Blackwell Science, LtdOxford, UKADDAddiction0965-2140© 2005 Society for the Study of Addiction 100 Original Article ‘Wanting’ and ‘liking’ for alcohol and amphetamine P. Willner et al.
RESEARCH REPORT
Excessive alcohol consumption and dependence on amphetamine are associated with parallel increases in subjective ratings of both ‘wanting’ and ‘liking’ Paul Willner1,2, Darren James1 & Michael Morgan2 Department of Psychology, University of Wales Swansea, Swansea1 and Department of Psychology, University of Sussex, Brighton, UK2
Correspondence to: Paul Willner, Department of Psychology University of Wales Swansea Swansea SA2 8PP UK E-mail:
[email protected] Submitted 22 December 2004; initial review completed 13 March 2005; final version accepted 17 May 2005
RESEARCH REPORT
ABSTRACT Aim One of the tenets of the incentive sensitization theory of drug addiction is that ‘as drugs come to be wanted more and more, they often come to be liked less and less’. The aim of this study was to test whether this assumption holds true. Specifically, the study aimed to test the hypothesis that in non-clinical samples, dependence on amphetamines and excessive alcohol use are associated with increased ‘wanting’ but decreased ‘liking’ for the drug. Design, setting and participants In two studies, the Desires for Alcohol Questionnaire (DAQ) was administered to 380 recreational drinkers, and the Desires for Speed Questionnaire (DSQ) and the Leeds Dependence Questionnaire (LDQ) were administered to 174 amphetamine users. Scales were derived from the DAQ and DSQ representing craving, on one hand, and positive and negative reinforcement on the other hand. Craving and positive reinforcement were taken as measures of alcohol or amphetamine ‘wanting’ and ‘liking’, respectively. Findings Scores on all three DAQ scales increased monotonically as a function of the extent of alcohol consumption. Scores on all three DSQ scales increased monotonically as a function of dependence, as measured by the LDQ. ‘Liking’ for amphetamine was unrelated to time since the drug was last taken. (These data were not available for alcohol.) Conclusions The finding that ‘wanting’ increased as a function of dependence on amphetamine or level of consumption in the case of alcohol is as predicted by the incentive sensitization theory, but the finding that ‘liking’ also increased as a function of dependence or excess is the opposite of the predicted effects While not refuting the incentive sensitization theory directly, the study questions the validity of one of the tenets of the theory. KEYWORDS Alcohol, amphetamine, dependence, human, incentive sensitization, liking, negative reinforcement, wanting.
INTRODUCTION On the basis of a wide array of research findings, principally from animal literature on the effects of stimulant drugs, Robinson & Berridge (1993) argued that the compulsive drug-seeking and drug-taking behaviours that characterize addiction are often not motivated by either © 2005 Society for the Study of Addiction
the desire to obtain pleasure or by the desire to relieve withdrawal. Instead they proposed an ‘incentive sensitization’ theory in which the psychological process of incentive salience is specifically responsible for instrumental drug-seeking and drug-taking behaviours, and that this is mediated by sensitization of particular brain systems (including the mesocorticolimbic dopamine sys-
doi:10.1111/j.1360-0443.2005.01222.x
Addiction, 100, 1487–1495
1488
P. Willner et al.
tem) as a result of repeated exposure to potentially addictive drugs. One of the features of addiction that this theory is intended to explain is that ‘as drugs come to be wanted more and more, they often come to be liked less and less’ (Robinson & Berridge 1993, p. 249). According to the incentive sensitization theory, ‘repeated drug use does not sensitise neural systems responsible for the subjective pleasurable effects of drugs, only those responsible for incentive salience—transforming “wanting” into craving’. They argued that addiction is characterized by an increasing dissociation between the incentive value of drugs (how much they are wanted) and their subjective pleasurable effects, such that the progressive increase in drug ‘wanting’ that characterizes addiction is not accompanied by an increase in the pleasure or ‘liking’ derived from the drugs. Indeed, it is specified that ‘the neural system responsible for the subjective pleasurable effects of drugs either does not change or else becomes hyposensitive (tolerant)’ (Robinson & Berridge 1993). The suggestion that ‘drugs can become pathologically wanted (“craved”) over the course of the development of an addiction even if they are liked less and less’ has been repeated in subsequent iterations of the theory (e.g. Robinson & Berridge 2001). However, at no time have Robinson & Berridge provided any direct evidence to support their hypothesis that subjective measures of drug ‘wanting’ and drug ‘liking’ become dissociated over the course of the development of drug dependence in humans, such that ‘wanting’ increases but ‘liking’ decreases. Indeed, in the original paper (Robinson & Berridge 1993) they acknowledge that ‘there is very little evidence linking escalation in dose with tolerance to the subjective effects of drugs’, and that there are a number of reports that addicts continue to experience euphoria even after years of drug use (e.g. Lasagna et al. 1955; McAuliff & Gordon 1974; McAuliff 1975). In their original presentation of the theory Robinson & Berridge (1993) cited, as evidence for their theory, studies that show that drug self-administration can be maintained in the absence of subjective pleasure. In one of these studies opiate ‘post-addicts’ were reported to work for an injection of a low dose of morphine but not placebo, despite the fact that four of the five participants could not distinguish the subjective effects of this dose of morphine from placebo (Lamb et al. 1991). In another study cocaine users reliably chose a low dose of cocaine over placebo despite the fact that this dose produces no subjective or cardiovascular effects (Fischman & Foltin 1992). These studies demonstrate clearly that reinforcement can occur without overt awareness, but they fail to show that subjective craving is dissociable from subjective liking. The human subjective evidence that Robinson & Berridge have cited more recently to support their theory © 2005 Society for the Study of Addiction
(Berridge & Robinson 1998; Robinson & Berridge 2003) is generally in agreement with the view that dopamine mediates ‘wanting’ or against the view that dopamine mediates ‘liking’. For example, they cite evidence that dopamine-receptor antagonists often do not suppress the subjective pleasure ratings of amphetamine or cigarettes in humans (Brauer et al. 1997, 2001; Wachtel et al. 2002). In a recent paper (Robinson & Berridge 2003) they cite positron emission tomography (PET) evidence, from [11C]raclopride binding studies, that mesolimbic dopamine might mediate interest in obtaining reward, rather than reward per se. Administration of damphetamine elevated dopamine, and changes in extracellular dopamine correlated with subjective reports of ‘drug wanting’ but not with mood (Leyton et al. 2002). Furthermore, although non-hedonic food stimulation (display of food without consumption) did not increase extracellular dopamine levels or the desire for food in a placebo condition, the food stimulation did increase dopamine levels and desire for food when preceded by methylphenidate (Volkow et al. 2002). Robinson & Berridge (2003) interpret this as evidence that activation of dopaminergic systems correlates better with subjective ratings of wanting for drug or food reward than with subjective ratings of pleasure. However, it is important to note that only one of the latter studies actually measured subjective wanting and mood in the same participants (Leyton et al. 2002), and none of these studies simultaneously measured subjective craving for, and liking of, drugs as a function of dependence or degree of exposure to the drug. Although not cited by Robinson & Berridge, other investigators have examined the relationship between craving and overt drug-seeking behaviour or drug consumption in laboratory studies. For example, De Wit (2000) reviewed the results of several laboratory studies in which both alcohol consumption and self-report measures of craving or desire for alcohol were measured, using subjects who were non-problem social drinkers. A priming dose of alcohol was used to increase craving and a dose of naltrexone was administered to decrease craving. In most of these studies self-reported ratings of desire for alcohol were positively associated with consumption of alcohol, although there were also instances in which one measure varied independently of the other, indicating that under certain circumstances they may be dissociable. For example, alcohol pre-loads increased ratings of desire for alcohol without altering behavioural preference for alcohol over money. However, although these studies measured craving or drug ‘wanting’, they did not measure subjective drug ‘liking’ and thus could not determine if drug ‘wanting’ and ‘liking’ become dissociated over the course of the development of drug dependence in humans. Addiction, 100, 1487–1495
‘Wanting’ and ‘liking’ for alcohol and amphetamine
A study by Volkow et al. (1997) probably comes closer than any other in the literature to testing the hypothesis that drug ‘wanting’ and ‘liking’ may become dissociated, in humans, over the course of the development of drug dependence. These investigators used PET to compare the responses of cocaine addicts, and controls with no history of drug abuse, to intravenous methylphenidate, a dopamine agonist. Addicts showed reduced dopamine release in the striatum and a reduced drug-induced ‘high’ relative to controls. In contrast, cocaine addicts showed an increased response to methylphenidate in the thalamus, which was associated with an elevated craving for cocaine and was not seen in control subjects. However, there are a number of problems with this study in relation to using these data to test the incentive sensitization theory: it measured drug-induced ‘high’ rather than drug liking; the observations were made well after the peak of the ‘high’ had subsided; the ‘high’ produced by one drug (methylphenidate) was compared with the effect of administration of that drug on ‘wanting’ (craving) for another (cocaine); and the controls were non-drug-abusers, rather than recreational but non-addicted cocaine users. Ideally, the optimum way of testing the Robinson & Berridge hypothesis directly would be to measure subjective ‘wanting’ for, and ‘liking’ of, the same drug, by users who are dependent on the drug to varying degrees. In the course of two studies carried out for the purpose of developing questionnaires to measure cravings for alcohol and amphetamine, we have gathered evidence that is relevant to this debate. The Desires for Alcohol Questionnaire (DAQ: Love et al. 1998) was based on ideas developed by Tiffany and colleagues, suggesting that cravings involve several relatively independent factors and are therefore best measured by multi-dimensional questionnaires (Tiffany & Drobes 1991; Tiffany et al. 1993; Singleton & Tiffany 1994). The Desires for Speed Questionnaire (DSQ: James et al. 2004) was developed by adapting and expanding the DAQ. For both the DAQ and the DSQ there was a separation between strong desires to take the drug, and reinforcement anticipated from so doing, which were represented by different factors in the solutions to principal components analyses of questionnaire items. The ‘strong desires’ factors in the DAQ and DSQ provide measures of ‘wanting’ for alcohol and amphetamine. The ‘reinforcement’ factors include both positive and negative reinforcement items, and the positive reinforcement items can be used to derive measures of ‘liking’ for alcohol and amphetamine. Therefore, in the present study, we have re-analysed these data to derive three separate scales, measuring ‘strong desires’ for alcohol or amphetamine (‘wanting’), positive reinforcement (‘liking’) and negative reinforcement. Our study of the DAQ (Love et al. 1998) also included information on the quantities of alcohol consumed, and © 2005 Society for the Study of Addiction
1489
our study of the DSQ (James et al. 2004) also included a measure of drug dependence, the Leeds Dependence Questionnaire (LDQ). This allowed us to test the hypothesis, derived directly from the incentive sensitization theory of drug addiction, that increasing dependence on alcohol or amphetamine should be associated with an increase in ‘wanting’ but a decrease in ‘liking’ for the drug. We report here the results of these analyses.
METHODS Alcohol study Participants A total of 380 recreational drinkers participated in the study, 238 of whom provided the data that contributed to the present analyses (others were excluded for reason of missing data). Participants were recruited at various times of day and in a variety of locations within the Swansea and Swansea Valley area of South Wales. The sample was 56% female, with an age range from 15 to 68 years (mean = 28.66, SD = 11.66). Participants were asked to complete the DAQ and a number of questions relating to their demographic details and their experience with alcohol. They also completed another alcohol craving questionnaire (see Love et al. 1998), the results of which do not contribute to the present analysis.
The DAQ The initial analyses of this data set are described elsewhere (Love et al. 1998). Briefly, participants were prompted on a Likert-type scale, with scores ranging between 1 and 7, as to how strongly they agreed or disagreed with each of 36 statements, following which principal components analysis was used to derive a factor structure for the DAQ. This analysis produced three factors, which were labelled ‘positive and negative reinforcement’, ‘strong desires and intentions to use’ and ‘mild desires and intentions to use’. For the purposes of the present analysis factor 2, ‘strong desires and intentions’, was taken as a measure of alcohol ‘wanting’. This factor contained 11 items. One item (‘nothing would be better than drinking now’) was ambiguous because the equivalent item on the DSQ loaded on a different factor (see below); this item was therefore deleted. The other 10 items all reflect craving for alcohol (e.g. ‘I want a drink so much I can almost taste it’), none of which make any reference to the consequences of using alcohol. Factor 1 of the DAQ contained 12 items, of which there were eight reflecting negatively reinforcing conseAddiction, 100, 1487–1495
1490
P. Willner et al.
quences of using alcohol (e.g. ‘drinking now would make the bad things in my life seem less bad’) and four positive reinforcement items (‘drinking now would make the good things in my life appear even better’). These items, but particularly those in the positive reinforcement scale, reflect anticipated liking for alcohol. For the purpose of the present analysis the positive and negative reinforcement items were computed as separate scales, with alpha coefficients of 0.88 and 0.96, respectively. The items comprising the three scales are shown in Appendix I.
Excessive drinking The number of units of alcohol consumed per week by each participant was calculated by multiplying the number of units usually drunk on any one occasion by the usual number of drinking sessions in any one week. Four levels of drinking were determined, using the Department of Health Guidelines in force at the time the study was conducted, which defined sensible drinking levels as up to 14 units of alcohol per week for women and up to 21 units per week for men. (The old guidelines were used because they are easier to translate into different levels of excessive drinking than the current guidelines, which are based on both volume and pattern of drinking.) The four levels were defined as shown in Table 1. (The table also shows the number of participants in each group.) The ‘Extreme’ classification is equivalent to a daily intake of greater than a whole bottle of wine (6 units) in women or 1.5 bottles of wine (9 units) in men. Actual median intakes in this group were 60 units per week (8.6 units/day) in women and 80 units per week (11.4 units/ day) in men.
Statistical analyses For all three DAQ-derived scales, unweighted scale scores were derived by summing the scores on each item and dividing by the number of items. These scores were analysed by two-way analyses of variance on the factors level of drinking and gender; however, as all of the level–gender interactions were non-significant (P > 0.05), the data
Table 1 Levels of drinking as defined for this study, based on the Department of Health Guidelines for Moderate Drinking in force at the time the study was carried out.
Moderate drinking Excessive drinking Heavy drinking Extreme drinking
Women
n
Men
n
0–14 units 15–28 units 29–42 units >42 units
76 37 11 7
0–21 units 22–42 units 43–63 units >63 units
44 40 13 10
© 2005 Society for the Study of Addiction
for women and men were combined for presentation. Dunnett’s test was used to compare scores in the three excessive drinking groups to those of the sensible drinking group.
Amphetamine study Participants A total of 174 participants were recruited using a variety of locations (see below), of whom 164 provided the data used in the present analyses. The data were collected from a convenience sample of volunteers who had taken amphetamine (‘speed’) within a 4-week period prior to completing the questionnaire. The sample comprised 123 males and 46 females (five participants did not provide this information) with participants’ ages ranging from 14 to 45 years (mean = 23.26, SD = 6.05). Approximately one-third of the sample were collected from the Swansea drug agency’s ‘drop-in centre’, with the remainder being collected from a number of small towns in the Merthyr and Swansea valleys. A small number of participants allowed the use of their homes as a meeting point, while other participants completed the questionnaire on an outdoor park bench, in a local pub or in the researcher’s car. Participants were asked to complete the LDQ, the DSQ and a number of questions relating to their demographic details and their experience with drugs.
The DSQ The initial analyses of this data set are described elsewhere (James et al. 2004). Briefly, participants were prompted on a Likert-type scale, with scores ranging between 1 and 7, as to how strongly they agreed or disagreed with each of 40 statements, following which a principal components analysis was used to derive a factor structure for the DSQ. This analysis produced four factors, which were labelled ‘positive and negative reinforcement’, ‘strong desires and intentions to use’, ‘mild desires and intentions to use’ and ‘control over drug use’. For the purposes of the present analysis factor 2, ‘strong desires and intentions’, was taken as a measure of amphetamine ‘wanting’. This factor contained 10 items, all of which reflect craving for amphetamine (e.g. ‘my desire for speed now seems overwhelming’), none of which makes any reference to the consequences of using amphetamine. Factor 1 of the DSQ contained eight items, of which there were seven reflecting negatively reinforcing consequences of using amphetamine (e.g. ‘even major problems in my life would not bother me now if I took some speed’) and only one positively reinforcement item (‘taking speed now would make things seem just perfect’). Addiction, 100, 1487–1495
‘Wanting’ and ‘liking’ for alcohol and amphetamine
However, there were a further eight items reflecting consequences of using amphetamine that also loaded onto this factor with a loading >0.4, the criterion for inclusion, but were excluded because they also crossloaded onto other factors (Table 2 in James et al. 2004); seven of these items reflected positively reinforcing consequences (e.g. ‘taking speed now would make me feel on top of the world’), and there was a single negative reinforcement item (‘taking speed now would make me feel less tense’). For the purposes of the present analysis, these 16 items were used to create two eight-item scales, made up of positively and negatively reinforcing items. One of the positive items (‘nothing would be better than taking some speed now’) was ambiguous because the equivalent item in the DAQ loaded onto a different factor (see above); this item was therefore deleted, leaving seven items in this scale. Alpha coefficients for the two scales were 0.90 and 0.92, respectively. These items all refer simply to the consequences of using amphetamine, and none of them includes any reference to desires or intentions to use. These items, but particularly those in the positive reinforcement scale, reflect anticipated liking for amphetamine. The items comprising the three scales are shown in Appendix II. The LDQ The LDQ is a self-completion questionnaire containing 10 items, each of which is scored on a scale of 0–3, yielding a dependence score between 0 and 30. For the purposes of analysis, participants were divided into groups (n = 18–36) corresponding to six score ranges (0–5, 6– 10, 11–15, 16–20, 21–25, 26–30). Time since last drug use The questionnaire also included an item on time since last drug use. It was predicted that drug ‘wanting’ should vary as a function of time since last drug use; however, if ratings of anticipated positive and negative reinforcement are based on a memory of the experienced effects of the drug, then these variables should be independent of time since last drug use. Statistical analyses For all three DSQ-derived scales, unweighted scale scores were derived by summing the scores on each item and dividing by the number of items. These scores were analysed by one-way analyses of variance on the factors LDQ score range and time since last drug use. In the LDQ analyses, Dunnett’s test was used to compare scores in the five more dependent groups to those of the least dependent group. © 2005 Society for the Study of Addiction
1491
RESULTS Alcohol study Positive and negative reinforcement scores were strongly correlated (r = 0.91, P < 0.001), and both were strongly correlated with ‘wanting’ scores (r = 0.77, 0.68, respectively, P < 0.001). As shown in Fig. 1, scores on all three scales increased monotonically as a function of dependence (wanting: F3,223 = 11.19, P < 0.001; positive reinforcement: F3,230 = 10.51, P < 0.001; negative reinforcement: F3,219 = 9.27, P < 0.001). The rise was somewhat steeper for the positive reinforcement scale, as significant increases were seen in the ‘excessive drinking’ group, whereas for the wanting and negative reinforcement scales the increase was significant only in the ‘heavy drinking’ and ‘extreme drinking’ groups. In order to examine whether there might be a subgroup of heavy drinkers with high ‘wanting’ but low ‘liking’ scores, the difference between scores on the ‘wanting’ and ‘positive reinforcement’ scales was computed for each individual. The range of these difference scores was -4.50 to +1.11, with 70% of participants falling within the range -1.0 to +1.0. Seventy individuals had scores outside this range, of whom 69 had negative scores of (i.e. ‘wanting’ < ‘liking’); only a single individual had a score that was marginally greater than +1.0 (‘wanting’ > ‘liking’). The data provide no evidence that ‘liking’ for alcohol decreases in very heavy drinkers. Amphetamine study Positive and negative reinforcement scores were strongly correlated (r = 0.80, P < 0.001), and both were modestly correlated with ‘wanting’ scores (r = 0.58, 0.54, respectively, P < 0.001). As shown in Fig. 2, scores on all three scales increased monotonically as a function of dependence (wanting: F5,158 = 13.44, P < 0.001; positive reinforcement: F5,159 = 9.62, P < 0.001; negative reinforcement: F5,159 = 7.93, P < 0.001). The rise was somewhat steeper for the reinforcement scales, as significant increases were seen in the 11–15 range of LDQ scores, whereas for the wanting scale the increase was not significant until the 16–20 range of LDQ scores. Essentially the same picture was seen when the analysis was restricted to those individuals (n = 41) who reported that amphetamine was their main drug (Table 2). Correlations between ‘wanting’ and reinforcement scores were a little higher in this subgroup (r = 0.77 and 0.61 for positive and negative reinforcement, respectively, P < 0.001) and, again, monotonic increases in all three measures were seen with increasing LDQ scores (F5,35 = 12.63, 12.67 and 4.90, P < 0.002, for wanting, positive reinforcement and negative reinforcement, respectively). Addiction, 100, 1487–1495
1492
P. Willner et al.
5
* 3
***
* *
2
Scale score
4
4
Scale score
5
***
+ve reinforcement
1
*** 3
*
2 1
Wanting
-ve reinforcement 0
0 Moderate Excessive
Heavy
Moderate Excessive
Extreme
Heavy
Extreme
Level of drinking
Level of drinking
Figure 1 Anticipated reinforcement from drinking alcohol (left panel) and ‘wanting’ for alcohol (right panel), as a function of extent of excessive drinking. *P < 0.05, ***P < 0.001, relative to moderate drinking group
*** ***
7
*
5 4 3
*
2
**
*** ***
+ve reinforcement
1
***
6
-ve reinforcement
0
Scale score
6
Scale score
7
***
***
5 4
*
3 2 Wanting
1 0
0–5
6–10 11–15 16–20 21–25 25–30
LDQ range
0–5
6–10 11–15 16–20 21–25 25–30
LDQ range
Figure 2 Anticipated reinforcement from using amphetamine (left panel) and ‘wanting’ for amphetamine (right panel), as a function of extent of dependence. *P < 0.05, **P < 0.01, ***P < 0.001, relative to 0–5 group; n = 18–36 per group.
Table 2 Mean (± SEM) ‘wanting’ and ‘liking’ scores in individuals for whom amphetamine was their main drug. Reinforcement LDQ range
n
Wanting
Positive
Negative
0–5 6–10 11–15 16–20 21–25 26–30
6 8 6 4 10 7
1.33 (± 0.52) 2.79 (± 0.45) 2.25 (± 0.52) 4.70 (± 0.63) 4.68 (± 0.40) 5.93 (± 0.48)
3.00 (± 0.40) 4.16 (± 0.35) 4.69 (± 0.40) 5.46 (± 0.49) 6.13 (± 0.31) 6.57 (± 0.37)
2.52 (± 0.60) 3.34 (± 0.52) 3.83 (± 0.60) 4.06 (± 0.74) 5.61 (± 0.47) 5.38 (± 0.56)
© 2005 Society for the Study of Addiction
In order to examine whether there might be a subgroup of dependent individuals with high ‘wanting’ but low ‘liking’ scores, the difference between scores on the ‘wanting’ and ‘positive reinforcement’ scales was computed for each individual. The range of these difference scores was -5.90 to +2.64, with the vast majority of subjects (91%) showing negative scores (i.e. ‘wanting’ < ‘liking’). Within the most dependent group of participants (LDQ range 26–30, n = 18) there were 10 individuals for whom the difference score was in the range -1 to +1, and eight who had larger differentials. However, in all eight, ‘liking’ scores were larger than ‘wanting’ scores, rather than smaller. Within the whole sample, there were only three individuals whose ‘wanting’ for amphetamine exceeded their ‘liking’ for amphetamine by more than Addiction, 100, 1487–1495
‘Wanting’ and ‘liking’ for alcohol and amphetamine
1493
Table 3 Mean (± SEM) ‘wanting’ and ‘liking’ scores as a function of time since last drug use. Reinforcement Time since last use
n
Wanting
Positive
Negative
1–3 hours 4–12 hours 13–24 hours 25–48 hours 3–7 days >7 days
59 28 26 14 19 16
3.09 (± 0.22) 3.41 (± 0.31) 2.36 (± 0.33) 1.81 (± 0.44) 3.10 (± 0.39) 2.50 (± 0.42)
4.67 (± 0.21) 5.18 (± 0.29) 5.01 (± 0.31) 4.47 (± 0.41) 4.59 (± 0.35) 4.36 (± 0.40)
3.98 (± 0.21) 4.90 (± 0.31) 4.00 (± 0.32) 3.84 (± 0.44) 4.01 (± 0.38) 4.00 (± 0.41)
one unit: their LDQ scores were 2, 7 and 25. In short, we could find no evidence for the existence of a subgroup of dependent individuals who ‘wanted’ but did not ‘like’ amphetamine. Scores on the three scales as a function of time since last drug use are shown in Table 3. ‘Wanting’ for amphetamine varied significantly as a function of time since last drug use (F5,155 = 2.66, P < 0.025) with the lowest levels of ‘wanting’ at intermediate intervals: that is, wanting was high during the period of drug intoxication and after several days of abstinence. However, scores for positive and negative reinforcement (i.e. the anticipated effects of the drug) did not vary significantly as a function of time since last drug use (F5,155 = 0.88, 1.48, respectively, not significant). In order to establish whether this different effect of time since last use on ‘wanting’ and reinforcement was significant, a further analysis of variance was conducted in which the three scales were treated as repeated measures. This analysis confirmed a significant scale–time interaction (F10,300 = 1.98, P < 0.05), and a significant quadratic component to the interaction (F5,150 = 2.69, P < 0.025), confirming that the difference between the scales was greatest at intermediate time intervals.
DISCUSSION Our re-analyses indicate that subjective ratings of craving for, and anticipated liking of, alcohol and amphetamine both increased monotonically with indicators of dependence. Thus, we found no evidence for a dissociation between subjective craving for, and anticipated liking of, drugs as a function of the development of dependence or degree of exposure to the drug. These results directly contradict one of the assumptions of incentive sensitization theory (Robinson & Berridge 1993; 2001) that subjective measures of drug ‘wanting’ and drug ‘liking’ should become dissociated over the course of the development of drug dependence in humans, such that ‘wanting’ increases while ‘liking’ ‘either does not change or else becomes hyposensitive’ (Robinson & Berridge 1993). © 2005 Society for the Study of Addiction
Our results might be considered to be at variance with the findings of Volkow et al. (1997) who reported that cocaine addicts showed a reduced methylphenidate-induced ‘high’ relative to non-dependent controls. However, as indicated earlier, there are a number of problems with the latter study in relation to using these data to test the incentive sensitization theory: it measured drug-induced ‘high’ rather than drug liking; the observations were made well after the peak of the ‘high’ had subsided; the ‘high’ produced by one drug (methylphenidate) was compared with the effect of administration of that drug on ‘wanting’ (craving) for another (cocaine); and the controls were non-drug-abusers, rather than recreational but non-addicted cocaine users. It should also be noted that the cocaine addicts had all just emerged from a treatment programme. If it is true that there is a dissociation between wanting and liking in this group (which the study has not actually shown), it may be that this dissociation is associated with the decision to enter treatment, rather than with addiction per se. That is, it may be that addiction in general involves increased wanting and liking but if, for some reason, the liking system were to become disabled, then this could cause addicts to view their situation as problematic and so to seek treatment. Finally, yet another possibility is that the reduced high in the study of Volkow et al. (1997) could be a consequence of treatment, given that the addicts were several weeks post-detoxification. There are five limitations to our present findings. First, the methodology may not be adequate to detect differences between ‘wanting’ and ‘liking’. For example, it is possible that dissociations between ‘wanting’ and ‘liking’ might be discernible in more dependent individuals. We note, however, that some participants in the alcohol study had extremely high daily intakes, and some participants in the amphetamine study had dependence scores close to the maximum measurable with the LDQ. It is also possible that our scales fail to separate ‘wanting’ and ‘liking’ adequately. However, while it is true that these scales were quite highly correlated, they were derived Addiction, 100, 1487–1495
1494
P. Willner et al.
empirically by factor analysis, and the items were retained on the basis that ‘wanting’ and ‘liking’ items did not cross-load. At least in the case of amphetamine the two scales do measure different things, as they differed in their behaviour as a function of time since last drug use. Secondly, the data were derived from a retrospective reanalysis of existing data. However, there is no reason to believe that participants would not have behaved in an identical fashion in similar alcohol and amphetamine studies that had set out to test the incentive sensitization theory a priori—given that the participants would remain ignorant of the specific hypotheses under investigation. In fact, we have recently completed two prospective studies of alcohol craving that essentially replicate the present data, in showing monotonic increases in both ‘wanting’ and ‘liking’ for alcohol, using the present DAQderived measures, as a function of increasing dependence; in these studies, dependence was measured using the LDQ rather than, as in the present alcohol study, relying on measures of consumption (Morgan et al. in preparation). Thirdly, the study is based on responses to questionnaires items measuring ‘wanting’ and ‘liking’, and different findings might emerge if behavioural measures were used. Fourthly, it is acknowledged that the present studies were not measuring drug ‘liking’ directly but were instead measuring ‘anticipated liking’. Nevertheless, we consider it unlikely that anticipated drug liking in experienced drug users would vary much from their self-reported liking while actually taking the drug in question, particularly where very heavy users are concerned. Support for this position is provided by the finding that unlike drug ‘wanting’, which varied as a function of time since last drug use, positive and negative reinforcement scores did not vary significantly as a function of time since last use. This suggests that participants’ ratings of the anticipated effects of amphetamine are based accurately on their memory of the actual effects of the drug. (Unfortunately, comparable data are not available for alcohol.) Nevertheless, the question of how accurately anticipated liking reflects actual liking is one that should be addressed directly. Finally, it is possible that participants may have experienced a response bias towards maintaining consistency between self-reported ‘wanting’ and ‘liking’ responses. A bias towards consistency could in principle contribute to our failure to observe any evidence of dissociation of drug ‘wanting’ and ‘liking’ as a function of the development of drug dependence in the present study. However, this criticism, if valid, would apply to all studies using multi-dimensional questionnaires to investigate craving. All these issues are empirical questions that should be investigated in more depth in future prospective studies. The present results cast doubt on the assumption that ‘liking’ for drugs and alcohol decreases during the © 2005 Society for the Study of Addiction
development of addiction. This in turn questions the significance of increases in ‘wanting’ for drugs and alcohol as the primary driver of addictive behaviours, at least in non-clinical samples of alcohol and drug users. The incentive sensitization theory was introduced to explain the supposed fact that ‘as drugs come to be wanted more and more, they often come to be liked less and less’ (Robinson & Berridge 1993, p. 249). However, the present data provide no support for the proposition that with increasing dependence, drugs ‘come to be liked less and less’. On the contrary, the results are consistent with the simpler hypothesis that addiction reflects an increase in the positive and/or negative reinforcing properties of addictive drugs. The present data do not refute the incentive sensitization theory directly: drug ‘wanting’ did increase with use, as predicted by the incentive sensitization theory, so the data are compatible with the position that this factor might underlie addiction. However, the study does question the validity of one of the most important ‘facts’ on which the incentive sensitization theory is based. Further work is needed to establish whether ‘wanting’ and ‘liking’ also increase in parallel in clinical samples, and to elucidate the significance of the findings if they do not.
References Berridge, K. C. & Robinson, T. E. (1998) What is the role of dopamine in reward: hedonic impact, reward learning, or incentive salience? Brain Research Review, 28, 309– 369. Brauer, L. H., Cramblett, M. J., Paxton, D. A. & Rose, J. E. (2001) Haloperidol reduces smoking of both nicotine-containing and denicotinized cigarettes. Psychopharmacology (Berlin), 159, 31–37. Brauer, L. H., Goudie, A. J. & De Wit, H. (1997) Dopamine ligands and the stimulus effects of amphetamine: animal models versus human laboratory data. Psychopharmacology (Berlin), 130, 2–13. De Wit, H. (2000) Laboratory-based assessment of alcohol craving in social drinkers. Addiction, 95, S165–S169. Fischman, M. W. & Foltin, R. W. (1992) Self-administration of cocaine by humans: a laboratory perspective. Ciba Foundation Symposium, 1992, 165–173; discussion 173–80. James, D., Davies, G. & Willner, P. (2004) The development and initial validation of a questionnaire to measure craving for amphetamine. Addiction, 99, 1181–1118. Lamb, R. J., Preston, K. L., Schindler, C. W., Meisch, R. A., Davis, F., Katz, J. L. et al. (1991) The reinforcing and subjective effects of morphine in post-addicts: a dose–response study. Journal of Pharmacology and Experimental Therapeutics, 259, 1165–1173. Lasagna, L., von Felsinger, J. M. & Beecher, H. K. (1955) Druginduced mood changes in man. I. Observations on healthy subjects, chronically ill patients, and postaddicts. JAMA, 157, 1006–1020. Leyton, M., Boileau, I., Benkelfat, C., Diksic, M., Baker, G. & Dagher, A. (2002) Amphetamine-induced increases in extracellular dopamine, drug wanting, and novelty seeking: a PET/ Addiction, 100, 1487–1495
‘Wanting’ and ‘liking’ for alcohol and amphetamine 11
1495
C raclopride study in healthy men. Neuropsychopharmacology, 27, 1027–1035. Love, A., James, D. & Willner, P. (1998) A comparison of two alcohol craving questionnaires. Addiction, 93, 1091–1102. McAuliffe, W. E. (1975) A second look at first effects: the subjective effects of opiates on nonaddicts. Journal of Drug Issues, 6, 369–399. McAuliffe, W. E. & Gordon, R. A. (1974) A test of Lindensmith’s theory of addiction: the frequency of euphoria among longterm addicts. American Journal of Sociology, 79, 795–840. Robinson, T. E. & Berridge, K. C. (1993) The neural basis of drug craving: an incentive-sensitization theory of addiction. Brain Research Reviews, 18, 247–291. Robinson, T. E. & Berridge, K. C. (2001) Incentive sensitization and addiction. Addiction, 96, 103–114. Robinson, T. E. & Berridge, K. C. (2003) Addiction. Annual Review of Psychology, 54, 25–53. Singleton, E. G. & Tiffany, S. T. (1994) Alcohol Craving Questionnaire: ACQ-Now: Background and Administration Manual. Baltimore: NIDA Addiction Research Centre.
Tiffany, S. T. & Drobes, D. J. (1991) The development and initial validation of a questionnaire of smoking urges. British Journal of Addiction, 86, 1467–1476. Tiffany, S. T., Singleton, E., Haertzen, C. A. & Henningfield, J. E. (1993) The development of a cocaine craving questionnaire. Drug and Alcohol Dependence, 34, 19–28. Volkow, N. D., Wang, G. J., Fowler, J. S., Logan, J., Gatley, S. J., Hitzemann, R. et al. (1997) Decreased striatal dopaminergic responsiveness in detoxified cocaine-dependent subjects. Nature, 386, 830–833. Volkow, N. D., Wang, G. J., Fowler, J. S., Logan, J., Jayne, M., Franceschi, D. et al. (2002) ‘Nonhedonic’ food motivation in humans involves dopamine in the dorsal striatum and methylphenidate amplifies this effect. Synapse, 44, 175– 180. Wachtel, S. R., Ortengren, A. & de Wit, H. (2002) The effects of acute haloperidol or risperidone on subjective responses to methamphetamine in healthy volunteers. Drug and Alcohol Dependence, 68, 23–33.
APPENDIX I Scales derived from the DAQ.
APPENDIX II Scales derived from the DSQ.
Strong desires and intentions to use alcohol (wanting) 03 I am missing having a drink now 06 I need a drink now 07 My desire to drink now seems overwhelming 13 I would do almost anything to have a drink now 15 I crave a drink now 24 I have an urge to drink now 25 I want a drink so much I can almost taste it 29 I am thinking of ways to get alcohol 32 I will have a drink now whatever gets in the way 34 I am going to drink as soon as I possibly can
Strong desires and intentions to use amphetamine (wanting) 07 I need some speed now 08 My desire for speed now seems overwhelming 11 I am making plans to take some speed now 15 I would do almost anything to take some speed now 17 I crave some speed now 26 I have an urge to take speed now 27 I want some speed so much I can almost feel it 32 I am thinking of ways to get speed 35 I will have some speed now whatever gets in the way 38 I am going to have some speed as soon as I possibly can
Positive reinforcement from using alcohol (liking) 02 Drinking now would make the good things in my life appear even better 10 Drinking now would make me feel on top of the world 21 Drinking would make me feel good 33 Drinking now would make things seem just perfect Negative reinforcement from using alcohol 04 It would feel as if the bad things in m life had completely disappeared if I drank now 08 Even major problems in my life would not bother me now if I drank 11 Drinking now would make me feel less tense 14 Drinking now would make the bad things in my life seem less bad 22 If I drank now the small daily hassles would feel less important 27 I would probably feel less worried about my daily problems if I drank now 31 Drinking would make me feel less stressed 36 All my tension would completely disappear if I drank now
© 2005 Society for the Study of Addiction
Positive reinforcement from using amphetamine (liking) 02 Taking speed now would make the good things in my life appear even better 12 Taking speed now would make me feel on top of the world 14 Taking speed would be satisfying now 20 Taking some speed would be wonderful now 23 Taking some speed now would make me feel good 29 Taking speed would be pleasant now 37 Taking speed now would make things seem just perfect Negative reinforcement from using amphetamine 04 It would feel as if the bad things in my life had completely disappeared if I took some speed now 10 Even major problems in my life would not bother me now if I took some speed 13 Taking speed now would make me feel less tense 16 Taking speed now would make the bad things in my life seem less bad 24 If I took some speed now the small daily hassles would feel less important 30 I would probably feel less worried about my daily problems if I took some speed now 34 Taking speed now would make me feel less stressed 40 All my tension would completely disappear if I took some speed now
Addiction, 100, 1487–1495
