., 1982) reported facilitated EXT, albeit in an appetitive operant paradigm. It is possible that EXT of appetitive operant conditioning, where the omission of appetitive reinforcement constitutes a loss of control and a so-called frustrative non-reward, may comprise a situation substantially

238

different from aversive Pavlovian EXT. This becomes more apparent when speculating about the possible adaptive functions of these processes. Port et al. (1998) suggested that enhanced EXT due to elevated levels of corticosterone may be adaptive as it allows an organism to accommodate changes in environmental contingencies more rapidly. While this sounds plausible for appetitive situations, the premature EXT of an aversive contingency might be dangerous and lower the chance of an organism’s survival. In signal detection theory terms, it may be beneficial in the appetitive domain to strive for correct rejection at the risk of a miss. In the aversive domain, however, this may prove lethal, whereas a strategy allowing for false alarms may save an organism’s life. Hence, the null results reported here might make sense from an evolutionarily point of view. Both enhanced consolidation of a discrete emotional event (Chapter 2) and the absence of premature EXT, at least in aversive scenarios, may promote survival. The other two experiments examined CS1-CS2 associations, employing SPC and SOC paradigms. In SPC, two motivationally neutral stimuli, A and X, are presented together. In the next stage, one of these stimuli (X) is conditioned to a motivationally significant reinforcer. In a final test stage, responding to A is greater than to a control stimulus. This depends on the formation of an association between A and X. The question of interest was whether glucocorticoids could affect SPC. This had not been investigated previously. A flavour aversion paradigm was employed, and flavours A and X were presented to rats as a compound in stage 1. This was followed immediately by administration of DEX or SAL. In the next stage, one of the flavours was conditioned to illness. In a final test phase, A was consumed less than control flavour B, demonstrating SPC. However, as in the CS-nothing studies, there was no effect of DEX. SPC might constitute the paradigm that comes closest to

239

assessing non-motivated learning in animals. The results are therefore in line with Buchanan and Lovallo’s (2001) report in humans that glucocorticoids do not modulate memory for non-arousing, neutral material. A second paradigm that comprises CS1-CS2 (A-X) associations is SOC. However, SOC differs from SPC in the temporal relation of the A-X stage to the conditioning of X. While in SPC this stage precedes conditioning, assuring entirely non-motivated learning, the A-X pairing follows (first-order) conditioning of X in SOC. X may therefore possess emotional characteristics when paired with A. Further, in addition to the associative chain and fantastic conditioning accounts, SOC may also occur by S-R mechanisms. So far, the effects of glucocorticoids on SOC had not been investigated. However, Nader and LeDoux (1999) dissociated the two paradigms pharmacologically before, finding that the D2 dopamine receptor agonist quinpirole blocked the acquisition of SOC but not SPC. Despite the null results in LI, EXT and SPC, one might have argued that the differences to SPC, such as potential acquired emotional characteristics of X, would possibly allow for glucocorticoid memory modulation to occur. DEX was administered immediately following the second-order stage. While SOC was demonstrated successfully, again no effect of DEX was found. In a series of four experiments, all employing learning procedures that lack the presentation of a discrete US, no effects of DEX were found. The dose used reliably induced memory enhancement in both appetitive and aversive discrete-cue Pavlovian conditioning paradigms of Chapter 2. These results suggest that glucocorticoid memory modulation relies on the presence of a discrete US, and support reports from the human literature that glucocorticoids only modulate emotional memory (Buchanan & Lovallo, 2001). The presence of a discrete US might result in further physiological processes that may be necessary for glucocorticoids to modulate

240

memory. For example, it has been suggested that glucocorticoids interact with the central noradrenergic system (Quirarte et al., 1997; Roozendaal, 2001). Hence, one might speculate that emotional events (US) might lead to the release of NA that glucocorticoids might depend on to exert a memory enhancing effect (see Chapter 3 for a more detailed discussion). Alternatively, the dose of DEX may not have been sufficient in order to modulate memory. The level of additional glucocorticoids released endogenously as a consequence of the training experience is likely to have been lower than in the experiments of Chapter 2 where footshock- or food-US occurred. One should bear in mind that the results presented here rely on experimental manipulations that may not have ecological validity. The adaptive nature of memory modulation itself, discriminating important from less important learning, would be corrupted if the underlying neurobiological system kicked in every time learning between neutral stimuli or about the absence of a reinforcer occurs. The aim of Chapter 3 was to examine if emotional significance and memory enhancement could be simulated by administering exogenous glucocorticoids alone, or whether these effects might depend on certain environmental situations (US presence). The results suggest the latter, as no effects of DEX on any of the paradigms were found, and one might hence speculate about possible neural interactions with other modulatory systems. This is also interesting in the light of Pugh et al’s (1997b) proposed role of glucocorticoids in context formation, a process that could be regarded as S-S learning. However, the four paradigms of Chapter 3, despite their common feature of lacking a discrete US, might also differ from each other in their degrees of emotionality. Whereas LI and SPC are likely to constitute purely non-emotional procedures, EXT and SOC may well contain emotional elements. Therefore, the reasons that DEX did

241

not result in any effect may differ between paradigms. Further experiments, e.g. administering different doses of DEX or concurrent NA, are necessary to clarify these questions.

6.1.4 Differential modulation of different reinforcer properties After it had been established from the studies of Chapter 3 that the presence of a US appears to be necessary for glucocorticoids to modulate memory, the question arose if this memory modulation selectively affects different reinforcer (US) properties. Reinforcers can be described regarding different types of properties. A distinction can be made between the motivational properties of a US, which determine their direction of emotional valence (appetitive vs. aversive), and their sensory properties. The latter reflect the exact physical characteristics of reinforcers (taste, texture etc.), making them distinct from each other, even within the same direction of emotional valence. As mentioned before, the great majority of studies investigating the role of glucocorticoids in memory have employed aversive learning procedures, typically involving a footshock US. Given its extremely brief duration but yet unpleasant nature, it is difficult to describe an aversive US such as footshock in terms of its sensory vs. motivational properties. Hence, there had not been a reinforcer-specific analysis of glucocorticoid memory modulation. In contrast to footshock however, appetitive food reinforcers, such as pellets or sucrose, can be analysed more readily in terms of both their motivational and sensory properties. A paradigm that allows such analysis is PIT (e.g. Kruse et al., 1983). One can make use of the fact that Pavlovian stimuli exert control over instrumental behaviour by a general as well as an outcomespecific process. The fact that responding during any reward signal may be energised

242

due to non-specific arousal reflects the general process. This effect relies on the motivational valence of a US. However, a selective elevation of responding occurs during presentation of a reward signal that shares the same outcome with that instrumental action. This outcome-specific transfer effect relies on the sensory properties of the reinforcer. In a PIT design, rats would typically undergo a Pavlovian conditioning stage during which two different stimuli might be paired with two different reinforcers (tone-X and clicker-Y). They would also receive lever training in which pressing of one lever (A) might be rewarded with X, and pressing of another lever (B) might result in Y. In a latter extinction test, increased general responding during either stimulus (as compared to ISI) constitutes the general transfer effect, allowing analysis of the motivational properties, whereas the outcome-specific transfer effect relies on the sensory properties to result in increased responding of A compared to B during tone presentations. The aim of Chapter 4 was to use such a paradigm to test if glucocorticoid memory modulation selectively acts on dissociable reinforcer properties. Using a within-subject design, DEX was administered immediately after Pavlovian pairings of one stimulus and reward, whereas SAL treatment followed alternative stimulus-reward pairings. In line with the findings and conclusions of Chapter 2, DEX was found to enhance appetitive Pavlovian conditioning. This had not been shown in a within-subject paradigm before. In the transfer test, DEX did not affect the general process but impaired the outcomespecific transfer effect, suggesting that DEX treatment after conditioning compromises associations with the sensory properties of the reinforcer. Given that Pavlovian conditioning was enhanced, it was concluded that post-training DEX memory enhancement might depend on strengthened associations with the motivational properties of the reinforcer. The procedure may not have been sensitive

243

enough to reflect this as an increase of the general transfer effect. In other words, glucocorticoid memory modulation seems to be bidirectional, enhancing the motivational properties of a reinforcer at the cost of the sensory processing, possibly by some overshadowing mechanism. In a more indirect way, the results of Chapter 4 might also be of interest to the study of amygdala functioning in emotional learning. Lesions of the BLA produce a strikingly similar pattern of results in PIT procedures, i.e. they impair the outcomespecific transfer effect while leaving the general transfer effect intact (Blundell et al., 2001). The BLA has been described as a critical structure in glucocorticoid memory modulation (e.g. Roozendaal & McGaugh, 1997a; Roozendaal, 2000), integrating hormonal and neuromodulatory factors on memory consolidation. Further, the BLA has been proposed to play an important role in the processing of the sensory properties of the reinforcer, integrating their representation alongside the motivational properties (Killcross and Blundell, in press). The fact that DEX, a glucocorticoid that modulates memory, appears to create a functional lesion of a structure (BLA) that has been regarded as crucial in glucocorticoid memory modulation appears paradoxical at first. However, one could speculate that processing of sensory and motivational properties of a reinforcer are in competition with each other, and that the memory modulatory effects of glucocorticoids in the BLA occur at the expense of its ability to generate a mental representation of the sensory reinforcer properties. This could be seen as a functional shift in the BLA with increasing levels of stress. The motivational properties of the US might take priority in processing, possibly by BLA activation of other brain structures such as the phylogenetically older CEA, while processing of the sensory properties is reduced. This shift may constitute the neural basis of the overshadowing mechanism mentioned above.

244

All demonstrations of glucocorticoid memory modulations, both in this thesis and elsewhere, involved relatively simple, contextual or discrete-cue, first-order Pavlovian or simple instrumental learning paradigms. Hence, glucocorticoidmodulated memories seem to be of a simpler type, i.e. restricted to the motivational aspects of an outcome, possibly coming at the price of losing detailed information. Such a model of cognitive processing in stressful or arousing situations has appeal from an evolutionary standpoint. Even though this is highly speculative (as always when considering evolution), it seems sensible that it is not knowledge about the sensory properties of a threat or reward that enhances survival chances at the next encounter, but rather knowledge about its motivational properties, which determine classification as a threat or reward in the first place. An overconsolidation of these motivational properties however might be maladaptive, possibly leading to the development and maintenance of psychopathological conditions such as PTSD and phobias (aversive stimuli) or drug addiction (appetitive stimuli). Chapter 4 provides support for real-life examples of anecdotal character, where the motivational or emotional meaning of an event is remembered a lot better than specific detail. A former smoker who always used to smoke in the pub or after a meal is likely to be reminded of how rewarding the cigarette felt when encountering these situations, rather than how the smoke tingled in his or her throat. Similarly, many people might well remember how a book or a film that they read/saw a long time ago made them feel really good or bad, but when asked about the plot, they realise they cannot remember much of it. The findings also provide some support for the null results of Chapter 3. If glucocorticoids only enhance associations with the motivational properties of a reinforcer, it seems plausible that no enhancement is observed when such

245

motivational properties may not be present or play a more minor role, as it could be argued to be the case in the procedures used in Chapter 3. All in all, the results of Chapter 4 may not only provide useful insights into the study of memory consolidation and the role of glucocorticoids. They may also have implication for general amygdala functioning, as well as for the development of certain psychopathological conditions such as post-traumatic stress disorder and drug addiction. Furthermore, given the loss of sensory detail during stress, they may provoke some thought concerning the validity of eye-witness testimonies of crime victims.

6.1.5 Individual differences in learning between high and low responders to stress The aim of Chapter 5, the final empirical chapter of this thesis, was to investigate differences in learning related to stress hormones, without employing pharmacological treatment. Instead, established individual differences in HPA reactivity were made use of, in order to set up an endogenous influence of glucocorticoid levels. The question of interest was if individual differences in HPA reactivity result in individual differences in basic emotional learning. If that was the case, a number of interesting implications would arise. Piazza is one of the discoverers of intrastrain-individual differences in HPA reactivity (e.g. Piazza et al., 1991a). He also developed a screening technique, based on locomotor activity to novelty, to divide a population of rats into HR and LR. His main interests constitute the interaction of glucocorticoids with the mesolimbic DA system during rewardrelated behaviour (e.g. Piazza & Le Moal, 1997). For example, Piazza and colleagues showed that HR self-administer amphetamine more readily than LR. (Deminiere et al., 1989; Piazza et al., 1989; Piazza et al., 1990). They argued that this was due to an

246

increased degree of reinforcement from amphetamine in HR, due to a glucocorticoid modulation of phasic mesolimbic DA (for a review see Piazza & Le Moal, 1996). Piazza and colleagues suggested from these findings that HR may have a greater propensity for sensation-seeking and drug addiction. However, in their papers Piazza et al. did not consider or mention the effects of glucocorticoids on memory consolidation (e.g. Roozendaal, 2000; this thesis), even though self-administration of amphetamine is assessed in a discrimination learning task. Chapter 5 therefore attempted to converge the ideas of Piazza and the glucocorticoid memory modulation literature. In order to do so, a population of rats was divided, on the basis of their locomotor activity to novelty, into HR and LR. They were then tested in a variety of learning paradigms, some of which had been shown to be influenced by exogenous glucocorticoid administration (Chapter 2). The hypothesis was that HR would show better learning than LR. In other words, the distinction between HR and LR simulated, by virtue of presumed endogenous differences, the pharmacological manipulations (DEX vs. SAL) employed in Chapter 2. If HR learned better than LR, this might also imply that the observed differences between HR and LR in amphetamine self-administration could, at least to some extent, result from better learning rather than merely from an increase of reinforcer value. The first experiment compared the two groups in an appetitive discrete-cue Pavlovian conditioning paradigm, employing auditory stimuli as CS and food pellets as US (as in Experiments 2.3 and 2.4). As hypothesised, HR showed better learning than LR. These differences were not attributable to baseline differences. While these results give support to the idea that differences in amphetamine self-administration between HR and LR could result from learning differences, it is, on the contrary, also

247

possible to attribute the differences observed in the current study, presumably in learning, to differences in the value of the food pellet reinforcer. While both accounts, enhanced learning and enhanced reinforcer value, would account for the observed results of enhanced appetitive conditioning, they can be dissociated in an aversive conditioning procedure. If learning per se is enhanced due to glucocorticoid memory modulation, then HR should show better learning in an aversive procedure also. However, if the differences in HPA activation stem primarily from rewarding features of glucocorticoids, by virtue of enhancing mesolimbic DA transmission in response to either appetitive or aversive reinforcers, one might expect that the salience of the aversive footshock would be compromised by glucocorticoids, leading to relatively impaired learning in HR. A discrete-cue Pavlovian fear-conditioning procedure with a visual CS and a footshock US was carried out, measuring conditioned suppression as in Experiment 2.2. As hypothesized, HR showed greater levels of conditioned suppression than LR, giving tentative support to the idea that higher levels of amphetamine self-administration in HR depend, at least to some degree, on differences in learning processes, rather than reinforcer values. Even though the results do not support the mechanism which Piazza and colleagues proposed why HR might be predisposed to develop drug addiction, the findings are in concordance with that general idea. Better learning between stimuli in the environment and the reinforcing properties of a drug might make people more likely to engage in consumption of the drug, promoting the development of addiction. By the same token, better learning about aversive stimuli might make HR individuals more likely to develop anxiety disorders such as PTSD or phobias. At least with respect to the influence of emotional learning, one could speculate about drug addiction and PTSD being related constructs with different directions of emotional

248

valence. Further, insufficient emotional learning might lead to reduced avoidance and approach of aversive and appetitive stimuli respectively, which might contribute towards the development of depression. The final experiment of Chapter 5 investigated differences between HR and LR in an appetitive instrumental learning procedure, in which lever pressing was rewarded with sucrose. No differences were found between groups in the rate of lever press acquisition. In line with the results of appetitive Pavlovian conditioning, HR showed greater levels of magazine approach during the first lever training session. These results suggest that HPA-related individual differences in learning might be selective for Pavlovian learning. However, the aim of this last study was not merely to assess instrumental acquisition rates but also to look at possible differences between the groups regarding shifts in the nature of associations governing instrumental performance with extended training. Instrumental behaviour is thought to be based on two types of associations. A-O associations reflect goal-directed behaviour and can be inferred from an animal’s sensitivity to a change in reinforcer value, e.g. a devaluation. In contrast, mechanistic S-R associations or habits are independent of the current value of the reinforcer, reflecting an absence of purposive action. According to some researchers, during the course of training, the relative contribution of these associations to behaviour are thought to shift from predominantly A-O responding to S-R habits (Adams & Dickinson, 1981; Adams, 1982; Dickinson, 1985). In order to test for possible differences between HR and LR in the governing of instrumental behaviour, a reversible reinforcer-specific devaluation treatment was applied repeatedly, after different levels of instrumental training, and responding was tested in extinction. Reinforcer devaluation took place by pre-feeding the animal with sucrose prior to test. It was found that HR lost their sensitivity to the reinforcer devaluation

249

procedure more readily than LR. With a low level of training, both groups responded more vigorously when the reinforcer had not been devalued. This was also the case with a medium level of training in LR, but not in HR. After a high training level, both groups responded equally in the devalued and non-devalued condition. The results can be interpreted as HR forming S-R habits more readily than LR. Alternatively, HR might have undergone an overshadowing process of motivational over sensory reinforcer properties (see Chapter 4) which could also have led to indiscriminate responding between the devaluation and non-devaluation conditions. Regardless of the mechanism, it seems intuitively sensible that behaviour becomes increasingly simple and ‘short-cut’ as more training occurs, and as the behaviour becomes increasingly procedural. With such a reduction of complexity, the loss of sensoryspecific representations of an outcome, possibly by a functional shift within the amygdala (see Chapter 4), may well be the first step, followed eventually by the development of habits in which behaviour becomes purely stimulus-response driven, with the outcome no longer represented at all. On a neural level, suggestions have been made about a shift from activity in the prelimbic cortex to the infralimbic cortex (Coutureau & Killcross, 2001) in the course of this transition. The results are also interesting in the light of drug addiction theories. It has been argued that drug addiction can be understood as the product of aberrant Pavlovian and instrumental learning processes (Robbins & Everitt, 2001). Both the formation of habits, as well as the (over-) consolidation of motivational properties of an outcome have been implicated in drug addiction. For example, Everitt et al. (2001) suggested that drug-seeking behaviour is initiated under the control of an A-O process but becomes a compulsive habit governed by S-R processes, as addiction sets in. Berridge (1996) proposed a model of drug addiction that dissociates liking and

250

wanting of a drug. Whereas drug liking is thought to depend on the drugs’ sensory properties, Berridge believes that associations between environmental stimuli and the motivational properties of the drug mediate drug wanting and thereby addiction. Again, the results reported here give support to the claim of Piazza and colleagues that HR, as compared to LR, might be more prone to drug addiction. However, while Piazza and colleagues (e.g. Piazza & Le Moal, 1996) argue that this is due to an enhancement of reinforcer value in HR, the results presented here emphasise differences in Pavlovian learning rates (Experiments 6.2 and 6.3), as well as differences in how rapidly shifts occur in the nature of instrumental behaviour. Therefore, the findings highlight the importance of simple learning and memory processes, and how individual differences in such processes may result in differences in the development of psychpathology. Specifically, this means that despite the possible importance of an interplay between glucocorticoids and mesolimbic DA in predisposing an animal to drug addiction, it is also important to consider the memory modulatory influences of glucocorticoids.

6.2 LIMITATIONS

There are a number of limitations to the findings presented in this thesis. First and foremost, all pharmacological studies in this thesis involved systemic drug administrations. Therefore, it is impossible to make any inferences regarding the neural basis of glucocorticoid memory modulation. Any suggestions made on that level, for example regarding the role of the amygdala in the processing of sensory and motivational reinforcer properties, albeit being of high theoretical interest, remain

251

purely speculative. In order to investigate the role of different brain structures in different aspects of glucocorticoid memory modulation, it would be necessary to employ lesion studies or to use localized microinfusions. Drugs like the selective GR agonist RU28362 and the selective GR antagonist RU38486 would be suitable for microinfusion studies. This would constitute an interesting future direction in the field of glucocorticoid memory modulation. Secondly, it should be noted that dose-response studies were only carried out in some of the procedures (Experiments 2.1 and 3.1). In general, the established dose of DEX that reliably enhanced memory in simple Pavlovian conditioning (1.2 mg/kg) was used. Given that glucocorticoids have been reported to act according to an inverted U-shaped relationship, it is possible that some of the null results, for example, would have looked different with a different, possibly higher, dose. It would therefore be interesting to repeat some of these studies, particularly the EXT and SOC procedures, with different doses of DEX or other glucocorticoids. Given that NA has been implicated to interact with glucocorticoids in the modulation of memory, it might furthermore be worth studying the effects of DEX on learning with concurrent administration of NA or ß-blockers. Finally, the entire project was based on manipulations that enhanced glucocorticoid levels. It might be worth attempting to replicate some of the current findings employing administration of GR antagonists, such as RU38486, or carrying out short-term ADX. In order to learn more about levels of endogenous glucocorticoid release resulting from different training events (e.g. footshock), and how this might interact with exogenous treatments, it would also be a good idea to measure plasma corticosterone levels. This technique could also be used to verify greater stressinduced glucocorticoid release in HR Lister-Hooded rats.

252

6.3 CONCLUDING REMARKS

This PhD project was concerned with the acute role of glucocorticoids in associative learning and the modulation of memory consolidation. Employing only a narrow range of learning paradigms, this field had so far been studied primarily in terms of possible neural bases of this modulation and interactions between glucocorticoids and other neurotransmitters. In most studies, such questions were investigated in aversive contextual learning procedures. Not much attention had been given to the distinction between emotional and neutral learning, nor had there been a great deal of interest in the role of different reinforcer properties in relation to glucocorticoids’ role in memory. The same goes for the study of HPA-related individual differences in learning. The goal of the current PhD project was to investigate these questions more closely, and to conduct a behavioural, stimulus-centred, analysis of how glucocorticoids modulate memory. Figure 6.3.1 constitutes an attempt to schematically summarise the findings and conclusions of this thesis, and to converge and encapsulate them into an overall model of the role of glucocorticoids in associative learning that might constitute an interesting expansion or alternative to the model outlined at the start of this thesis (Figure 1.2.6.1). For this model to be developed, psychopharmacological experiments in rats were carried out. It was found that glucocorticoids modulate memory not only in contextdependent procedures but also in discrete-cue Pavlovian conditioning. Further, glucocorticoids were found to modulate memory in appetitive procedures, suggesting a common general arousal mechanism mediating this modulation. However, the effect could not be shown in learning procedures lacking discrete US presentations, suggesting

that

glucocorticoids

selectively

253

modulate

emotional

general arousal

Ch.5

LR MeR

HPA activation

GR

HR

sensory properties (functional amygdala shift ?)

emotional event (US)

Ch.2

-

* motivational properties /

Ch.4

no HPA activation

Ch.3 non-emotional event (no US / S neutral)

Ch.3

discrete stimulus (CS)

+ MeR

HR

overconsolidation maladaptive response psychopathology

drug addiction ?

GR

exogenous (e.g. DEX)

Memory modulation

(NA-interaction ?)

* overshadowing

experimental treatment

Memory modulation

Ch.2

endogenous (CORT)

LR

adaptive response to situation

PTSD ? phobias ?

approach to appetitive stimuli

clinical speculations underconsolidation maladaptive response psychopathology

avoidance of aversive stimuli

depression ?

Ch.5/6

depression ?

Figure 6.3.1. Overall model of glucocorticoids in associative learning and memory modulation, derived from the empirical results, as well as from the ideas and speculations, of this thesis. The findings on instrumental learning and habit formation (Experiment 5.4) are omitted for the sake of clarity. Ch. refers to the chapter that contributed to the particular part of the model. Faces (happy/sad) indicate the direction of valence of the emotional event that triggered the modulation. HR = High Responders, MeR = Medium Responder, LR = Low Responders. Thick arrow heads signal different levels of CORT release (HR vs. LR). Thick, dotted arrow with + indicates strengthening of association

through memory modulation, whereas thin arrow with - signals a decrease in associative strength.

254

learning. A closer analysis of the different reinforcer properties implied that glucocorticoid memory modulation comes at a cost. Glucocorticoids seem to strengthen associations with the motivational properties while overshadowing processing of the sensory properties. Finally, it was hypothesized and tentatively confirmed that individual differences in stress-reactivity might result in individual differences in learning. These results reported here may not only make some important suggestions regarding the role of glucocorticoids (and more generally emotion) in associative learning and memory. They may also be of indirect interest to the study of brain structures that have been implicated to play an important role in emotional learning, particularly the amygdala. Last but not least, the results of this thesis may have some value

towards

a

better

understanding

of

the

development

of

certain

psychopathological conditions that might relate to emotional learning, such as PTSD or drug addiction.

- - THE END - -

255

APPENDIX Pilots, parameters and drug selection

256

In order to investigate the effects of post-training glucocorticoids on memory consolidation, a number of pilot experiments had to be conducted in the initial phases of this PhD project to establish the correct parameters for initial learning procedures. These are described in this Appendix. As glucocorticoid memory modulation has mainly been demonstrated in inhibitory avoidance or spatial tasks, and as there are even claims that its role is limited to learning about the context (see Chapter 3), investigating it in a discrete-cue Pavlovian conditioning paradigm appeared to be a worthwhile venture. Moreover, an appropriate drug had to be selected to allow experimental manipulation of glucocorticoid activity in future experiments, and its preparation and dosage had to be determined. This Appendix also discusses the selection of the GR agonist DEX that was used in many studies throughout this project.


In this first pilot experiment, the aim was to set up a simple fear-conditioning procedure to observe how an aversive CS-US association is formed, and how it extinguishes over repeated non-reinforced trials of the CS. The correct procedure and parameters were sought for future studies of fear conditioning that might involve pharmacological manipulations. The fear-conditioning procedure employed involved the assessment of freezing as an index of fear. Freezing is a species-specific fear response in the rat and suppresses all ongoing behaviour and movement apart from breathing (see Figure A.1.1). Freezing becomes a conditioned response (CR) when it

257

occurs during presentation of a CS, as a consequence of learning about the relationship between that CS and an aversive event (e.g. footshock US).

Figure A.1.1. A rat that displays freezing in one of the pilot experiments. The lit bulb on top of the chamber indicates the current occurrence of an aversive auditory CS.

Method Subjects Subjects were eight male, Lister Hooded rats (Harlan, UK), weighing between 290 g and 315 g at the start of the experiment (M = 301 g). They were housed in pairs in a temperature-controlled colony room (21±2ºC) and maintained on a 12:12-hrlight-dark cycle (lights on at 08:00 am) with free access to water. Prior and during the experiment, rats were fed approximately 15g of food pellets per day to maintain their body weights. Apparatus All behavioural procedures took place in a modified Campden chamber (Campden Instruments Ltd., UK; 24.5-cm-wide x 20.5-cm-high x 22-cm-deep) located inside a soundproof cabinet, equipped with a fan providing a 62-dB

258

background noise. The chamber was constructed from aluminium (side walls and ceiling) and plexiglass (hand hinged front door). The floor of the chamber consisted of 16 stainless steel bars spaced 1.5 cm apart centre-to-centre. Bars were wired and connected to a Coulbourn precision-regulated animal shocker (model E13-14, Coulbourn Instruments, USA), set to deliver scrambled footshock as US (0.3 mA, 0.5 s). The shock intensity was based on what is generally considered a mild footshock in the literature. A house-light on the centre of the ceiling was lit during the experiment. A tone (3 kHz, 80 dB, 10 s) served as the auditory CS and was delivered by two speakers in the left wall of the box. The door of the chamber remained open during sessions and the crucial parts of the procedure were filmed with a video camera and later assessed manually. A green and a red light, as well as a lit timer outside the box, indicated the various stages of the experiment. A BBC Master 128 microcomputer (Acorn, UK) equipped with a SPIDER extension for on-line control (Paul Fray Ltd, UK) controlled the equipment and recorded the data. Rats were run one at a time. Behavioural procedure Habituation and Conditioning. On day 1, rats were placed into the chamber and received a conditioning session that lasted 60 min. After 3 and 6 min, they were presented with a non-reinforced presentation of the tone, in order to assess and habituate any unconditioned freezing. In three further trials after 15, 30 and 45 min the tone was co-terminated with the occurrence of a mild footshock. Extinction. On day 2, rats were placed into the same chamber at roughly the same time of day as the acquisition session took place. Extinction sessions lasted 60 min and comprised five non-reinforced presentations of the tone, starting after 10 min with a fixed interstimulus interval (ISI) of 9 min 50 s. Over two weeks later, on day 18, a second extinction session of the same format took place. At the end of each session, rats were

259

returned to their home cage. Assessment of freezing. All stimulus presentations were filmed, including a preCS phase of equivalent length (10 s). Freezing was defined as the absence of any movement other than breathing. In order to assess freezing during the preCS and CS periods, two raters independently watched the recordings while listening to a sequence of beeps (1/s) created by BBC Master 128 microcomputer, programmed for this purpose. From the beginning of the preCS period to the end of the CS period, the rater made a decision during each beep whether freezing took place during that moment or not. See Table A.1.1 for an outline of the experimental design. Statistical analysis All data were first processed with Microsoft Excel (Microsoft Corporation, USA) on a DAN personal computer (Dan Technology Ltd., UK) and then analysed with the statistical software package CLRAnova (Clear Lake Research, USA) on a LC475 Macintosh computer (Apple Computer Inc., USA). Analyses of variance (ANOVA) and, when appropriate, simple-effect analyses (Levine, 1991) and posthoc tests (Newman-Keuls Pairwise Comparisons) were conducted. Pearson Productmoment correlations (r) were computed to assess inter-rater reliability. Two animals were excluded from the analysis due to technical failures, leaving six subjects altogether.

Table A.1.1. Design of Experiment A.1. A represents the tone CS. + indicates aversive reinforcement by mild footshock. HAB trials are not included.

PHASE

-


Extinction 1

Extinction 2

Day 1

Day 2

Day 18

2xA 3 x A+

5xA

5xA

260

Results Freezing was assessed independently by two raters. 10-s stimulus periods were taken into account and Pearson’s Product-moment correlation coefficient (r) was calculated. Scores between the two raters correlated significantly, r = 0.72, p < .001. Only the results of the rater that was not also the experimenter are reported. Figure A.1.2 illustrates freezing behaviour in the preCS- and CS period across the habituation (HAB), conditioning (COND) and extinction (EXT) phases. Freezing during CS presentations increased following conditioning. The data were analysed in two ways. In the first analysis, extinction trials 5 to 10 were omitted and the remaining nine trials were divided into three 3-trial phases. The first phase, HAB, contained the two non-reinforced tone presentations plus the first conditioning trial (HAB 1 + 2, COND 1). During presentation of the tone in the first conditioning trial, rats were still naïve about their relationship to the footshock so that functionally, this trial belonged to the HAB phase. The second phase, COND, contained the second and third conditioning trial plus the first extinction trial (COND 2 + 3, EXT 1). Again, the latter belonged functionally to the COND phase. Having experienced two tonefootshock pairings before, rats were yet unaware of the non-reinforced nature of this trial. The last phase, EXT, contained EXT trials 2 to 4, all following non-reinforced trials. A three-way, 3 x 2 x 3 ANOVA was carried out with three repeated measures, PHASE (HAB, COND, EXT), STIMULUS-STATE (preCS, CS) and TRIAL (1-3). There was a significant main effect of PHASE, F(2, 10) = 10.02, p < .01. A NewmanKeuls posthoc analysis revealed that freezing was significantly lower during HAB (M = 1.50) than COND (M = 3.61) and EXT (M = 3.81), but that the latter two phases did not differ. The increase in overall freezing following HAB suggests that learning about the footshock had occurred. There was a significant interaction of PHASE x

261

Freezing (s) during 10-s period

10 9

preCS CS

8 7 6 5 4 3 2 1 0 1

2

HAB

1

2

3

1

2

3

4

5

6

7

8

9 10

EXT

COND

Trial & Phase

Figure A.1.2. Freezing during the preCS and CS period across trials during the HAB, COND and EXT phases. Note that phases are labelled according to their actual trials, rather than according to the labels of the first analysis.

262

TRIAL, F(4, 20) = 3.86, p < .02. Simple effects revealed an effect of TRIAL in HAB (p < .03), indicating a decrease of unconditioned freezing (habituation), an effect of TRIAL in COND (p < .01) indicating an increase in conditioned freezing (conditioning), and no effect of TRIAL in EXT. Further, there was a significant main effect of STIMULUS-STATE, F(1, 5) = 15.45, p < .02, with overall preCS freezing (M = 1.26) being lower than CS freezing (M = 4.69). The interaction of PHASE x STIMULUS-STATE was also significant, F(2, 10) = 5.54, p < .05. Simple-effects analysis revealed that during the CS, freezing in habituation (M = 2.56) was significantly lower than in conditioning (M = 5.83) and extinction (M = 5.67), p < .005. The results imply that conditioning occurred to the tone but not to the context. In a second analysis, only extinction trials were taken into consideration and a two-way, 2 x 10 ANOVA with two repeated measures, STIMULUS-STATE (preCS, CS) and TRIAL (1-10) was carried out. There was a significant main effect of STIMULUSSTATE, F(1,5) = 37.63; p < .005, with freezing in the CS period (M = 5.25) being higher than freezing in the preCS period (M = 0.98). There was also a significant main effect of TRIAL, F(9, 45) = 2.65, p < .02. A Newman-Keuls post-hoc analysis revealed that there was significantly less freezing during the last extinction trial than in trials 2, 3 and 5, indicating that some extinction had taken place. The interaction of STIMULUS-STATE x TRIAL was marginally significant, F(9, 45) = 1.98, p = .064, showing a decline in CS and no change in preCS freezing levels, the latter of which remained at floor.

Discussion The pilot experiment resulted in a number of useful conclusions. First of all, learning between the CS (tone) and the US (footshock) had successfully been

263

demonstrated, and freezing as the dependent variable proved sensitive to indicate this relationship. Inter-rater reliability was reasonably high, suggesting that one blind rater would suffice in future assessments. CS scores might have been at ceiling during the beginning of non-reinforced trials so that extinction became apparent only towards later trials. Fewer, i.e. one or two, conditioning trials would probably be sufficient for animals to learn about the predictive relationship of tone and footshock. In fact, if memory-enhancing manipulations were to be employed, it would be important not to have a ceiling effect as this may conceal possible results. The large time gap of over two weeks between extinction trial 5 and 6 might have caused spontaneous recovery and contributed to the slow extinction rate. With just a single weaker shock, five extinction trials might suffice. If that were not the case, a greater number of extinction trials, with different sessions not greatly separated in time, ensuring the occurrence of extinction, might also prove useful to uncover memory differences. PreCS scores were low throughout the experiment, indicating that conditioning was specific to the tone. There was little evidence for context conditioning. For future studies, a CSpreCS score might prove useful.


The second pilot study built on the first one, attempting to improve parameters and eliminate some flaws and problems. Again, it constituted a simple fearconditioning procedure in which freezing was the dependent variable. However, this study employed a single-trial learning procedure, hence with only one CS-US pairing. Furthermore, a second auditory stimulus, a train of clicks was introduced in addition to the tone. In this within-subject design, all subjects learned an association between

264

tone and footshock on one day, and between clicker and footshock on another. Both stimuli were presented non-reinforced in the extinction session. The study was designed to found the basis of future experiments, in which pharmacological manipulations could take place. Footshock was lowered in comparison to (pilot-) Experiment A.1 from 0.3 mA to 0.25 mA, and the number of subjects was increased to 12.

Method Subjects Subjects were 12 male, Lister Hooded rats (Harlan, UK), weighing between 270 g and 295 g at the start of the experiment (M = 276 g). For housing and feeding information, refer to Experiment A.1. Apparatus The apparatus was the same as described for Experiment A.1 with the following modifications: 1. The footshock was lowered to 0.25 mA. 2. An additional auditory cue was available in form of a 10-Hz train of clicks (clicker) with a sound level of 75 dB. It was produced by a heavy-duty relay mounted behind the left wall of the chamber. 3. The sound level of the tone was matched to that of the clicker (75 dB). Behavioural procedure Habituation and conditioning. On day 1, rats were placed into the chamber and received a conditioning session that lasted 25 min After 3 and 6 min, they were presented with a non-reinforced presentation (A) of either tone (T) or clicker (C), in order to habituate to it and remove any unconditioned freezing. The same stimulus (A) was then paired with a mild footshock after 15 min. After 25 min, the session was

265

terminated. On day 2, another conditioning session was given at the same time of day, only this time the other stimulus (B) was presented, so that all subjects received conditioning with both stimuli, the order of which was counterbalanced. Extinction. On day 3, rats were placed into the chamber, again at the same time of day, and received an extinction session that lasted 30 min. It comprised five non-reinforced presentations of each stimulus in a pseudo-randomised order (T-C-C-T-C-C-T-T-CT). Inter-trial intervals were 2 min 50 s long. At the end of each session, rats were returned to their home cage. See Table A.2.1 for an outline of the experimental design. Assessment of freezing. Refer to Experiment A.1. Freezing was assessed by one rater, blind to experimental allocation. Statistical analysis Refer to Experiment A.1.

Table A.2.1. Design of Experiment A.2. A and B are either tone or clicker (counterbalanced). + indicates aversive reinforcement by mild footshock. HAB trials are not included. PHASE

DESIGN

Conditioning 1

Conditioning 2

Extinction

Day 1

Day 2

Day 3

1st (A) vs. 2nd (B)

A-B-B-A-B-B-A-A-B-A A+

B+

conditioning day

B-A-A-B-A-A-B-B-A-B

Results A CS-preCS difference score of freezing was used as the primary dependent variable, although preCS and CS scores alone were also examined. Data were analysed in two ways, comparing conditioning on day 1 vs. conditioning on day 2 266

(counterbalanced over tone and clicker), as well as comparing conditioning to tone vs. conditioning to clicker (counterbalanced over day 1 and 2). Figures A.2.1 and A.2.2 display these data. Freezing during CS presentations was increased following conditioning, irrespective of the day on which the stimuli were conditioned. Further, there was greater freezing to the clicker than to the tone in the later trials of extinction. Day 1 vs. day 2. A two-way, 2 x 8 ANOVA with two repeated measures, DAY (1 + 2) and TRIAL (HAB 1 + 2, COND, EXT 1 - 5) revealed no significant main effect of DAY, F(1, 11) = 1.31, p = .28, (Ms: day 1 = 3.53 , day 2 = 4.18), nor a significant interaction of DAY x TRIAL, F(7, 77) = 1.02, p = .43. This suggests that the day on which conditioning took place did not affect unconditioned or conditioned freezing. There was a significant main effect of TRIAL, F(7, 77) = 33.62, p < .01. Newman-Keuls posthoc analysis revealed that freezing during all EXT trials was greater than freezing during the HAB and COND trials, confirming the successful occurrence of conditioning between the CS and the US. Tone vs. Clicker. A two-way, 2 x 8 ANOVA with two repeated measures, STIMULUS (tone, clicker) and TRIAL (HAB 1 + 2, COND, EXT 1 - 5) was carried out. There was no significant main effect of STIMULUS, F(1, 11) = 1.50, p = .25 (Ms: tone = 3.51, clicker = 4.20). The significant main effect of TRIAL (reported above) remained, as the analyses were identical with respect to TRIAL. There was furthermore a significant interaction of STIMULUS x TRIAL, F(7, 77) = 4.03, p < .001. Simple-effect analysis revealed that freezing was greater to clicker than to tone in EXT trials 3 (p < .05) and 4 (p < .01). This suggests that extinction of clicker-shock associations proceeded more slowly than that of the tone-shock association. However, the process of extinction itself is not made evident by this analysis. In order to examine this, a separate analysis of extinction trials only was carried out. A two-way, 2 x 5 ANOVA with two repeated

267

10

Freezing (s), CS-preCS

9

Day 1 Day 2

8 7 6 5 4 3 2 1 0 -1 1

2

C

1

2

3

4

5

EXT

HAB & COND

Trial & Phase

Figure A.2.1. Freezing across trials to stimuli conditioned on day 1 and 2, during the HAB, COND and EXT phases.

268

10

Freezing (s), CS-preCS

9

Tone Clicker

8 7 6 5 4 3 2 1 0 -1 1

2

C

1

2

3

4

5

EXT

HAB & COND

Trial & Phase

Figure A.2.2. Freezing across trials to tone and clicker, during the HAB, COND and EXT phases.

269

measures, STIMULUS (tone, clicker) and TRIAL (EXT 1-5) revealed a marginally significant main effect of STIMULUS, F(1, 11) = 4.33, p = .06 (Ms: tone = 4.93, clicker = 6.50). There was no significant main effect of TRIAL, F(4, 44) = 1.46, p = .23 but a significant interaction STIMULUS by TRIAL, F(4, 44) = 3.71, p < .02. Simple-effect analysis revealed a difference of freezing between trials to tone (p < .04) but not to clicker (p = .30), suggesting that extinction only occurs of associations with the former but not the latter. PreCS scores. PreCS scores were at floor throughout trials (Ms: pre-clicker = 0.16, pre-tone = 0.14). A two-way, 2 x 8 ANOVA with two repeated measures, STIMULUS (pre-tone, pre-clicker), and TRIAL (HAB 1 + 2, COND, EXT 1 - 5) revealed no significant main effects or interactions in preCS freezing (all Fs < 1). CS scores. When considering CS rather than CS-preCS data, a two-way, 2 x 8 ANOVA with two repeated measures, STIMULUS (tone, clicker), and TRIAL (HAB 1 + 2, COND, EXT 1 - 5) revealed exactly the same pattern of results as with CS-preCS scores (data not shown).

Discussion The aim of this pilot study was to remove some of the problems that occurred in Experiment A.1. It introduced a second auditory stimulus and hence created the basis for a within-subject design. Each stimulus was paired with footshock in two conditioning sessions occurring over two days. The order in which clicker and tone were conditioned was counterbalanced. It was a one-trial learning study with the strength of the footshock slightly reduced, in order to avoid a ceiling effect early on in extinction. There was no difference in freezing during extinction between stimuli conditioned on the first and on the second day, supporting the within-subject design. However, freezing to clicker was greater than to tone in the later stages of extinction,

270

suggesting a greater salience of the clicker. Significant extinction only occurred to tone but not to clicker. The lack of a difference earlier on in extinction might have been due to a ceiling effect. This possible difference in salience between the two CS might cause an undesired increase in variance for upcoming studies employing experimental manipulations. For future studies, the salience of the two different CS would have to be better matched, for example by reducing the salience of the clicker. Furthermore, a greater number of extinction trials would be necessary, as extinction did not occur fully (tone) or at all (clicker) over the five trials. A further decrease in US magnitude might be necessary to avoid a ceiling effect at the start of extinction. PreCS freezing was virtually absent suggesting that no contextual conditioning occurred. As a consequence, results for CS-preCS and CS alone scores were virtually identical. In summary, Experiment A.2 resulted in useful insights regarding the parameters of a successful fear conditioning design. These were considered when planning Experiment A.3.


The aim of this pilot study was to further eliminate the problems that became apparent first two studies, i.e. the slow rate of extinction and the difference in salience between tone and clicker. Hence, the study was almost identical to its predecessor with some minor exceptions. First, the shock that served as the US was reduced to 0.225 mA (as compared to 0.25 mA and 0.3 mA before), in order to decrease the strength of the association and make the process of extinction more apparent. For the same reason, the heavy-duty relay that produced the clicker was detached from the

271

wall of the actual chamber and re-attached to a different position somewhat further away within the chamber-containing cabinet, and the relay now switched every 15 ms, i.e. a reduction of the clicker frequency to 6.66 Hz. It was anticipated that the clicker would now become less salient. The tone was made slightly louder than the clicker (80 dB), again to reduce the gap in salience between the two stimuli. The CS-US pairing now occurred after 20 rather than 15 min so that post-training administration of drugs in future studies would be closer to the learning event. Finally, the extinction session was prolonged to 60 min, comprising 10 rather than 5 non-reinforced presentations of each stimulus.

Method Subjects Subjects were six male, Lister Hooded rats (Harlan, UK), weighing between 290 g and 310 g at the start of the experiment (M = 300 g). For housing and feeding information, refer to Experiment A.1. Apparatus The apparatus was the same as described for Experiment A.2 with the following modifications: 1. The footshock was lowered to 0.225 mA. 2. The heavyduty relay that produced the clicker was detached from the wall of the actual chamber and re-attached to a different position somewhat further away within the chambercontaining cabinet. 3. The frequency of the clicker was reduced to 6.66 Hz. 4. The sound level of the tone was increased to 80 dB. Behavioural procedure Habituation and conditioning. Habituation and conditioning occurred in the same way as in Experiment A.2, with the modification that the CS-US pairing now

272

took place after 20 min rather than 15 min, i.e. 5 rather than 10 min before the session terminates. Extinction. The extinction procedure differed from that of Experiment A.2 only in that the session lasted 60 (rather than 30) min and comprised 10 (rather than 5) non-reinforced presentations of each stimulus in a pseudo-randomised order (A-B-B-A-B-B-A-A-B-A-B-A-A-B-A-A-B-B-A-B). Assessment of freezing. Freezing was assessed as in Experiment A.2. See Table A.3.1 for an outline of the experimental design. Statistical analysis Refer to Experiment A.1.

Table A.3.1. Design of Experiment A.3. A and B are either tone or clicker (counterbalanced). + indicates aversive reinforcement by mild footshock. HAB trials are not included. PHASE DESIGN 1st (A) vs. 2nd (B) conditioning day

Conditioning 1

Conditioning 2

Extinction

Day 1

Day 2

Day 3

A+

B+

10 x A 10 x B

Results Figure A.3.1 illustrates freezing behaviour to tone and clicker across extinction. There was no difference in freezing between the two stimuli. CS scores were used as the dependent variable and only the extinction data was analysed. A two-way, 2 x 10 ANOVA with two repeated measures STIMULUS (tone, clicker) and TRIAL (extinction trials 1-10) was carried out. There was no significant main effect of STIMULUS, F(1, 5) < 1, indicating that the stimuli, clicker (M = 6.22) and tone (M = 5.80) did not differ significantly from each other in their salience. A significant main effect of TRIAL, F(5, 9) = 2.38, p < .03, confirmed that extinction took place. 273

10 Tone Clicker


9 8 7 6 5 4 3 2 1 0 1

2

3

4

5

6

7

8

9

10

Extinction Trial

Figure A.3.1 Freezing across 10 extinction trials to tone and clicker.

274

There was no significant interaction between STIMULUS and TRIAL, F(9, 45) = 1.14, p = .36.

Discussion The third pilot study successfully improved the parameters of the fearconditioning paradigm. As anticipated, the difference in conditioning to the two auditory stimuli was eliminated. Extinction was observed over ten trials with comparable rates to both stimuli. The parameters were set for the first study involving the experimentally manipulative administration of a drug to assess its effect in this paradigm.

A.4 WHICH DRUG TO SELECT ?

There are a number of ways in which one can investigate modulatory effects of glucocorticoids on memory pharmacologically. For example, there are different points in the time course of learning at which a drug can be administered. Conceptually, learning and memory is often divided into three processes:

1.

Acquisition – the point in time when the learning event, e.g. the CSUS association, is experienced.

2.

Consolidation – the period after the learning event, presumably during which a cascade of physiological events result in the storage of that event into long-term-memory.

3.

Retrieval – a later point in time when the learned event is recalled.

275

These concepts are useful in understanding the possible problems that one needs to be aware of when studying memory modulation. This project’s topic of investigation is the ability of glucocorticoids to modulate, i.e. influence either by enhancement or deterioration, the formation of memory as a consequence of their release after the learning event. In a real-life situation, increased levels of glucocorticoids are not present at the time when the learning event begins to take place but only shortly afterwards. Hence, it is important to dissociate memory modulation by glucocorticoids from possible effects on attentional, motivational or sensory-perceptual mechanisms that may operate during acquisition. The retrieval process is also not the focus of investigation as such, although it serves as a test for memory acquisition and consolidation under experimental conditions. When studying memory consolidation pharmacologically, it is therefore important that subjects are drug-free when the learning events take place (acquisition) and when memory is later tested (retrieval). A solution to this is constituted by the post-training administration technique (Breen & McGaugh, 1961; McGaugh, 1966). When drugs are given immediately after a learning event but are no longer present when the learned memory is retrieved at a later point, one can safely eliminate aforementioned effects on acquisition or retrieval, or of state-dependency, and conclude that any effect of the drug occurred during the consolidation phase. After deciding at which stage to give the drug, it is now important to decide which drug to select. Conventionally when studying the role of a certain endogenous substance, e.g. a neurotransmitter or a hormone, one could either attempt to mimic the actions of that substance by administering an agonist, or attempt to block its actions by administering an antagonist. As described in Chapter 1, glucocorticoids bind to two types of receptors: the MR and the GR. Most evidence suggests that the memory

276

modulation by glucocorticoids is mediated by GRs (e.g. Veldhuis et al., 1985; Oitzl & de Kloet, 1992; Sandi & Rose, 1994a; see Chapter 1). A number of compounds acting on GRs are briefly described below and their suitability for this project is discussed.

Corticosterone (natural glucocorticoid) Corticosterone is the primary endogenous glucocorticoid in rodents. The human equivalent is cortisol. Corticosterone first binds to MRs until those are fully saturated. It then binds to GRs, usually during stressful events. Post-training corticosterone has been shown to reverse ADX-impairment in a forced-swimming task only at very high doses, as compared to DEX and RU28362 (Veldhuis et al., 1985).

DEX (synthetic GR agonist) DEX binds more readily, i.e. with a ten-fold higher affinity, to GRs than to MRs, both in rodents and man (de Kloet, 1991). DEX has been used in a number of post-training memory modulation studies, usually employing inhibitory avoidance or spatial tasks (e.g. Roozendaal & McGaugh, 1996a, 1997b; Veldhuis et al., 1985). In those experiments, DEX was usually administered sc. The applied doses ranged from 0.1 mg/kg (Veldhuis et al., 1985) to 8.0 mg/kg (Flood et al., 1978) with an inverted Ushaped effect of memory enhancement (e.g. Power, Roozendaal, & McGaugh, 2000). DEX is best known for its clinical use in the so-called dexamethasone suppression test, helping to detect a hyperactive HPA axis common to a subgroup of clinically depressed patients (Carroll, Curtis, & Mendels, 1976).

277

RU28362 (synthetic, selective GR agonist) RU28362 is a selective GR agonist that does not bind to MRs at all. It has been used in a number of memory modulation studies and is most often administered centrally into specific brain structures in quantities of a few (0.1-10) ng (e.g. Roozendaal & McGaugh, 1997a,b), but has also been given systemically (Veldhuis et al., 1985).

RU38486 (synthetic, selective GR antagonist) RU38486 is a selective GR antagonist. It binds to GRs without any agonist activity. It has been used in a number of memory modulation studies both by icv (e.g. 0.3 – 3ng, Roozendaal & McGaugh, 1997a; 10 and 100 ng, Oitzl et al., 1998a) and sc (e.g. 10 mg/kg, Pugh et al., 1997a) administration. In another domain and under the name of RU486, it has become very widely known as the abortion pill.

In order to demonstrate the effect of an endogenous substance, mimicking the effect exogenously seems to be the most direct and convincing way. This speaks in favour of the use of an agonist (corticosterone, DEX, RU28362) rather than an antagonist (RU38486). As both, MRs and GRs have a similar affinity to corticosterone, while memory modulation is probably mediated by GRs, the use of a selective GR agonist seems more appropriate. Synthetic glucocorticoids are thought to be up to 1000 times more potent than corticosterone. Due to the reliably reported modulatory effects of DEX, mainly in the inhibitory avoidance task by Roozendaal and McGaugh’s group, DEX was selected as the pharmacological manipulation in this project. It was also more readily available in the UK (Sigma) than RU28362, which in most studies was donated by Roussel Uclaf, France.

278

However, there are a few problems associated with DEX that should be addressed. First of all, DEX is quite difficult to dissolve, requiring the use of ethanol (2-4% final concentration). One of the (peripheral) effects of DEX is the suppression of stress-induced endogenous ACTH in the pituitary and, as a consequence, glucocorticoid release (de Kloet, Wallach, & McEwen, 1975). In other words, treatment with DEX can create a state of “chemical-“ or “functional adrenalectomy”, i.e. chronic administration can result in the depletion of the brain’s endogenous glucocorticoids. DEX can also induce aptosis (cell death) in the hippocampus and striatum (Haynes, Griffiths, Hyde, Barber, & Mitchell, 2001), an effect also observed after ADX (Sloviter, et al., 1989). DEX does not enter the brain very readily (de Kloet et al., 1975). This was found to be due to the drug-transporting P-glycoprotein (Schinkel, et al., 1994). It is encoded by the mdr1a gene in mice and is expressed in the apical membrane of the endothelial cells of the blood-brain barrier and limits access of its substrate molecules to the brain to protect it from xenophobic agents (de Kloet, 1997). However, its capacity is limited and some DEX can still enter the brain if given in high enough doses (de Kloet et al., 1975; de Kloet, 1997). Very high doses of DEX however, can change the MR/GR balance with possibly destabilizing and threatening consequences for excitability and viability of neurons (de Kloet, 1997). It is therefore important to consider all these issues when using DEX in memory modulatory experiments. It should only be administered acutely, limited to one or very few treatments, rather than chronically, to prevent long-lasting functional adrenalectomy and aptosis. This however is already achieved by the very nature of the experimental design of memory modulation studies where administration usually occurs only after a single or very few training sessions. Considering possible effects of acute DEX on the endogenous glucocorticoid production - this system is shut down

279

for hours afterwards (M. S. Oitzl, personal communication, June 2000) - further training or test sessions should ideally be spaced 24 hr or longer from the point of administration. Again, this is usually the case in most experimental designs. The administered dose needs to be high enough to allow penetration of the blood-brain barrier by some DEX (probably about 0.1 mg/kg and more), but not too high as this makes dissolving difficult, and it might destabilize the MR/GR balance. Finally, doses that are too high may not be useful, as glucocorticoids are likely to exert their effects on memory according to an inverted U-shaped function (Roozendaal, 2000). Hence DEX only proves useful for memory modulation studies within quite a narrow dose window. This window seems to differ between laboratories and paradigms. Flood et al. (1978), for example, found memory-enhancing effects with doses as high as 8.0 mg/kg. This however seems to have been exceptionally high. The doses used by Roozendaal and McGaugh’s group (0.3 – 3 mg/kg) seem to fulfil the abovementioned criteria best and have been successful in demonstrating memory modulation, i.e. memory enhancement with post-training administration. One might argue that these effects were due to peripheral actions of DEX, i.e. negative feedback, at the pituitary and subsequent suppression of endogenous ACTH and corticosterone. Hence, the memory enhancing effects might have been due to lower levels of ACTH, or lower rather than higher levels of central corticosterone, with the endogenous release being shut down. Furthermore, as CRH release from the hypothalamus is not inhibited and can even be increased given the lack of negative feedback from endogenous corticosterone, one might also argue that CRH mediates the memory modulatory effects of DEX. These explanations, however, are highly unlikely. A number of reasons suggest a central effect of DEX instead. First of all, the results of many studies that employed DEX administration (e.g. Roozendaal & McGaugh, 1996a,b;

280

Roozendaal et al., 1996a,b; Setlow et al., 2000) were in the opposite direction of those memory-impairing ones observed with ADX (Roozendaal & McGaugh, 1996b; Roozendaal et al., 1998), with administration of the corticosterone synthesis inhibitor metyrapone (Roozendaal et al., 1996a), and with central (Sandi & Rose, 1994a; Roozendaal & McGaugh, 1997a,b, Oitzl et al., 1998a) or peripheral (Pugh et al., 1997a) administration of the GR antagonists RU38486. Secondly, the effects are in the same direction as those observed with central (Roozendaal & McGaugh, 1997a,b) and peripheral (Veldhuis et al., 1985) administration of the GR agonist 28362 and peripheral administration of corticosterone (e.g. Cordero & Sandi, 1998; Veldhuis et al., 1985). Moreover, the effects of DEX follow an inverted U-shape (e.g. Roozendaal, 2000) and can be reversed by central administration of the GR antagonist RU38486 (e.g. Roozendaal et al., 1999a), which shifts the DEX response curve to the right. Finally, DEX can also enhance memory in animals that underwent ADX (Roozendaal et al., 1996b). Considering both the advantages and disadvantages of DEX, it was decided to employ it as the pharmacological treatment in experiments on glucocorticoid memory modulation in this project.


Aim of this study was to employ the improved fear conditioning parameters in a study involving the post-training administration of DEX. DEX is a GR agonist that has been shown to enhance memory in a number of inhibitory avoidance studies (e.g. Roozendaal & McGaugh, 1996a; Roozendaal et al., 1999a; Setlow et al., 2000) but which has not been used in discrete-cue Pavlovian conditioning. The parameters of

281

Experiment A.3 were adopted and DEX and saline (SAL) were administered immediately after the conditioning session by sc injection. The dose of 0.3 mg/kg was adopted from Roozendaal and McGaugh (1996a,b). Drug treatment constituted a within-subject manipulation. All subjects received both treatments over the course of two days with stimulus-treatment constellations, as well as stimulus and drug order, counterbalanced.

Method Subjects Subjects were eight male, Lister Hooded rats (Harlan, UK) weighing between 295 g and 305 g at the start of the experiment (M = 299 g). For housing and feeding information, refer to Experiment A.1. Apparatus The apparatus was the same as in Experiment A.3. Drugs and injection procedure DEX (Sigma, UK) was injected sc immediately, i.e. within few min, posttraining at a dose of 0.3 mg/kg in a volume of 2.0 ml/kg. This dose was based on Roozendaal and McGaugh (1996a,b). DEX was first dissolved in absolute ethanol and subsequently diluted in 0.9% sodium chloride according to the appropriate dose. For the within-subject control condition, SAL (i.e. 0.9% sodium chloride) with a matched ethanol concentration of 2%. Drugs were prepared freshly on both days of conditioning and labelled RED and BLUE by an independent researcher so that the experimenter and rater remained blind as to which treatment was given at any stage.

282

Behavioural procedure Habituation and conditioning. These phases were the same as in Experiment A.3, except that drug treatment was administered immediately after the end of the session. Following that, animals were returned to their home cage. All subjects received both DEX and SAL, across the two conditioning dates. The stimulustreatment constellation, clicker and tone with DEX and SAL, as well as the order of stimuli and drugs were counterbalanced. Extinction. The extinction procedure was the same as in Experiment A.3, except that a further extinction session of the same kind was run on day 4. See Table A.5.1 for an outline of the experimental design. Assessment of freezing. Refer to Experiment A.3. Statistical analysis Refer to Experiment A.1.

Table A.5.1. Design of Experiment A.5. A and B are either tone or clicker, and X and Y are either DEX or SAL (all counterbalanced). + indicates reinforcement by mild footshock. PHASE Conditioning 1 Conditioning 2 DESIGN st

nd

1 (AX) vs. 2 (BY) conditioning day

Extinction 1

Extinction 2

Day 1

Day 2

Day 3

Day 4

A+ Treatment X

B+ Treatment Y

A-B-B-A-B-B-A-A-B-A B-A-A-B-A-A-B-B-A-B

see Day 3

Results Figure A.5.1 shows the level of freezing across extinction during stimuli that were followed by either DEX or SAL treatment in conditioning. Freezing did not differ between the two treatment conditions. CS scores were used as the dependent variable and a two-way, 2 x 20 ANOVA with two repeated measures, TREATMENT

283

10 DEX SAL


9 8 7 6 5 4 3 2 1 0 2

4

6

8

10

12

14

16

18

20

Extinction Trial

Figure A.5.1. Freezing across 20 extinction trials to stimuli followed either by administration of DEX or SAL in condition.

284

(DEX, SAL) and TRIAL (extinction trials 1-20) was carried out. There was no significant main effect of TREATMENT, F(1, 6) < 1, indicating that post-training treatment with DEX (M = 4.68) did not have an overall effect on freezing in extinction, as compared to post-training treatment with SAL (M = 4.55). There was a significant main effect of TRIAL, F(6, 19) = 18.10; p < .001, showing that extinction took place. There was no significant interaction of TREATMENT x TRIAL, F(19,114) = 1.52, p = .09.

Discussion Experiment A.5 made use of the insights and findings of its three predecessors and employed the optimised parameters in a within-subject fearconditioning paradigm. In order to ensure complete extinction, two sets of ten extinction trials were run. The GR agonist DEX was selected as a pharmacological treatment and administered after one of the two conditioning sessions, while SAL was given as vehicle after the other session. The constellations of stimulus type and drug treatment, as well as the order of stimuli and drug treatments, were counterbalanced. The advantage of such a design is that all animals are exposed to all stimuli and all drug treatments. This excludes the possibility that any effect of DEX observed in extinction may be due to general side effects. It was hypothesised that freezing might be greater to those stimuli whose conditioning session was followed by DEX rather than SAL treatment. This was not the case. However, the behavioural parameters of the study seemed optimised and the lack of a drug effect might have well been due to the dose of 0.3 mg/kg not being sufficient. The dose was based on experiments by Roozendaal and McGaugh (1996a,b), in which it proved sufficient to have a memory enhancing effect. However, their task was of a different nature, i.e. inhibitory

285

avoidance rather than discrete-cue Pavlovian conditioning, and might have been more aversive. This might have lead to a greater additional release of endogenous glucocorticoids, requiring a smaller dose of exogenous DEX. Given the assumed nonlinear, inverted U-shape function by which glucocorticoids are thought to exert an effect on memory (Roozendaal, 2000), one cannot assume that the level of endogenous glucocorticoid release does not interact with the level of exogenous treatment. Furthermore, it is always difficult to compare doses across laboratories due to the general parameters of both the learning paradigm and the drug preparation. In summary, the parameters for fear conditioning with freezing assessment were set so that pharmacological manipulation using DEX could be carried out in the future. A first pilot experiment with a low dose of DEX did not find any effects. However, a dose-response study with a number of different doses, from 0.3 mg/kg upwards, was planned to be carried out next (Experiment 2.1).

General Discussion This Appendix presented a series of pilot experiments that were concerned with optimising parameters for discrete-cue Pavlovian fear conditioning experiments. Auditory CS, a footshock US and freezing as the dependent variable were utilised in these studies, and parameters were improved gradually in order to make the procedure sensitive to the effects of experimental pharmacological manipulations. The Appendix furthermore discussed various natural and synthetic glucocorticoids, from which the GR agonist DEX was selected as treatment for further experiments in this project.

286

REFERENCES

287

Abercrombie, E. D., Keefe, K. A., DiFrischia, D. S., & Zigmond, M. J. (1989). Differential effect of stress on in vivo dopamine release in striatum, nucleus accumbens, and medial frontal cortex. Journal of Neurochemistry, 52(5), 1655-8.

Adams, C. D. (1982). Variations in the sensitivity of instrumental responding to reinforcer devaluation. Quarterly Journal of Experimental Psychology: Comparative and Physiological Psychology, 2, 77-98.

Adams, C. D., & Dickinson, A. (1981). Instrumental responding following reinforcer devaluation. Quarterly Journal of Experimental Psychology: Comparative and Physiological Psychology, 2, 109-121.

Adolphs, R., Cahill, L., Schul, R., & Babinsky, R. (1997). Impaired declarative memory for emotional material following bilateral amygdala damage in humans. Learning and Memory, 4(3), 291-300.

Adolphs, R., Denburg, N. L., & Tranel, D. (2001). The amygdala's role in long-term declarative memory for gist and detail. Behavioral Neuroscience, 115(5), 983-992.

Adolphs, R., Tranel, D., & Denburg, N. (2000). Impaired emotional declarative memory following unilateral amygdala damage. Learning and Memory, 7(3), 180-186.

Akil, H., Campeau, S., Cullinan, W. E., Lechan, R. M., Toni, R., Watson, S. J., & Moore, R. M. (1998). Neuroendocrine Systems I: Overview, Thyroid and Adrenal Axes. In F. E. Bloom, S. C. Landis, J. L. Roberts, L. R. Squire, & M. J. Zigmond (Eds.), Fundamental Neuroscience (pp. 1127-1150). New York: Academic Press.

288

Akirav, I., & Richter-Levin, G. (1999a). Priming stimulation in the basolateral amygdala modulates synaptic plasticity in the rat dentate gyrus. Neuroscience Letters, 270(2), 83-6.

Akirav, I., & Richter Levin, G. (1999b). Biphasic modulation of hippocampal plasticity by behavioral stress and basolateral amygdala stimulation in the rat. Journal of Neuroscience, 19(23), 10530-10535.

Arbel, I., Kadar, T., Silbermann, M., & Levy, A. (1994). The effects of longterm corticosterone administration on hippocampal morphology and cognitive performance of middle-aged rats. Brain Research, 657(1-2), 227-35.

Barbany, G., & Persson, H. (1993). Adrenalectomy attenuates kainic acidelicited increases of messenger RNAs for neurotrophins and their receptors in the rat brain. Neuroscience, 54(4), 909-22.

Barrot, M., Marinelli, M., Abrous, D. N., Rouge-Pont, F., Le Moal, M., & Piazza, P. V. (2000). The dopaminergic hyper-responsiveness of the shell of the nucleus accumbens is hormone-dependent. European Journal of Neuroscience, 12(3), 973-9.

Beatty, P. A., Beatty, W. W., Bowman, R. E., & Gilchrist, J. C. (1970). The effects of ACTH, adrenalectomy and dexamethasone on the acquisition of an avoidance response in rats. Physiology and Behavior, 5(8), 939-944.

Beckwith, B. E., Petros, T. V., Scaglione, C., & Nelson, J. (1986). Dosedependent effects of hydrocortisone on memory in human males. Physiology and Behavior, 36(2), 283-6.

Belanoff, J. K., Gross, K., Yager, A., & Schatzberg, A. F. (2001). Corticosteroids and cognition. Journal of Psychiatric Research, 35(3), 127-45.

289

Bennett, M. C., Diamond, D. M., Fleshner, M., & Rose, G. M. (1991). Serum corticosterone level predicts the magnitude of hippocampal primed burst potentiation and depression in urethane-anesthetized rats. Psychobiology, 19(4), 301-307.

Beylin, A. V., & Shors, T. J. (1999). Stress-induced facilitation of trace eyeblink conditioning is dependent on the presence of glucocorticoids. Eastern Psychological Association, 70, 63.

Blundell, P., Hall, G., & Killcross, S. (2001). Lesions of the basolateral amygdala disrupt selective aspects of reinforcer representation in rats. Journal of Neuroscience, 21(22), 9018-9026.

Bohus, B. (1970). Central nervous structures and the effect of ACTH and corticosteroids on avoidance behavior: a study with intracerebral inplantation of corticosteroids in the rat. Progress in Brain Research, 32, 171-84.

Bohus, B., & Lissak, K. (1968). Adrenocortical Hormones and Avoidance Behaviour of Rats. International Journal of Neuropharmacology, 7(4), 301-306.

Bouton, M. E. (1984). Differential control by context in the inflation and reinstatement paradigms. Journal of Experimental Psychology: Animal Behavior Processes, 10(1), 56-74.

Bouton, M. E. (1991). Context and Retrieval in Extinction and in Other Examples of Interference in Simple Associative Learning. In L. W. Dachowski & C. F. Flaherty (Eds.) (pp. 25-53). Hillsdale, NJ: Erlbaum.

Bouton, M. E. (1994). Conditioning, remembering, and forgetting. Journal of Experimental Psychology: Animal Behavior Processes, 20(3), 219-231.

Bouton, M. E., & Bolles, R. C. (1979). Contextual control of the extinction of conditioned fear. Learning and Motivation, 10(4), 445-466.

290

Bradley, M. M., Greenwald, M. K., Petry, M. C., & Lang, P. J. (1992). Remembering pictures: Pleasure and arousal in memory. Journal of Experimental Psychology: Learning, Memory, and Cognition, 18(2), 379-390.

Bramham, C. R., Southard, T., Ahlers, S. T., & Sarvey, J. M. (1998). Acute cold stress leading to elevated corticosterone neither enhances synaptic efficacy nor impairs LTP in the dentate gyrus of freely moving rats. Brain Research, 789(2), 245255.

Breen, R. A., & McGaugh, J. L. (1961). Facilitation of maze learning with posttrial injections of picrotoxin. Journal of Comparative and Physiological Psychology, 54, 498-501.

Broadbear, J. H., Winger, G., & Woods, J. H. (1999). Glucocorticoidreinforced responding in the rhesus monkey. Psychopharmacology, 147(1), 46-55. Brogden, W. J. (1939). Sensory pre-conditioning. Journal of Experimental Psychology, 25, 323-332.

Brown, T. H., Chapman, P. F., Kairiss, E. W., & Keenan, C. L. (1988). Longterm synaptic potentiation. Science, 242(4879), 724-728.

Buchanan, T. W., & Lovallo, W. R. (2001). Enhanced memory for emotional material

following

stress-level

cortisol

treatment

in

humans.

Psychoneuroendocrinology, 26(3), 307-317.

Burke, A., Heuer, F., & Reisberg, D. (1992). Remembering emotional events. Memory and Cognition, 20(3), 277-90.

Caggiula, A. R., Epstein, L. H., Antelman, S. M., Saylor, S. S., Perkins, K. A., Knopf, S., & Stiller, R. (1991). Conditioned tolerance to the anorectic and corticosterone-elevating effects of nicotine. Pharmacology Biochemistry & Behavior, 40(1), 53-9.

291

Cahill, L. (1997). The neurobiology of emotionally influenced memory. Implications for understanding traumatic memory. Annals of the New York Academy of Sciences, 821, 238-46.

Cahill, L. (1999). A neurobiological perspective on emotionally influenced, long-term memory. Seminars in Clinical Neuropsychiatry, 4(4), 266-73.

Cahill, L. (2000a). Modulation of long-term memory in humans by emotional arousal: adrenergic activation & the amygdala. In J. P. Aggleton (Ed.), The amygdala: A functional analysis (2nd ed., pp. 425–444). London: Oxford University Press.

Cahill, L. (2000b). Neurobiological mechanisms of emotionally influenced, long-term memory. Progress in Brain Research, 126, 29-37.

Cahill, L., Babinsky, R., Markowitsch, H. J., & McGaugh, J. L. (1995). The amygdala and emotional memory. Nature, 377(6547), 295-296.

Cahill, L., Haier, R. J., Fallon, J., Alkire, M. T., Tang, C., Keator, D., Wu, J., & McGaugh, J. L. (1996). Amygdala activity at encoding correlated with long-term, free recall of emotional information. Proceedings of the National Academy of Science U S A, 93(15), 8016-21.

Cahill, L., Haier, R. J., White, N. S., Fallon, J., Kilpatrick, L., Lawrence, C., Potkin, S. G., & Alkire, M. T. (2001). Sex-related differences in amygdala activity during emotionally influenced memory storage. Neurobiology of Learning and Memory, 75(1), 1-9.

Cahill, L., & McGaugh, J. L. (1990). Amygdaloid complex lesions differentially affect retention of tasks using appetitive and aversive reinforcement. Behavioral Neuroscience, 104(4), 532-543.

Cahill, L., & McGaugh, J. L. (1995). A novel demonstration of enhanced memory associated with emotional arousal. Consciousness & Cognition, 4(4), 410-21.

292

Cahill, L., & McGaugh, J. L. (1996). Modulation of memory storage. Current Opinion in Neurobiology, 6(2), 237-42.

Cahill, L., & McGaugh, J. L. (1998). Mechanisms of emotional arousal and lasting declarative memory. Trends in Neurosciences, 21(7), 294-299.

Cahill, L., Prins, B., Weber, M., & McGaugh, J. L. (1994). Beta-Adrenergic activation and memory for emotional events. Nature, 371(6499), 702-704.

Cahill, L., Roozendaal, B., & McGaugh, J. L. (1997). The neurobiology of memory for aversive emotional events. In M. E. Bouton & M. S. Fanselow (Eds.), Learning, motivation, and cognition: The functional behaviorism of Robert C. Bolles. (pp. 369-384). Washington, DC: American Psychological Association.

Cahill, L., Weinberger, N. M., Roozendaal, B., & McGaugh, J. L. (1999). Is the amygdala a locus of "conditioned fear"? Some questions and caveats. Neuron, 23(2), 227-8.

Canli, T., Zhao, Z., Brewer, J., Gabrieli, J. D. E., & Cahill, L. (2000). Eventrelated activation in the human amygdala associates with later memory for individual emotional response. Journal of Neuroscience, 20(19).

Carroll, B. J., Curtis, G. C., & Mendels, J. (1976). Neuroendocrine regulation in depression: II. Discrimination of depressed from nondepressed patients. Archives of General Psychiatry, 33(9), 1051-1058.

Christianson, S. A., & Loftus, E. F. (1991). Remembering emotional events: The fate of detailed information. Cognition and Emotion, 5(2), 81-108.

Coe, C. L., Stanton, M. E., & Levine, S. (1983). Adrenal responses to reinforcement and extinction: Role of expectancy versus instrumental responding. Behavioral Neuroscience, 97(4), 654-657.

293

Conrad, C. D., Lupien, S. J., & McEwen, B. S. (1996). Glucocorticoid (GC), but not mineralocorticoid (MC), receptor agonist and antagonist block spatial memory performance on the Y-maze. Society for Neuroscience Abstracts, 22(2), 809.22.

Conrad, C. D., Lupien, S. J., & McEwen, B. S. (1999). Support for a bimodal role for Type II adrenal steroid receptors in spatial memory. Neurobiology of Learning and Memory, 72(1), 39-46.

Cools, A. R., & Gingras, M. A. (1998). Nijmegen high and low responders to novelty: A new tool in the search after the neurobiology of drug abuse liability. Pharmacology, Biochemistry and Behavior, 60(1), 151-159.

Cools, A. R., Rots, N. Y., Ellenbroek, B., & de Kloet, E. R. (1993). Bimodal shape of individual variation in behavior of Wistar rats: The overall outcome of a fundamentally different make-up and reactivity of the brain, the endocrinological and the immunological system. Neuropsychobiology, 28(1-2), 100-105.

Coover, G. D., Goldman, L., & Levine, S. (1971). Plasma corticosterone increases produced by extinction of operant behavior in rats. Physiology and Behavior, 6(3), 261-263.

Coover, G. D., Murison, R., Sundberg, H., Jellestad, F., & Ursin, H. (1984). Plasma corticosterone and meal expectancy in rats: effects of low probability cues. Physiology and Behavior, 33(2), 179-84.

Coover, G. D., Sutton, B. R., Welle, S. L., & Hart, R. P. (1978). Corticosterone responses, hurdle-jump acquisition, and the effects of dexamethasone using classical conditioning of fear. Hormones and Behavior, 11(3), 279-294.

Cordero, M. I., Merino, J. J., & Sandi, C. (1998). Correlational relationship between shock intensity and corticosterone secretion on the establishment and subsequent expression of contextual fear conditioning. Behavioral Neuroscience, 112(4), 885-891.

294

Cordero, M. I., & Sandi, C. (1998). A role for brain glucocorticoid receptors in contextual fear conditioning: Dependence upon training intensity. Brain Research, 786(1-2), 11-17.

Corodimas, K. P., LeDoux, J. E., Gold, P. W., & Schulkin, J. (1994). Corticosterone potentiation of conditioned fear in rats. Annals of the New York Academy of Sciences, 746, 392-3.

Costa-Miserachs, D., Portell-Cortes, I., Aldavert-Vera, L., Torras-Garcia, M., & Morgado-Bernal, I. (1993). Facilitation of a distributed shuttlebox conditioning with post-training epinephrine in rats. Behavioral & Neural Biology, 60(1), 75-8.

Costa-Miserachs, D., Portell-Cortes, I., Aldavert-Vera, L., Torras-Garcia, M., & Morgado-Bernal, I. (1994). Long-term memory facilitation in rats by posttraining epinephrine. Behavioral Neuroscience, 108(3), 469-74.

Cottrell, G. A., & Nakajima, S. (1977). Effect of corticosteroids in the hippocampus on passive avoidance behavior in the rat. Pharmacology, Biochemistry and Behavior, 7(3), 277-280.

Cremer, H., Lange, R., Christoph, A., Plomann, M., Vopper, G., Roes, J., Brown, R., Baldwin, S., Kraemer, P., Scheff, S., Barthels, D., Rajewsky, K., & Wille, W. (1994). Inactivation of the N-CAM gene in mice results in size reduction of the olfactory bulb and deficits in spatial learning. Nature, 367(6462), 455-459.

Dachir, S., Kadar, T., Robinzon, B., & Levy, A. (1993). Cognitive deficits induced in young rats by long-term corticosterone administration. Behavioral & Neural Biology, 60(2), 103-9.

Dallman, M. F., Akana, F. S., Scribner, K. A., Bradburt, Y. M. J., Walker, C.D., Strack, A. M., & Cascio, C. S. (1991). Stress, feedback and facilitation in the hypothalamic-pituitary-adrenal axis. Journal of Neuroendocrinology, 5, 517-526.

295

Dantzer, R., Arnone, M., & Mormede, P. (1980). Effects of frustration on behaviour and plasma corticosteroid levels in pigs. Physiology and Behavior, 24(1), 1-4.

Davis, M. (1992). The role of the amygdala in fear and anxiety. Annual Review of Neuroscience, 15, 353-375.

Dayas, C. V., Buller, K. M., & Day, T. A. (1999). Neuroendocrine responses to an emotional stressor: evidence for involvement of the medial but not the central amygdala. European Journal of Neuroscience, 11(7), 2312-22.

de Boer, S. F., de Beun, R., Slangen, J. L., & Van der Gugten, J. (1990). Dynamics of plasma catecholamine and corticosterone concentrations during reinforced and extinguished operant behavior in rats. Physiology and Behavior, 47(4), 691-698.

de Kloet, E. R. (1991). Brain Corticosteroid Receptor Balance and Homeostatic Control. Frontiers in Neuroendocrinology, 12(2), 95-164.

de Kloet, E. R. (1997). Why Dexamethasone Poorly Penetrates in Brain. Stress, 2(1), 13-20.

de Kloet, E. R., Oitzl, M. S., & Joels, M. (1999). Stress and cognition: Are corticosteroids good or bad guys? Trends in Neurosciences, 22(10), 422-426.

de Kloet, E. R., Wallach, G., & McEwen, B. S. (1975). Differences in corticosterone and dexamethasone binding to rat brain and pituitary. Endocrinology, 96(3), 598-609.

de Quervain, D. J. F., Roozendaal, B., & McGaugh, J. L. (1998). Stress and glucocorticoids impair retrieval of long-term spatial memory. Nature, 394(6695), 787790.

296

de Quervain, D. J. F., Roozendaal, B., Nitsch, R. M., McGaugh, J. L., & Hock, C. (2000). Acute cortisone administration impairs retrieval of long-term declarative memory in humans. Nature Neuroscience, 3(4), 313-314.

de Wied, D. (1967). Opposite effects of ACTH and glucocorticoids on extinction of conditioned avoidance behavior. In L. Martini, F. Fraschini, & M. Motta (Eds.), Proc. Second Intern. Congr. Hormonal Steroids, Milan, May 1966, Excerpta Medica, Intern. Congr. Ser. No. 132 (pp. 945-951). Amsterdam: Excerpta Medica Foundation.

Delamater, A. R. (1996). Effects of several extinction treatments upon the integrity of Pavlovian stimulus-outcome associations. Animal Learning and Behavior, 24(4), 437-449.

Dellu, F., Mayo, W., Piazza, P. V., Le Moal, M., & Simon, H. (1993). Individual differences in behavioral responses to novelty in rats: Possible relationship with the sensation-seeking trait in man. Personality and Individual Differences, 15(4), 411-418.

Dellu, F., Mayo, W., Vallee, M., Maccari, S., Piazza, P. V., Le Moal, M., & Simon, H. (1996b). Behavioral reactivity to novelty during youth as a predictive factor of stress-induced corticosterone secretion in the elderly--a life-span study in rats. Psychoneuroendocrinology, 21(5), 441-53.

Dellu, F., Piazza, P. V., Mayo, W., Le Moal, M., & Simon, H. (1996a). Novelty-seeking in rats: Biobehavioural characteristics and possible relationship with the sensation-seeking trait in man. Neuropsychobiology, 34(3), 136-145.

Deminiere, J. M., Piazza, P. V., Le Moal, M., & Simon, H. (1989). Experimental approach to individual vulnerability to psychostimulant addiction. Neuroscience and Biobehavioral Reviews, 13(2-3), 141-147.

Deroche, V., Piazza, P. V., Deminiere, J. M., Le Moal, M., & Simon, H. (1993). Rats orally self-administer corticosterone. Brain Research, 622(1-2), 315-20. 297

Diamond, D. M., Bennett, M. C., Engstrom, D. A., Fleshner, M., & Rose, G. M. (1989). Adrenalectomy reduces the threshold for hippocampal primed burst potentiation in the anesthetized rat. Brain Research, 492(1-2), 356-360.

Diamond, D. M., Bennett, M. C., Fleshner, M., & Rose, G. M. (1992). Inverted-U relationship between the level of peripheral corticosterone and the magnitude of hippocampal primed burst potentiation. Hippocampus, 2(4), 421-30.

Diamond, D. M., Fleshner, M., Ingersoll, N., & Rose, G. (1996). Psychological

stress

impairs

spatial

working

memory:

Relevance

to

electrophysiological studies of hippocampal function. Behavioral Neuroscience, 110(4), 661-672.

Diamond, D. M., Fleshner, M., & Rose, G. M. (1994). Psychological stress repeatedly blocks hippocampal primed burst potentiation in behaving rats. Behavioural Brain Research, 62(1), 1-9.

Diamond, D. M., & Park, C. R. (2000). Predator exposure produces retrograde amnesia and blocks synaptic plasticity. Progress toward understanding how the hippocampus is affected by stress. Annals of the New York Academy of Sciences, 911, 453-5.

Diamond, D. M., Park, C. R., Heman, K. L., & Rose, G. M. (1999). Exposing rats to a predator impairs spatial working memory in the radial arm water maze. Hippocampus, 9(5), 542-552.

Dickinson, A. (1985). Actions and habits: the development of behavioural autonomy. Philosophical Transactions Royal Society London, B 308, 67-78.

Dixon, R. B., & Christy, N. P. (1980). On the various forms of corticosteroid withdrawal syndrome. American Journal of Medicine, 68(2), 224-30.

298

Doyle, P., Rohner-Jeanrenaud, F., & Jeanrenaud, B. (1993). Local cerebral glucose utilization in brains of lean and genetically obese (fa/fa) rats. American Journal of Physiology, 264(1 Pt 1), E29-36.

Ellenbroek, B. A., Knobbout, D. A., & Cools, A. R. (1997). The role of mesolimbic and nigrostriatal dopamine in latent inhibition as measured with the conditioned taste aversion paradigm. Psychopharmacology, 129(2), 112-120.

Everitt,

B.

J.,

Dickinson,

A.,

&

Robbins,

T.

W.

(2001).

The

neuropsychological basis of addictive behaviour. Brain Research Reviews, 36(2-3), 129-38.

Everitt, B. J., & Robbins, T. W. (1992). Amygdala-Ventral Striatal Interactions and Reward-Related Processes. In J. Aggleton (Ed.), The Amygdala: Neurobiological Aspects of Emotion, Memory, and Mental Dysfunction (pp. 401-429). New York: Wiley-Liss.

Fanselow, M. S. (1990). Factors governing one-trial contextual conditioning. Animal Learning and Behavior, 18(3), 264-270.

Fanselow, M. S., & LeDoux, J. E. (1999). Why we think plasticity underlying Pavlovian fear conditioning occurs in the basolateral amygdala. Neuron, 23(2), 22932.

Fehm-Wolfsdorf, G., Reutter, K., Zenz, H., Born, J., & Fehm, H. L. (1993). Are circadian variations in taste thresholds cortisol-dependent? Journal of Psychophysiology, 7(1), 65-72.

Ferry, B., & McGaugh, J. L. (1999). Clenbuterol administration into the basolateral amygdala post-training enhances retention in an inhibitory avoidance task. Neurobiology of Learning and Memory, 72(1), 8-12.

299

Ferry, B., Roozendaal, B., & McGaugh, J. L. (1999a). Basolateral amygdala noradrenergic influences on memory storage are mediated by an interaction between beta- and alpha-adrenoceptors. Journal of Neuroscience, 19(12), 5119-5123.

Ferry, B., Roozendaal, B., & McGaugh, J. L. (1999b). Role of norepinephrine in mediating stress hormone regulation of long-term memory storage: A critical involvement of the amygdala. Biological Psychiatry, 46(9), 1140-1152.

Ferry, B., Roozendaal, B., & McGaugh, J. L. (1999c). Involvement of alpha1adrenoceptors in the basolateral amygdala in modulation of memory storage. European Journal of Pharmacology, 372(1), 9-16.

Fleshner, M., Pugh, C. R., Tremblay, D., & Rudy, J. W. (1997). DHEA-S selectively impairs contextual-fear conditioning: Support for the antiglucocorticoid hypothesis. Behavioral Neuroscience, 111(3), 512-517.

Flood, J. F., Vidal, D., Bennett, E. L., Orme, A. E., Vasquez, S., & Jarvik, M. E. (1978). Memory facilitating and anti-amnesic effects of corticosteroids. Pharmacology, Biochemistry and Behavior, 8(1), 81-87.

Foy, M. R., Stanton, M. E., Levine, S., & Thompson, R. F. (1987). Behavioral stress impairs long-term potentiation in rodent hippocampus. Behavioral & Neural Biology, 48(1), 138-49.

Fuchs, E., Fluegge, G., Ohl, F., Lucassen, P., Vollmann Honsdorf, G. K., & Michaelis, T. (2001). Psychosocial stress, glucocorticoids, and structural alterations in the tree shrew hippocampus. Physiology and Behavior, 73(3), 285-291.

Fuller, R. W., & Snoddy, H. D. (1981). Elevation of serum corticosterone concentrations in rats by pergolide and other dopamine agonists. Endocrinology, 109(4), 1026-32.

300

Gallagher, M., Kapp, B. S., Musty, R. E., & Driscoll, P. A. (1977). Memory formation: evidence for a specific neurochemical system in the amygdala. Science, 198(4315), 423-5.

Galvez, R., Mesches, M. H., & McGaugh, J. L. (1996). Norepinephrine release in the amygdala in response to footshock stimulation. Neurobiology of Learning and Memory, 66(3), 253-257.

Garrud, P., Gray, J. A., & de Wied, D. (1974). Pituitary-adrenal hormones and extinction of rewarded behaviour in the rat. Physiology and Behavior, 12(1), 109-119.

Garrud, P., Gray, J. A., Rickwood, L., & Coen, C. (1977). Pituitary-adrenal hormones and effects of partial reinforcement on appetitive behaviour in the rat. Physiology and Behavior, 18(5), 813-818.

Garvey, W. T., Huecksteadt, T. P., Lima, F. B., & Birnbaum, M. J. (1989). Expression of a glucose transporter gene cloned from brain in cellular models of insulin resistance: dexamethasone decreases transporter mRNA in primary cultured adipocytes. Molecular Endocrinology, 3(7), 1132-41.

Gold, P. E. (1989). Neurobiological features common to memory modulation by many treatments. Animal Learning and Behavior, 17(1), 94-100.

Gold, P. E., & McCarty, R. (1981). Plasma catecholamines: Changes after footshock and seizure-producing frontal cortex stimulation. Behavioral and Neural Biology, 31(3), 247-260.

Gold, P. E., & McGaugh, J. L. (1975). A single-trace, two-process view of memory storage processes. In D. Deutsch & J. A. Deutsch (Eds.), Short-Term Memory (pp. 355-378). New York: Academic Press.

Gold, P. E., & van Buskirk, R. B. (1975). Facilitation of time-dependent memory processes with posttrial epinephrine injections. Behavioral Biology, 13(2), 145-53. 301

Gold, P. E., & van Buskirk, R. B. (1976). Effects of posttrial hormone injections on memory processes. Hormones and Behavior, 7(4), 509-517.

Gold, P. E., & van Buskirk, R. B. (1978). Posttraining brain norepinephrine concentrations: correlation with retention performance of avoidance training and with peripheral epinephrine modulation of memory processing. Behavioral Biology, 23(4), 509-20.

Goldman, L., Coover, G. D., & Levine, S. (1973). Bidirectional effects of reinforcement shifts on pituitary adrenal activity. Physiology and Behavior, 10(2), 209-214.

Greendale, G. A., Kritz-Silverstein, D., Seeman, T., & Barrett-Connor, E. (2000). Higher basal cortisol predicts verbal memory loss in postmenopausal women: Rancho Bernardo Study. Journal of the American Geriatric Society, 48(12), 1655-8.

Groenewegen, H. J., Berendse, H. W., Meredith, G. E., Haber, S. N., Voorn, P., Wolters, J. G., & M., L. A. H. (1991). Functional Anatomy of the Ventral, Limbic System-Innervated Striatum. In P. Willner & J. Scheel-Krueger (Eds.), The Mesolimbic Dopamine System: From Motivation to Action (pp. 19-59): John Wiley & Sons Ltd.

Hall, G., & Channell, S. (1985). Differential effects of contextual change on latent inhibition and on the habituation of an orienting response. Journal of Experimental Psychology: Animal Behavior Processes, 11(3), 470-481.

Hall, G., & Honey, R. C. (1989). Contextual effects in conditioning, latent inhibition, and habituation: Associative and retrieval functions of contextual cues. Journal of Experimental Psychology: Animal Behavior Processes, 15(3), 232-241.

Hall, G., & Schachtman, T. R. (1987). Differential effects of a retention interval on latent inhibition and the habituation of an orienting response. Animal Learning and Behavior, 15(1), 76-82.

302

Hamann, S. B., Cahill, L., McGaugh, J. L., & Squire, L. R. (1997). Intact enhancement of declarative memory for emotional material in amnesia. Learning and Memory, 4(3), 301-309. Hamann, S. B., Cahill, L., & Squire, L. R. (1997). Emotional perception and memory in amnesia. Neuropsychology, 11(1), 104-113.

Hamann, S. B., Ely, T. D., Grafton, S. T., & Kilts, C. D. (1999). Amygdala activity related to enhanced memory for pleasant and aversive stimuli. Nature Neuroscience, 2(3), 289-293.

Hamann, S. B., & Mao, H. (2000). Emotional memory and reactions to verbal and nonverbal emotional stimuli: a functional MRI study. Society for Neuroscience Abstracts, 26(1), 403.3.

Hamann, S. B., Monarch, E. S., & Goldstein, F. C. (2000). Memory enhancement for emotional stimuli is impaired in early Alzheimer's disease. Neuropsychology, 14(1), 82-92.

Hatfield, T., & McGaugh, J. L. (1999). Norepinephrine infused into the basolateral amygdala posttraining enhances retention in a spatial water maze task. Neurobiology of Learning and Memory, 71(2), 232-239.

Haynes, L. E., Griffiths, M. R., Hyde, R. E., Barber, D. J., & Mitchell, I. J. (2001). Dexamethasone induces limited apoptosis and extensive sublethal damage to specific subregions of the striatum and hippocampus: implications for mood disorders. Neuroscience, 104(1), 57-69.

Hebb, D. O. (1949). The organization of behavior: a neuropsychological theory. New York: Science Editions.

Hennessy, J. W., Cohen, M. E., & Rosen, A. J. (1973). Adrenocortical influences upon the extinction of an appetitive runway response. Physiology and Behavior, 11(6), 767-770.

303

Heuer, F., & Reisberg, D. (1990). Vivid memories of emotional events: the accuracy of remembered minutiae. Memory and Cognition, 18(5), 496-506.

Holsboer, F. (2001). Prospects for antidepressant drug discovery. Biological Psychology, 57(1-3), 47-65.

Honey, R. C., & Good, M. (1993). Selective hippocampal lesions abolish the contextual specificity of latent inhibition and conditioning. Behavioral Neuroscience, 107(1), 23-33.

Honma, K. I., Honma, S., & Hiroshige, T. (1984). Feeding-associated corticosterone peak in rats under various feeding cycles. American Journal of Physiology, 246(5 Pt 2), R721-6.

Horner, H. C., Munck, A., & Lienhard, G. E. (1987). Dexamethasone causes translocation of glucose transporters from the plasma membrane to an intracellular site in human fibroblasts. Journal of Biological Chemistry, 262(36), 17696-702.

Ikeda, M., Mori, E., Hirono, N., Imamura, T., Shimomura, T., Ikejiri, Y., & Yamashita, H. (1998). Amnestic people with Alzheimer's disease who remembered the Kobe earthquake. British Journal of Psychiatry, 172, 425-8.

Ikegaya, Y., Saito, H., & Abe, K. (1994). Attenuated hippocampal long-term potentiation in basolateral amygdala-lesioned rats. Brain Research, 656(1), 157-64.

Ikegaya, Y., Saito, H., & Abe, K. (1995). High-frequency stimulation of the basolateral amygdala facilitates the induction of long-term potentiation in the dentate gyrus in vivo. Neuroscience Research, 22(2), 203-7.

Imperato, A., Puglisi-Allegra, S., Casolini, P., & Angelucci, L. (1991). Changes in brain dopamine and acetylcholine release during and following stress are independent of the pituitary-adrenocortical axis. Brain Research, 538(1), 111-7.

304

Imperato, A., Puglisi-Allegra, S., Grazia Scrocco, M., Casolini, P., Bacchi, S., & Angelucci, L. (1992). Cortical and limbic dopamine and acetylcholine release as neurochemical correlates of emotional arousal in both aversive and non-aversive environmental changes. Neurochemistry International, 20(Supplement), 265S-270S.

Introini-Collison, I. B., Miyazaki, B., & McGaugh, J. L. (1991). Involvement of

the

amygdala

in

the

memory-enhancing

effects

of

clenbuterol.

Psychopharmacology, 104(4), 541-4.

Introini-Collison, I. B., Saghafi, D., Novack, G. D., & McGaugh, J. L. (1992). Memory-enhancing effects of post-training dipivefrin and epinephrine: involvement of peripheral and central adrenergic receptors. Brain Research, 572(1-2), 81-6.

Iuvone, P. M., Morasco, J., & Dunn, A. J. (1977). Effect of corticosterone on the synthesis of [3H]catecholamines in the brains of CD-1 mice. Brain Research, 120(3), 571-6.

James, W. (1890). The Principles of Psychology (Vol. 1). New York: Dover Publications, Inc.

Jefferys, D., Copolov, D., Irby, D., & Funder, J. (1983). Behavioural effect of adrenalectomy: reversal by glucocorticoids or [D-Ala2,Met5]enkephalinamide. European Journal of Pharmacology, 92(1-2), 99-103.

Jefferys, D., & Funder, J. W. (1987). Glucocorticoids, adrenal medullary opioids, and the retention of a behavioral response after stress. Endocrinology, 121(3), 1006-9.

Joels, M., & de Kloet, E. R. (1989). Effects of glucocorticoids and norepinephrine on the excitability in the hippocampus. Science, 245(4925), 1502-5.

Joels, M., & de Kloet, E. R. (1994). Mineralocorticoid and glucocorticoid receptors in the brain. Implications for ion permeability and transmitter systems. Progress in Neurobiology, 43(1), 1-36 305

Joseph, M. H., Peters, S. L., Moran, P. M., Grigoryan, G. A., Young, A. M., & Gray, J. A. (2000). Modulation of latent inhibition in the rat by altered dopamine transmission in the nucleus accumbens at the time of conditioning. Neuroscience, 101(4), 921-30.

Kabbaji, M., Devine, D. P., Savage, V. R., & Akil, H. (2000). Neurobiological correlates of individual differences in novelty-seeking behavior in the rat: Differential expression of stress-related molecules. Journal of Neuroscience, 20(18), 6983-6988.

Kalat, J. W. (1977). Status of "learned safety" or "learned noncorrelation" as a mechanism in taste aversion learning. In L. M. Barker, M. R. Best, & M. Domjan (Eds.), Learning mechanisms in food selection (pp. 273-293). Waco, TX: Baylor University Press.

Kalat, J. W. (1995). Biological Psychology (5th ed.). Belmont, CA: Brooks/Cole Publ.

Kalimi, M., Shafagoj, Y., Loria, R., Padgett, D., & Regelson, W. (1994). Antiglucocorticoid effects of dehydroepiandrosterone (DHEA). Journal of Molecular and Cellular Biochemistry, 131(2), 99-104.

Kaneto, H. (1997). Learning/memory processes under stress conditions. Behavioural Brain Research, 83(1-2), 71-74.

Kasprow, W. J., Catterson, D., Schachtman, T. R., & Miller, R. R. (1984). Attenuation of latent inhibition by post-acquisition reminder. Quarterly Journal of Experimental Psychology: Comparative and Physiological Psychology, 1, 53-63.

Kazui, H., Mori, E., Hashimoto, M., Hirono, N., Imamura, T., Tanimukai, S., Hanihara, T., & Cahill, L. (2000). Impact of emotion on memory: Controlled study of the influence of emotionally charged material on declarative memory in Alzheimer's disease. British Journal of Psychiatry, 177, 343-347.

306

Kiernan, M. J., & Westbrook, R. F. (1993). Effects of exposure to a to-beshocked environment upon the rat's freezing response: evidence for facilitation, latent inhibition, and perceptual learning. Quarterly Journal of Experimental Psychology B, 46(3), 271-88.

Killcross, S. (2001). Loss of latent inhibition in conditioned taste aversion following exposure to a novel flavour before test. Quarterly Journal of Experimental Psychology: Comparative and Physiological Psychology, 3, 271-288.

Killcross, S., & Balleine, B. (1996). Role of primary motivation in stimulus preexposure effects. Journal of Experimental Psychology: Animal Behavior Processes, 22(1), 32-42.

Killcross, S., & Blundell, P. (in press). Associative Representations of emotionally significant outcomes. In S. Moore & M. Oaksford (Eds.), Emotional Cognition . Amsterdam & Philadelphia: John Benjamins.

Killcross, S., & Coutureau, E. (2001). The role of the medial prefrontal cortex in actions and habits. Society for Neuroscience Abstracts, 27, 533.9.

Killcross, S., Dickinson, A., & Robbins, T. W. (1994a). Amphetamineinduced disruptions of latent inhibition are reinforcer mediated: implications for animal models of schizophrenic attentional dysfunction. Psychopharmacology, 115(12), 185-95.

Killcross, S., Dickinson, A., & Robbins, T. W. (1994b). Effects of the neuroleptic alpha-flupenthixol on latent inhibition in aversively- and appetitivelymotivated

paradigms:

evidence

for

dopamine-reinforcer

interactions.

Psychopharmacology, 115(1-2), 196-205.

Killcross, S., & Robbins, T. W. (1993). Differential effects of intra-accumbens and systemic amphetamine on latent inhibition using an on-baseline, within-subject conditioned suppression paradigm. Psychopharmacology, 110(4), 479-489.

307

Killcross, S., Robbins, T. W., & Everitt, B. J. (1997). Different types of fearconditioned behaviour mediated by separate nuclei within amygdala. Nature, 388(6640), 377-380. Kilpatrick, L., & Cahill, L. (2001). Gender-related differences in explicit memory for emotional events: relation to lateralized amygdala processing. Society for Neuroscience Abstracts, 27(1), 72.3.

Kim, J. J., Lee, H. J., Han, J. S., & Packard, M. G. (2001). Amygdala is critical for stress-induced modulation of hippocampal long-term potentiation and learning. Journal of Neuroscience, 21(14), 5222-5238.

Kirschbaum, C., Wolf, O. T., May, M., Wippich, W., & Hellhammer, D. H. (1996). Stress- and treatment-induced elevations of cortisol levels associated with impaired declarative memory in healthy adults. Life Sciences, 58(17), 1475-1483.

Konorski, J. (1967). Integrative activity of the brain. An interdisciplinary approach. Chicago & London: The University of Chicago Press.

Korte, S. M. (2001). Corticosteroids in relation to fear, anxiety and psychopathology. Neuroscience and Biobehavioral Reviews, 25(2), 117-142.

Kovacs, G. L., Telegdy, G., & Lissak, K. (1976). 5-hydroxytryptamine and the mediation of pituitary-adrenocortical hormones in the extinction of active avoidance behaviour. Psychoneuroendocrinology, 1(3), 219-230.

Kovacs, G. L., Telegdy, G., & Lissak, K. (1977). Dose-dependent action of corticosteroids on brain serotonin content and passive avoidance behavior. Hormones and Behavior, 8(2), 155-165.

Kraemer, P. J., & Roberts, W. A. (1984). The influence of flavor preexposure and test interval on conditioned taste aversions in the rat. Learning and Motivation, 15(3), 259-278.

308

Kraemer, P. J., & Spear, N. E. (1992). The effect of nonreinforced stimulus exposure on the strength of a conditioned taste aversion as a function of retention interval: Do latent inhibition and extinction involve a shared process? Animal Learning and Behavior, 20(1), 1-7.

Krieger, D. T., & Hauser, H. (1978). Comparison of synchronization of circadian corticosteroid rhythms by photoperiod and food. Proceedings of the National Academy of Science U S A, 75(3), 1577-81.

Kruse, J. M., Overmier, J. B., Konz, W. A., & Rokke, E. (1983). Pavlovian conditioned stimulus effects upon instrumental choice behavior are reinforcer specific. Learning and Motivation, 14(2), 165-181.

Kwak, S. P., Young, E. A., Morano, I., Watson, S. J., & Akil, H. (1992). Diurnal corticotropin-releasing hormone mRNA variation in the hypothalamus exhibits a rhythm distinct from that of plasma corticosterone. Neuroendocrinology, 55(1), 74-83.

LaBar, K. S., & Phelps, E. A. (1998). Arousal-mediated memory consolidation: Role of the medial temporal lobe in humans. Psychological Science, 9(6), 490-493.

LeDoux, J. E. (1994). Emotion, memory and the brain. Scientific American, 270(6), 50-7.

LeDoux, J. E. (1995). Emotion: Clues from the brain. Annual Review of Psychology, 46, 209-235.

Levine, G. (1991). A Guide to SPSS for Analysis of Variance. Hillsdale, NJ: Lawrence Erlbaum Associates, Publishers.

Levine, S., & Coover, G. D. (1976). Environmental control of suppression of the pituitary-adrenal system. Physiology and Behavior, 17(1), 35-37.

309

Liang, K. C., Chen, L. L., & Huang, T. E. (1995). The role of amygdala norepinephrine in memory formation: involvement in the memory enhancing effect of peripheral epinephrine. Chinese Journal of Physiology, 38(2), 81-91.

Liang, K. C., Juler, R. G., & McGaugh, J. L. (1986). Modulating effects of posttraining epinephrine on memory: Involvement of the amygdala noradrenergic system. Brain Research, 368(1), 125-133.

Liang, K. C., & McGaugh, J. L. (1983a). Lesions of the stria terminalis attenuate the amnestic effect of amygdaloid stimulation on avoidance responses. Brain Research, 274(2), 309-18.

Liang, K. C., & McGaugh, J. L. (1983b). Lesions of the stria terminalis attenuate the enhancing effect of post-training epinephrine on retention of an inhibitory avoidance response. Behavioural Brain Research, 9(1), 49-58.

Liang, K. C., McGaugh, J. L., & Yao, H. Y. (1990). Involvement of amygdala pathways in the influence of post-training intra-amygdala norepinephrine and peripheral epinephrine on memory storage. Brain Research, 508(2), 225-33.

Liu, L., Tsuji, M., Takeda, H., Takada, K., & Matsumiya, T. (1999). Adrenocortical suppression blocks the enhancement of memory storage produced by exposure to psychological stress in rats. Brain Research, 821(1), 134-140.

Loscertales, M., Rose, S. P. R., & Sandi, C. (1997). The corticosteroid synthesis inhibitors metyrapone and aminoglutethimide impair long-term memory for a passive avoidance task in day-old chicks. Brain Research, 769(2), 357-361.

Lovely, R. H., Pagano, R. R., & Paolino, R. M. (1972). Shuttle-box-avoidance performance and basal corticosterone levels as a function of duration of individual housing in rats. Journal of Comparative and Physiological Psychology, 81(2), 331335.

310

Lubow, R. E. (1973). Latent inhibition. Psychological Bulletin, 79(6), 398407.

Lubow, R. E., & Moore, A. U. (1959). Latent inhibition: The effect of nonreinforced pre-exposure to the conditional stimulus. Journal of Comparative and Physiological Psychology, 52, 415-419.

Lubow, R. E., Weiner, I., & Schnur, P. (1981). Conditioned attention theory. In G. H. Bower (Ed.), The psychology of learning and motivation (Vol. 15, pp. 1-49). New York: Academic Press.

Luine, V. N., Spencer, R. L., & McEwen, B. S. (1993). Effects of chronic corticosterone ingestion on spatial memory performance and hippocampal serotonergic function. Brain Research, 616(1-2), 65-70.

Luine, V. N., Villegas, M., Martinez, C., & McEwen, B. S. (1994). Repeated stress causes reversible impairments of spatial memory performance. Brain Research, 639(1), 167-70.

Lupien, S., Lecours, A. R., Lussier, I., Schwartz, G., Nair, N. P. V., & Meaney, M. J. (1994). Basal cortisol levels and cognitive deficits in human aging. Journal of Neuroscience, 14(5, Pt 1), 2893-2903.

Lupien, S. J., Gillin, C. J., & Hauger, R. L. (1999). Working memory is more sensitive than declarative memory to the acute effects of corticosteroids: A doseresponse study in humans. Behavioral Neuroscience, 113(3), 420-430.

Lupien, S. J., & McEwen, B. S. (1997). The acute effects of corticosteroids on cognition: Integration of animal and human model studies. Brain Research Reviews, 24(1), 1-27.

Lupien, S. L., Gillin, C., Frakes, D., Soefje, S., & Hauger, R. L. (1995). Delayed but not immediate effects of a 100 minutes hydrocortisone infusion on

311

declarative memory performance in young normal adults. International Society Psychoneuroendocrinology Abstracts, 25.

Mantsch, J. R., Saphier, D., & Goeders, N. E. (1998). Corticosterone facilitates the acquisition of cocaine self-administration in rats: opposite effects of the type II glucocorticoid receptor agonist dexamethasone. Journal of Pharmacology and Experimental Therapeutics, 287(1), 72-80.

Maren, S., & Fanselow, M. S. (1996). The amygdala and fear conditioning: has the nut been cracked? Neuron, 16(2), 237-40.

Markowska, A. L., Koliatsos, V. E., Breckler, S. J., Price, D. L., & Olton, D. S. (1994). Human nerve growth factor improves spatial memory in aged but not in young rats. Journal of Neuroscience, 14(8), 4815-4824.

McCarty, R., & Gold, P. E. (1981). Plasma catecholamines: Effects of footshock level and hormonal modulators of memory storage. Hormones and Behavior, 15(2), 168-182.

McCormick, C. M., McNamara, M., Mukhopadhyay, S., & Kelsey, J. E. (1997). Acute corticosterone replacement reinstates performance on spatial and nonspatial memory tasks 3 months after adrenalectomy despite degeneration in the dentate gyrus. Behavioral Neuroscience, 111(3), 518-531.

McEwen, B. S. (1987). Glucocorticoid-biogenic amine interactions in relation to mood and behavior. Biochemistry & Pharmacology, 36(11), 1755-63. McEwen, B. S., & Sapolsky, R. M. (1995). Stress and cognitive function. Current Opinion in Neurobiology, 5(2), 205-16.

McGaugh, J. L. (1966). Time-dependent processes in memory storage. Science, 153(3742), 1351-1358.

McGaugh, J. L. (1973). Drug facilitation of learning and memory. Annual Review Pharmacology, 13, 229-41. 312

McGaugh, J. L. (2000). Memory: A century of consolidation. Science, 287(5451), 248-251.

McGaugh, J. L., & Cahill, L. (1997). Interaction of neuromodulatory systems in modulating memory storage. Behavioural Brain Research, 83(1-2), 31-38.

McGaugh, J. L., Cahill, L., & Roozendaal, B. (1996). Involvement of the amygdala in memory storage: interaction with other brain systems. Proceedings of the National Academy of Science U S A, 93(24), 13508-14.

McGaugh, J. L., Ferry, B., Vazdarjanova, A., & Roozendaal, B. (2000). Amygdala: Role in modulation of memory storage. In J. P. Aggleton (Ed.), The amygdala: a functional analysis (2nd ed., pp. 391-423). London: Oxford University Press.

Merali, Z., McIntosh, J., Kent, P., Michaud, D., & Anisman, H. (1998). Aversive and appetitive events evoke the release of corticotropin-releasing hormone and bombesin-like peptides at the central nucleus of the amygdala. Journal of Neuroscience, 18(12), 4758-4766.

Micco, D. J., McEwen, B. S., & Shein, W. (1979). Modulation of behavioral inhibition in appetitive extinction following manipulation of adrenal steroids in rats: Implications for involvement of the hippocampus. Journal of Comparative and Physiological Psychology, 93(2), 323-329.

Micheau, J., Destrade, C., & Soumireu Mourat, B. (1981). Intraventricular corticosterone injection facilitates memory of an appetitive discriminative task in mice. Behavioral and Neural Biology, 31(1), 100-104.

Micheau, J., Destrade, C., & Soumireu-Mourat, B. (1982). [Posttrial injections of corticosterone in dorsal hippocampus of the BALB/c mouse facilitate extinction of appetitive operant conditioning in the Skinner box]. Comptes rendus seances de l'academie des sciences III, 294(23), 1109-12.

313

Micheau, J., Destrade, C., & Soumireu-Mourat, B. (1985). Time-dependent effects of posttraining intrahippocampal injections of corticosterone on retention of appetitive learning tasks in mice. European Journal of Pharmacology, 106(1), 39-46.

Miller, R. R., Kasprow, W., J., & Schachtman, T., R. (1986). Retrieval variability: Sources and consequences. American Journal of Psychology, 99, 145-218.

Mitchell, J. B., & Meaney, M. J. (1991). Effects of corticosterone on response consolidation and retrieval in the forced swim test. Behavioral Neuroscience, 105(6), 798-803.

Moayeri, S. E., Cahill, L., Jin, Y., & Potkin, S. G. (2000). Relative sparing of emotionally influenced memory in Alzheimer's disease. Neuroreport: For Rapid Communication of Neuroscience Research, 11(4), 653-655.

Moore, B. R. (1973). The Role of Directed Pavlovian Reactions in Simple Instrumental Learning in the Pigeon. In R. A. Hinde & J. Stevenson-Hinde (Eds.), Constraints on Learning (pp. 159-186). London: Academic Press.

Mormede, P., & Dantzer, R. (1977). Effects of dexamethasone on fear conditioning in pigs. Behavioral and Neural Biology, 21(2), 225-235.

Morris, R. G., Garrud, P., Rawlins, J. N., & O' Keefe, J. (1982). Place navigation impaired in rats with hippocampal lesions. Nature, 297(5868), 681-683.

Nader, K., & LeDoux, J. (1999). The dopaminergic modulation of fear: Quinpirole impairs the recall of emotional memories in rats. Behavioral Neuroscience, 113(1), 152-165.

Newcomer, J. W., Craft, S., Hershey, T., Askins, K., & Bardgett, M. E. (1994). Glucocorticoid-induced impairment in declarative memory performance in adult humans. Journal of Neuroscience, 14(4), 2047-53.

314

Newcomer, J. W., Selke, G., Melson, A. K., Hershey, T., Craft, S., Richards, K., & Alderson, A. L. (1999). Decreased memory performance in healthy humans induced by stress-level cortisol treatment. Archives of General Psychiatry, 56(6), 52733. Nielson, K. A., & Jensen, R. A. (1994). Beta-adrenergic receptor antagonist antihypertensive medications impair arousal-induced modulation of working memory in elderly humans. Behavioral and Neural Biology, 62(3), 190-200.

O' Carroll, R. E., Drysdale, E., Cahill, L., Shajahan, P., & Ebmeier, K. P. (1999a). Memory for emotional material: A comparison of central versus peripheral beta blockade. Journal of Psychopharmacology, 13(1), 32-39.

O' Carroll, R. E., Drysdale, E., Cahill, L., Shajahan, P., & Ebmeier, K. P. (1999b). Stimulation of the noradrenergic system enhances and blockade reduces memory for emotional material in man. Psychological Medicine, 29(5), 1083-1088.

Ohl, F., Michaelis, T., Vollmann Honsdorf, G. K., Kirschbaum, C., & Fuchs, E. (2000). Effect of chronic psychosocial stress and long-term cortisol treatment on hippocampus-mediated memory and hippocampal volume: A pilot-study in tree shrews. Psychoneuroendocrinology, 25(4), 357-363.

Oitzl, M. S., & de Kloet, E. R. (1992). Selective corticosteroid antagonists modulate specific aspects of spatial orientation learning. Behavioral Neuroscience, 106(1), 62-71.

Oitzl, M. S., Fluttert, M., & de Kloet, E. R. (1998a). Acute blockade of hippocampal glucocorticoid receptors facilitates spatial learning in rats. Brain Research, 797(1), 159-162.

Oitzl, M. S., Fluttert, M., Sutanto, W., & de Kloet, E. R. (1998b). Continuous blockade of brain glucocorticoid receptors facilitates spatial learning and memory in rats. European Journal of Neuroscience, 10(12), 3759-66.

315

O'Keefe, J., & Nadel, L. (1978). The hippocampus as a cognitive map. Oxford: Clarendon Press.

Orchinik, M., Licht, P., & Crews, D. (1988). Plasma steroid concentrations change in response to sexual behavior in Bufo marinus. Hormones & Behavior, 22(3), 338-50.

Orchinik, M., Murray, T. F., & Moore, F. L. (1991). A corticosteroid receptor in neuronal membranes. Science, 252(5014), 1848-51.

Packard, M. G., Cahill, L., & McGaugh, J. L. (1994). Amygdala modulation of hippocampal-dependent and caudate nucleus-dependent memory processes. Proceedings of the National Academy of Science U S A, 91(18), 8477-81.

Packard, M. G., & McGaugh, J. L. (1996). Inactivation of hippocampus or caudate nucleus with lidocaine differentially affects expression of place and response learning. Neurobiology of Learning and Memory, 65(1), 65-72.

Parent, M. B., Quirarte, G. L., Cahill, L., & McGaugh, J. L. (1995). Spared retention of inhibitory avoidance learning after posttraining amygdala lesions. Behavioral Neuroscience, 109(4), 803-807.

Parkinson, J. A., Cardinal, R. N., & Everitt, B. J. (2000). Limbic corticalventral striatal systems underlying appetitive conditioning. Progress in Brain Research, 126, 263-85.

Parkinson, J. A., Robbins, T. W., & Everitt, B. J. (2000). Dissociable roles of the central and basolateral amygdala in appetitive emotional learning. European Journal of Neuroscience, 12(1), 405-13.

Pavlides, C., Kimura, A., Magarinos, A. M., & McEwen, B. S. (1994). Type I adrenal steroid receptors prolong hippocampal long-term potentiation. Neuroreport, 5(18), 2673-7.

316

Pavlides, C., Kimura, A., Magarinos, A. M., & McEwen, B. S. (1995a). Hippocampal homosynaptic long-term depression/depotentiation induced by adrenal steroids. Neuroscience, 68(2), 379-85.

Pavlides, C., Ogawa, S., Kimura, A., & McEwen, B. S. (1996). Role of adrenal steroid mineralocorticoid and glucocorticoid receptors in long-term potentiation in the CA1 field of hippocampal slices. Brain Research, 738(2), 229-235.

Pavlides, C., Watanabe, Y., Magarinos, A. M., & McEwen, B. S. (1995b). Opposing roles of type I and type II adrenal steroid receptors in hippocampal longterm potentiation. Neuroscience, 68(2), 387-94.

Pavlides, C., Watanabe, Y., & McEwen, B. S. (1993). Effects of glucocorticoids on hippocampal long-term potentiation. Hippocampus, 3(2), 183-92.

Pavlov, I. P. (1927). Conditioned Reflexes. London: Oxford University Press.

Pearce, J. M., & Hall, G. (1980). A model for Pavlovian learning: Variations in the effectiveness of conditioned but not of unconditioned stimuli. Psychological Review, 87(6), 532-552.

Pfeuffer, T. (1977). GTP-binding proteins in membranes and the control of adenylate cyclase activity. Journal of Biological Chemistry, 252(20), 7224-34.

Phelps, E. A., & Anderson, A. K. (1997). Emotional memory: what does the amygdala do? Current Biology, 7(5), R311-4.

Phillips, R. G., & LeDoux, J. E. (1992). Differential contribution of amygdala and hippocampus to cued and contextual fear conditioning. Behavioral Neuroscience, 106(2), 274-285.

Phillips, R. G., & LeDoux, J. E. (1994). Lesions of the dorsal hippocampal formation interfere with background but not foreground contextual fear conditioning. Learning and Memory, 1(1), 34-44. 317

Piazza, P. V., Deminiere, J. M., Le Moal, M., & Simon, H. (1989). Factors that predict individual vulnerability to amphetamine self-administration. Science, 245(4925), 1511-1513.

Piazza, P. V., Deminiere, J. M., Maccari, S., Mormede, P., Le Moal, M., & Simon, H. (1990). Individual reactivity to novelty predicts probability of amphetamine self-administration. Behavioural Pharmacology, 1(4), 339-345.

Piazza, P. V., Deroche, V., Deminiere, J. M., Maccari, S., Le Moal, M., & Simon, H. (1993). Corticosterone in the range of stress-induced levels possesses reinforcing properties: implications for sensation-seeking behaviors. Proceedings of the National Academy of Science U S A, 90(24), 11738-42.

Piazza, P. V., & Le Moal, M. (1996). Pathophysiological basis of vulnerability to drug abuse: Role of an interaction between stress, glucocorticoids, and dopaminergic neurons. Annual Review of Pharmacology and Toxicology, 36, 359378.

Piazza, P. V., & Le Moal, M. (1997). Glucocorticoids as a biological substrate of reward: physiological and pathophysiological implications. Brain Research Reviews, 25(3), 359-72.

Piazza, P. V., Maccari, S., Deminiere, J. M., Le Moal, M., Mormede, P., & Simon, H. (1991a). Corticosterone levels determine individual vulnerability to amphetamine self-administration. Proceedings of the National Academy of Science U S A, 88(6), 2088-92.

Piazza, P. V., Rouge-Pont, F., Deminiere, J. M., Kharoubi, M., Le Moal, M., & Simon, H. (1991b). Dopaminergic activity is reduced in the prefrontal cortex and increased in the nucleus accumbens of rats predisposed to develop amphetamine selfadministration. Brain Research, 567(1), 169-74.

Piazza, P. V., Rouge-Pont, F., Deroche, V., Maccari, S., Simon, H., & Le Moal, M. (1996). Glucocorticoids have state-dependent stimulant effects on the 318

mesencephalic dopaminergic transmission. Proceedings of the National Academy of Science U S A, 93(16), 8716-20.

Port, R. L., Sisak, M. E., Finamore, T. L., Soltrick, M. L., & Seybold, K. S. (1998). Role of corticosterone in extinction of an appetitive instrumental response. International Journal of Neuroscience, 96, 13-21.

Power, A. E., Roozendaal, B., & McGaugh, J. L. (2000). Glucocorticoid enhancement of memory consolidation in the rat is blocked by muscarinic receptor antagonism in the basolateral amygdala. European Journal of Neuroscience, 12(10), 3481-3487.

Pugh, C. R., Fleshner, M., & Rudy, J. W. (1997a). Type II glucocorticoid receptor antagonists impair contextual but not auditory-cue fear conditioning in juvenile rats. Neurobiology of Learning and Memory, 67(1), 75-79.

Pugh, C. R., Tremblay, D., Fleshner, M., & Rudy, J. W. (1997b). A selective role for corticosterone in contextual-fear conditioning. Behavioral Neuroscience, 111(3), 503-511.

Puia, G., Santi, M. R., Vicini, S., Pritchett, D. B., Purdy, R. H., Paul, S. M., Seeburg, P. H., & Costa, E. (1990). Neurosteroids act on recombinant human GABAA receptors. Neuron, 4(5), 759-65.

Quirarte, G. L., Galvez, R., Roozendaal, B., & McGaugh, J. L. (1998). Norepinephrine release in the amygdala in response to footshock and opioid peptidergic drugs. Brain Research, 808(2), 134-140.

Quirarte, G. L., Roozendaal, B., & McGaugh, J. L. (1997). Glucocorticoid enhancement of memory storage involves noradrenergic activation in the basolateral amygdala. Proceedings of the National Academy of Science U S A, 94(25), 14048-53.

319

Radulovic, J., Ruehmann, A., Liepold, T., & Spiess, J. (1999). Modulation of learning and anxiety by corticotropin-releasing factor (CRF) and stress: Differential roles of CRF receptors 1 and 2. Journal of Neuroscience, 19(12), 5016-5025.

Randich, A., Froehlich, J. C., Fraley, S. M., Fjermestad, J. L., & Brush, F. R. (1976). Pituitary-adrenal activity and appetitive reinforcement. Physiological Psychology, 4(3), 346-350.

Reddy, D. S., & Kulkarni, S. K. (1998). The effects of neurosteroids on acquisition and retention of a modified passive-avoidance learning task in mice. Brain Research, 791(1-2), 108-116.

Rescorla, R. A. (1996). Preservation of Pavlovian associations through extinction. Quarterly Journal of Experimental Psychology: Comparative and Physiological Psychology, 3, 245-258.

Rescorla, R. A., & Cunningham, C. L. (1978). Within-compound flavor associations. Journal of Experimental Psychology: Animal Behavior Processes, 4(3), 267-275.

Rescorla, R. A., & Wagner, A. R. (1972). A Theory of Pavlovian Conditioning in the Effectiveness of Reinforcement and Nonreinforcement. In A. H. Black & W. F. Prokasy (Eds.), Classical Conditioning (Vol. 2, pp. 64-99). New York: AppletonCentury-Crofts.

Reul, J. M., & de Kloet, E. R. (1985). Two receptor systems for corticosterone in rat brain: microdistribution and differential occupation. Endocrinology, 117(6), 2505-11.

Reul, J. M., Gesing, A., Droste, S., Stec, I. S., Weber, A., Bachmann, C., Bilang-Bleuel, A., Holsboer, F., & Linthorst, A. C. (2000). The brain mineralocorticoid receptor: greedy for ligand, mysterious in function. European Journal of Pharmacology, 405(1-3), 235-49.

320

Reul, J. M., Pearce, P. T., Funder, J. W., & Krozowski, Z. S. (1989). Type I and type II corticosteroid receptor gene expression in the rat: effect of adrenalectomy and dexamethasone administration. Molecular Endocrinology, 3(10), 1674-80.

Revusky, S. (1977). The concurrent approach to delay learning. In L. M. Barker, M. R. Best, & M. Domjan (Eds.), Learning mechanisms in food selection (pp. 319-366). Waco, TX: Baylor University Press.

Ricceri, L., Alleva, E., & Calamandrei, G. (1994). Impairment of passive avoidance learning following repeated administrations of antibodies against nerve growth factor in neonatal mice. Neuroreport, 5(11), 1401-4.

Richter-Levin, G., & Akirav, I. (2000). Amygdala-hippocampus dynamic interaction in relation to memory. Molecular Neurobiology, 22(1-3), 11-20.

Rizley, R. C., & Rescorla, R. A. (1972). Associations in second-order conditioning and sensory preconditioning. Journal of Comparative and Physiological Psychology, 81(1), 1-11.

Robbins, T. W., & Everitt, B. J. (2001). Limbic-striatal memory systems and drug addiction. Paper presented at the Seventh conference on the neurobiology of learning and memory: Orchestration of cells and systems. Making memories in the brain., University of California, Irvine, USA.

Roesler, R., Berlau, D. J., LaLumiere, R. T., Roozendaal, B., & McGaugh, J. L. (2001). Inhibitory avoidance requires contextual experience but not an instrumental response during training. Society for Neuroscience Abstracts, 27(2), 743.1.

Roozendaal, B. (2000). Glucocorticoids and the regulation of memory consolidation. Psychoneuroendocrinology, 25(3), 213-238.

Roozendaal, B., Bohus, B., & McGaugh, J. L. (1996a). Dose-dependent suppression of adrenocortical activity with metyrapone: Effects on emotion and memory. Psychoneuroendocrinology, 21(8), 681-693. 321

Roozendaal, B., Carmi, O., & McGaugh, J. L. (1996c). Adrenocortical suppression blocks the memory-enhancing effects of amphetamine and epinephrine. Proceedings of the National Academy of Science U S A, 93(4), 1429-33.

Roozendaal, B., de Quervain, D. J. F., Ferry, B., Setlow, B., & McGaugh, J. L. (2001a). Basolateral amygdala-nuclues accumbens interactions in mediating glucocorticoid enhancement of memory consolidation. Journal of Neuroscience, 21(7), 2518-2525.

Roozendaal, B., Koolhaas, J. M., & Bohus, B. (1991). Attenuated cardiovascular, neuroendocrine, and behavioral responses after a single footshock in central amygdaloid lesioned male rats. Physiology and Behavior, 50(4), 771-775.

Roozendaal, B., Koolhaas, J. M., & Bohus, B. (1992). Central Amygdaloid Involvement in Neuroendocrine Correlates of Conditioned Stress Responses. Journal of Neuroendocrinology, 4(4), 483-489.

Roozendaal, B., & McGaugh, J. L. (1996a). Amygdaloid nuclei lesions differentially affect glucocorticoid-induced memory enhancement in an inhibitory avoidance task. Neurobiology of Learning and Memory, 65(1), 1-8.

Roozendaal, B., & McGaugh, J. L. (1996b). The memory-modulatory effects of glucocorticoids depend on an intact stria terminalis. Brain Research, 709(2), 243250.

Roozendaal, B., & McGaugh, J. L. (1997a). Glucocorticoid receptor agonist and antagonist administration into the basolateral but not central amygdala modulates memory storage. Neurobiology of Learning and Memory, 67(2), 176-179.

Roozendaal, B., & McGaugh, J. L. (1997b). Basolateral amygdala lesions block the memory-enhancing effect of glucocorticoid administration in the dorsal hippocampus of rats. European Journal of Neuroscience, 9(1), 76-83.

322

Roozendaal, B., Nguyen, B. T., Power, A. E., & McGaugh, J. L. (1999b). Basolateral amygdala noradrenergic influence enables enhancement of memory consolidation

induced

by

hippocampal

glucocorticoid

receptor

activation.

Proceedings of the National Academy of Science U S A, 96(20), 11642-7.

Roozendaal, B., Portillo Marquez, G., & McGaugh, J. L. (1996b). Basolateral amygdala lesions block glucocorticoid-induced modulation of memory for spatial learning. Behavioral Neuroscience, 110(5), 1074-1083.

Roozendaal, B., Quirarte, G. L., & McGaugh, J. L. (2001b). Glucocorticoids interact with the basolateral amygdala beta-adrenoceptor-cAMP/PKA system in influencing memory consolidation. Society for Neuroscience Abstracts, 27(1), 312.15.

Roozendaal, B., Sapolsky, R. M., & McGaugh, J. L. (1998). Basolateral amygdala lesions block the disruptive effects of long-term adrenalectomy on spatial memory. Neuroscience, 84(2), 453-65.

Roozendaal, B., Williams, C. L., & McGaugh, J. L. (1999a). Glucocorticoid receptor activation in the rat nucleus of the solitary tract facilitates memory consolidation: involvement of the basolateral amygdala. European Journal of Neuroscience, 11(4), 1317-23.

Rose, S. P. R. (1995). Cell-adhesion molecules, glucocorticoids and longterm-memory formation. Trends in Neurosciences, 18(11), 502-506.

Rosen, J. B., & Schulkin, J. (1998). From normal fear to pathological anxiety. Psychological Review, 105(2), 325-350.

Rouge Pont, F., Piazza, P. V., Kharouby, M., Le Moal, M., & Simon, H. (1993). Higher and longer stress-induced increase in dopamine concentrations in the nucleus accumbens of animals predisposed to amphetamine self-administration: A microdialysis study. Brain Research, 602(1), 169-174.

323

Rouge-Pont, F., Deroche, V., Le Moal, M., & Piazza, P. V. (1998). Individual differences in stress-induced dopamine release in the nucleus accumbens are influenced by corticosterone. European Journal of Neuroscience, 10(12), 3903-7.

Rubinow, D. R., Post, R. M., Savard, R., & Gold, P. W. (1984). Cortisol hypersecretion and cognitive impairment in depression. Archives of General Psychiatry, 41(3), 279-83.

Rudy, J. W., Kuwagama, K., & Pugh, C. R. (1999). Isolation reduces contextual but not auditory-cue fear conditioning: A role for endogenous opioids. Behavioral Neuroscience, 113(2), 316-323.

Rudy, J. W., & Pugh, C. R. (1996). A comparison of contextual and generalized auditory-cue fear conditioning: evidence for similar memory processes. Behavioral Neuroscience, 110(6), 1299-1308.

Rudy, J. W., & Pugh, C. R. (1998). Time of conditioning selectively influences contextual fear conditioning: Further support for a multiple-memory systems view of fear conditioning. Journal of Experimental Psychology: Animal Behavior Processes, 24(3), 316-324.

Sandi, C. (1998). The role and mechanisms of action of glucocorticoid involvement in memory storage. Neural Plasticity, 6(3), 41-52.

Sandi, C., Loscertales, M., & Guaza, C. (1997). Experience-dependent facilitating effect of corticosterone on spatial memory formation in the water maze. European Journal of Neuroscience, 9(4), 637-42.

Sandi, C., & Rose, S. P. (1994a). Corticosteroid receptor antagonists are amnestic for passive avoidance learning in day-old chicks. European Journal of Neuroscience, 6(8), 1292-7.

324

Sandi, C., Rose, S. P., Mileusnic, R., & Lancashire, C. (1995). Corticosterone facilitates long-term memory formation via enhanced glycoprotein synthesis. Neuroscience, 69(4), 1087-93.

Sandi, C., & Rose, S. P. R. (1994b). Corticosterone enhances long-term retention in one-day-old chicks trained in a weak passive avoidance learning paradigm. Brain Research, 647(1), 106-112.

Sandi, C., & Rose, S. P. R. (1997). Training-dependent biphasic effects of corticosterone in memory formation for a passive avoidance task in chicks. Psychopharmacology, 133(2), 152-160.

Sapolsky, R. M. (1986). Glucocorticoid toxicity in the hippocampus: reversal by supplementation with brain fuels. Journal of Neuroscience, 6(8), 2240-4.

Sapolsky, R. M. (2000). Stress hormones: good and bad. Neurobiology of Disease, 7(5), 540-2.

Sawchenko, P. E. (1991). The final common path: Issues Concerning in Organization of Central Mechanism Controlling Corticotropin Secretion. In M. R. Brown, G. Koob, & C. Rivier (Eds.), Stress: Neurobiology & Neuroendocrinology (pp. 55-71). New York: Dekker.

Schinkel, A. H., Smit, J. J., van Tellingen, O., Beijnen, J. H., Wagenaar, E., van Deemter, L., Mol, C. A., van der Valk, M. A., Robanus-Maandag, E. C., te Riele, H. P., Berns, A. J. M., & Borst, P. (1994). Disruption of the mouse mdr1a Pglycoprotein gene leads to a deficiency in the blood-brain barrier and to increased sensitivity to drugs. Cell, 77(4), 491-502.

Schumacher, M. (1990). Rapid membrane effects of steroid hormones: An emerging concept in neuroendocrinology. Trends in Neurosciences, 13(9), 359-362.

325

Selden, N. R., Everitt, B. J., Jarrard, L. E., & Robbins, T. W. (1991). Complementary roles for the amygdala and hippocampus in aversive conditioning to explicit and contextual cues. Neuroscience, 42(2), 335-50.

Setlow, B., Roozendaal, B., & McGaugh, J. L. (2000). Involvement of a basolateral amygdala complex-nucleus accumbens pathway in glucocorticoid-induced modulation of memory consolidation. European Journal of Neuroscience, 12(1), 36775.

Shalev, U., Feldon, J., & Weiner, I. (1998). Latent inhibition is disrupted by acute or repeated administration of corticosterone. International Journal of Neuropsychopharmacology, 1(2), 103-113.

Shalev, U., & Weiner, I. (2001). Gender-dependent differences in latent inhibition following prenatal stress and corticosterone administration. Behavioural Brain Research, 126(1-2), 57-63.

Shors, T. J. (1998). Stress and Sex Effects on Associative Learning: For Better or for Worse. The Neuroscientist, 4(5), 353-364.

Shors, T. J. (2001). Acute stress rapidly and persistently enhances memory formation in the male rat. Neurobiology of Learning and Memory, 75(1), 10-29.

Shors, T. J., Beylin, A. V., Wood, G. E., & Gould, E. (2000). The modulation of Pavlovian memory. Behavioural Brain Research, 110(1-2), 39-52.

Shors, T. J., & Mathew, P. R. (1998). NMDA receptor antagonism in the lateral/basolateral but not central nucleus of the amygdala prevents the induction of facilitated learning in response to stress. Learning and Memory, 5(3), 220-230.

Shors, T. J., & Matzel, L. D. (1997). Long-term potentiation: What's learning got to do with it? Behavioral and Brain Sciences, 20(4), 597-655.

326

Shors, T. J., Pickett, J., Wood, G., & Paczynski, M. (1999). Acute stress persistently enhances estrogen levels in the female rat. Stress, 3(2), 163-71.

Shors, T. J., Weiss, C., & Thompson, R. F. (1992). Stress-induced facilitation of classical conditioning. Science, 257(5069), 537-9. Sloviter, R. S., Valiquette, G., Abrams, G. M., Ronk, E. C., Sollas, A. L., Paul, L. A., & Neubort, S. (1989). Selective loss of hippocampal granule cells in the mature rat brain after adrenalectomy. Science, 243(4890), 535-8.

Starkman, M. N., Gebarski, S. S., Berent, S., & Schteingart, D. E. (1992). Hippocampal formation volume, memory dysfunction, and cortisol levels in patients with Cushing's syndrome. Biological Psychiatry, 32(9), 756-765.

Takahashi, L. K., & Kim, H. (1994). Intracranial action of corticosterone facilitates the development of behavioral inhibition in the adrenalectomized preweanling rat. Neuroscience Letters, 176(2), 272-6.

Teyler, T. J., & DiScenna, P. (1987). Long-term potentiation. Annual Review of Neuroscience, 10, 131-61.

Thomas, S. R., Assaf, S. Y., & Iversen, S. D. (1984). Amygdaloid complex modulates neurotransmission from the entorhinal cortex to the dentate gyrus of the rat. Brain Research, 307(1-2), 363-5.

Timpl, P., Spanagel, R., Sillaber, I., Kresse, A., Reul, J. M., Stalla, G. K., Blanquet, V., Steckler, T., Holsboer, F., & Wurst, W. (1998). Impaired stress response and reduced anxiety in mice lacking a functional corticotropin-releasing hormone receptor 1. Nature Genetics, 19(2), 162-6.

Trapold, M. A., & Overmier, J. B. (1972). The second learning process in instrumental learning. In A. H. Black & W. F. Prokasy (Eds.), Classical conditioning II: Current research and theory (pp. 427-452). New York: Appleton-Century-Crofts.

327

Vale, W., Spiess, J., Rivier, C., & Rivier, J. (1981). Characterization of a 41residue ovine hypothalamic peptide that stimulates secretion of corticotropin and betaendorphin. Science, 213(4514), 1394-7.

van Stegeren, A. H., Everaerd, W., Cahill, L., McGaugh, J. L., & Gooren, L. J. G. (1998). Memory for emotional events: Differential effects of centrally versus peripherally acting beta-blocking agents. Psychopharmacology, 138(3-4), 305-310.

van Wimersma Greidanus, T. B. (1970). Effects of steroid on extinction of an avoidance response in rats. A structure-activity relationship study. Progress in Brain Research, 32, 185-91.

van Wimersma Greidanus, T. B., & De Wied, D. (1969). Effects on intracerebral implantation of corticosteroids on extinction of an avoidance response in rats. Physiology and Behavior, 4(3), 365-370.

Vazdarjanova, A., & McGaugh, J. L. (1998). Basolateral amygdala is not critical for cognitive memory of contextual fear conditioning. Proceedings of the National Academy of Science U S A, 95(25), 15003-7.

Vedhara, K., Hyde, J., Gilchrist, I. D., Tytherleigh, M., & Plummer, S. (2000). Acute stress, memory, attention and cortisol. Psychoneuroendocrinology, 25(6), 535549.

Veldhuis, H. D., De Korte, C. C., & De Kloet, E. R. (1985). Glucocorticoids facilitate the retention of acquired immobility during forced swimming. European Journal of Pharmacology, 115(2-3), 211-7.

Veldhuis, H. D., Van Koppen, C., Van Ittersum, M., & De Kloet, E. R. (1982). Specificity of the adrenal steroid receptor system in rat hippocampus. Endocrinology, 110(6), 2044-51.

328

Wagner, A. R. (1978). Expectancies and the Priming of STM. In S. H. Hulse, H. Fowler, & W. K. Honig (Eds.), Cognitive Processes in Animal Behavior (pp. 177193). Potomac, MD: Lawrence Erlbaum.

Wagner, A. R. (1981). SOP: A Model of Automatic Memory Processing in Animal Behavior. In N. E. Spear & R. R. Miller (Eds.), Information processing in animals: Memory mechanisms (pp. 5-47). Hillsdale, NJ: Erlbaum.

Ward-Robinson, J., Coutureau, E., Good, M., Honey, R. C., Killcross, A. S., & Oswald, C. J. P. (2001). Excitotoxic lesions of the hippocampus leave sensory preconditioning intact: Implications for models of hippocampal functioning. Behavioral Neuroscience, 115(6), 1357-1362.

Ward-Robinson, J., Symonds, M., & Hall, G. (1998). Context specificity of sensory preconditioning: Implications for processes of within-event learning. Animal Learning and Behavior, 26(2), 225-232.

Weiner, I. (1990). Neural substrates of latent inhibition: the switching model. Psychological Bulletin, 108(3), 442-61.

Wessel, I., & Merckelbach, H. (1998). Memory for threat-relevant and threatirrelevant cues in spider phobics. Cognition and Emotion, 12(1), 93-104.

Whelan, T. B., Schteingart, D. E., Starkman, M. N., & Smith, A. (1980). Neuropsychological deficits in Cushing's syndrome. Journal of Nervous and Mental Disease, 168(12), 753-7.

Williams, C. L., Men, D., Clayton, E. C., & Gold, P. E. (1998). Norepinephrine release in the amygdala after systemic injection of epinephrine or escapable footshock: contribution of the nucleus of the solitary tract. Behavioral Neuroscience, 112(6), 1414-22.

329

Wolf, O. T., Convit, A., McHugh, P. F., Kandil, E., Thorn, E. L., De Santi, S., McEwen, B. S., & de Leon, M. J. (2001a). Cortisol differentially affects memory in young and elderly men. Behavioral Neuroscience, 115(5), 1002-1011.

Wolf, O. T., Koester, B., Kirschbaum, C., Pietrowsky, R., Kern, W., Hellhammer, D. H., Born, J., & Fehm, H. L. (1997). A single administration of dehydroepiandrosterone does not enhance memory performance in young healthy adults, but immediately reduces cortisol levels. Biological Psychiatry, 42(9), 845-848.

Wolf, O. T., Schommer, N. C., Hellhammer, D. H., McEwen, B. S., & Kirschbaum, C. (2001b). The relationship between stress induced cortisol levels and memory differs between men and women. Psychoneuroendocrinology, 26(7), 711720.

Wolkowitz, O. M., Reus, V. I., Weingartner, H., Thompson, K., Breier, A., Doran, A., Rubinow, D., & Pickar, D. (1990). Cognitive effects of corticosteroids. American Journal of Psychiatry, 147(10), 1297-303.

Wood, G. E., & Shors, T. J. (1998). Stress facilitates classical conditioning in males, but impairs classical conditioning in females through activational effects of ovarian hormones. Proceedings of the National Academy of Science U S A, 95(7), 4066-71.

Wright, C. I., Beijer, A. V., & Groenewegen, H. J. (1996). Basal amygdaloid complex afferents to the rat nucleus accumbens are compartmentally organized. Journal of Neuroscience, 16(5), 1877-93.

Young, A. M., Ahier, R. G., Upton, R. L., Joseph, M. H., & Gray, J. A. (1998). Increased extracellular dopamine in the nucleus accumbens of the rat during associative learning of neutral stimuli. Neuroscience, 83(4), 1175-83.

Young, S. L., Bohenek, D. L., & Fanselow, M. S. (1994). NMDA processes mediate anterograde amnesia of contextual fear conditioning induced by hippocampal

330

damage: immunization against amnesia by context preexposure. Behavioral Neuroscience, 108(1), 19-29.

Yuille, J. C., & Cutshall, J. L. (1986). A case study of eyewitness memory of a crime. Journal of Applied Psychology, 71(2), 291-301.

Zorawski, M., & Killcross, S. (2000a). Posttraining glucocorticoid receptor agonist enhances memory in an appetitive discrete cue Pavlovian conditioning paradigm. European Journal of Neuroscience, 12(Supplement 11), 78.3.

Zorawski, M., & Killcross, S. (2000b). Effects of posttraining/postpreexposure glucocorticoid receptor agonist in appetitive and aversive discrete cue Pavlovian conditioning paradigms and latent inhibition. Journal of Psychopharmacology, Supplement to Volume 14 Number 3, A51.

Zorawski, M., & Killcross, S. (2000c). Post-training dexamethasone in appetitive and aversive cue conditioning and latent inhibition. Society for Neuroscience Abstracts, 26(2), 562.3.

Zorawski, M., & Killcross, S. (2001a). Posttraining glucocorticoids modulate CS-US but not CS-nothing associations. Journal of Psychopharmacology, Supplement to Volume 15 Number 3, A69.

Zorawski,

M.,

&

Killcross,

S.

(2001b).

Memory

modulation

by

glucocorticoids is selective for the motivational properties of the US - evidence from a Pavlovian-instrumental transfer study. Society for Neuroscience Abstracts, 84.11.

Zorawski, M., & Killcross, S. (accepted). Post-training glucocorticoid receptor agonist enhances memory in appetitive and aversive Pavlovian discrete-cue conditioning paradigms. Neurobiology of learning and memory.

331

Author information and contact details

I was born in Hamburg, Germany on March 3, 1974. I grew up and went to school in Kaltenkirchen. Between 1995 to 1998, I studied psychology (BSc Hons) at the University of York and, as an exchange student, at the University of California San Diego. I started this PhD at the University of York in October 1998, and transferred to Cardiff University in September 1999.

For correspondence regarding this dissertation, I can be contacted by electronic mail: [email protected]

Word count Chapters

67,340

References

10,321

Other

3,565

Total

81,226

332