Observational learning of instrumental ... - Semantic Scholar

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY 2006, 59 (11), 1909 –1920

Observational learning of instrumental discriminations in the rat: The role of demonstrator type A. L. Saggerson and R. C. Honey Cardiff University, Cardiff, UK

In two experiments, observer rats saw a pretrained demonstrator rat of either the same or a different strain engaging in a discrimination task in which the presentation of a discriminative stimulus indicated whether performing a particular response (pulling a chain) would be reinforced. In both experiments an effect of demonstrator familiarity was found: Observers of a demonstrator from a different strain behaved in a manner that was consistent with the demonstrator whereas observers of a demonstrator from the same strain did not. These results suggest that an effect akin to latent inhibition operates in the social domain: Familiarity with the demonstrator retards the readiness with which observational learning proceeds.

The issue of whether nonhuman animals exhibit anything akin to imitative learning has a venerable history (e.g., Heyes, 2001; Miller & Dollard, 1941; Romanes, 1884; Thorndike, 1911; Thorpe, 1956) and has been the focus of an increasing amount of experimental investigation. For example, Heyes and her colleagues have recently adopted a similar procedure to that used by Grindley (1932), the “bi-directional control”, to investigate observational learning in rats (Heyes & Dawson, 1990; Heyes, Dawson, & Nokes, 1992; Heyes, Jaldow, & Dawson, 1994). In these studies, demonstrator rats were first trained to push a joystick to the left or to the right in order to obtain food. Observer rats were then placed in a chamber that faced the demonstration chamber and could view a demonstrator pushing the

joystick to either the left or the right. When the observers were subsequently placed in the demonstration chamber they pushed the joystick in the same direction as the demonstrator; for example, an observer who had witnessed a demonstrator pushing the joystick to the right were more likely to push the joystick in that direction when placed in the demonstration chamber. The observers appeared to have learnt to perform a specific response as a result of being exposed to a demonstrator performing that response. It has recently become apparent, however, that the observers in these studies were preferentially pushing the joystick in the same direction as the demonstrator because the demonstrator had deposited odour cues (e.g., saliva, microscopic food particles) on the side of the joystick contralateral to the

Correspondence should be addressed to A. L. Saggerson or R. C. Honey, School of Psychology, Cardiff University, Cardiff CF10 3AT, UK. E-mail: [email protected] We thank Dennis Simmonds and Howard Thomas for their technical assistance and Rachel Gazey for helping to score the videotapes. # 2006 The Experimental Psychology Society http://www.psypress.com/qjep

1909 DOI:10.1080/17470210600705032

SAGGERSON AND HONEY

direction in which is was being pushed. If one makes the reasonable assumption that observers will be drawn to the side of the joystick on which odour cues had been deposited, then a secondary consequence of investigating this side of the joystick might be to increase the likelihood that the joystick will be moved in the direction that the demonstrator had pushed the joystick (see Mitchell, Heyes, Gardner, & Dawson, 1999). Further studies that have controlled for the influence of odour cues (e.g., by swapping chambers between observation and test, Campbell & Heyes, 2002; Ray & Heyes, 2002) have now indicated that in the absence of the odour-cue confound observer rats exhibit either demonstrator-inconsistent responding (pushing the joystick in the opposite direction to the demonstrator) or no observational learning effects (see also, Ray, Gardner, & Heyes, 2000). Given the fact that there remains little evidence of observational learning in rats, the experiments reported here examined whether or not such evidence can be secured and, if it can, then what conditions foster it. In particular, we were interested in whether the nature of the demonstrator (the relatedness, or visual and olfactory familiarity to the observer) influences the likelihood that an observer’s behaviour would come to match or resemble that of a demonstrator. A feature of standard laboratory practice is that rats are usually housed socially and most often in pairs. One consequence of this arrangement is that the rats used in studies of observational learning will usually have had a considerable amount of exposure to conspecifics of the same strain as the demonstrator prior to observational training. They will also have received food at the same time as their cage mates and engaged in synchronous feeding-related behaviours with them. This fact might well influence the outcome of observational learning studies. However, to date there has been surprisingly little research directed toward whether or not the nature of the demonstrator (e.g., demonstrator familiarity) influences the effects of observational experience in rats. In one study, Ray et al. (2000) manipulated the familiarity of the demonstrator in a study of

1910

observational learning. Observer rats either were housed together with their prospective demonstrators prior to observation training and testing (and hence the demonstrator was familiar), or were housed separately and first encountered their demonstrator during the first observation session. At test, observers pushed the joystick in the same direction as their demonstrator had done irrespective of whether their demonstrator had been familiar or novel. That is, there was no effect of demonstrator familiarity. Interpretation of this pattern of results is, however, complicated by the possibility that the observational learning effect was a product of odour cues deposited on the joystick (see Mitchell et al., 1999). Using the same procedure as Ray et al. (2000), Reed, Skiera, Adams, and Heyes (1996) examined whether social rearing influenced observational learning. In this study, half of the observer rats were housed socially, and the other half were housed in isolation. Socially reared observers (those for whom the demonstrator might be presumed to be more familiar) were more likely to exhibit demonstrator-consistent behaviour than those reared in isolation (which tended to show demonstrator-inconsistent behaviour). However, it is difficult to know whether this finding reflects a difference in the familiarity of the demonstrator or some other effect of social isolation (see Reed et al., 1996, p. 113). Moreover, interpretation of the effect is again complicated by the possibility that odour cues deposited on the manipulandum might have contaminated test performance to an unknown extent in the two groups of observers (see Mitchell et al., 1999). In the two experiments reported here, all rats were socially reared with conspecifics of the same strain, and we manipulated whether demonstrators were from the same strain as the observers and the observers’ cage mates (for group same) or from a different strain than the observers (for group different). Given the reasonable assumption that there would be more generalization between a cage mate and a demonstrator of the same strain than between a cage mate and a demonstrator of different strain, we supposed that the demonstrators would be more familiar to observers in

THE QUARTERLY JOURNAL OF EXPERIMENTAL PSYCHOLOGY, 2006, 59 (11)

OBSERVATIONAL LEARNING

group same than to those in group different. Of course, we could have manipulated familiarity more directly by (a) giving observer rats more or less exposure to a demonstrator, or (b) housing observers with rats of either the same strain or a different strain and then using demonstrators that match or do not match the strain of the observers’ cagemates. However, both of these experimental strategies have limitations. For example, rats are routinely weaned with dams of the same strain and with same-strain conspecifics, and this might compromise or otherwise dilute the subsequent efficacy of either of the alternative manipulations just considered. We therefore chose to attempt to maximize the potential effect of demonstrator type by simply manipulating whether or not the demonstrators were the same strain as that of the observers, the observers’ cagemates, and conspecifics with which they had been weaned. In both experiments, the observers received exposure to demonstrators that were performing a discrimination in which a specific response was rewarded in the presence of one discriminative stimulus but not another. During test sessions, observers were placed in the demonstration chamber, and their behaviour in the presence of the two discriminative stimuli was monitored. The issues of central interest were whether or not the observers’ behaviour during the discriminative stimuli would be influenced by (a) how the demonstrators behaved during these stimuli, and (b) the nature of the demonstrator. If our procedures reveal an observational learning effect then this effect will not be subject to the analysis that has been offered for previous reports of observational learning in rats by Heyes and her colleagues (see Campbell & Heyes, 2002; Heyes & Dawson, 1990; Heyes et al., 1992; Heyes et al., 1994; Mitchell et al., 1999; Ray et al., 2000; Ray & Heyes, 2002). This is because in our procedure, it is the auditory discriminative stimuli rather than the manipulanda that control the observer’s (and demonstrator’s) behaviour, and this fact eliminates the possibility that odour cues deposited on the manipulandum could produce an observational learning effect. If the type of demonstrator modulates observational

learning then this would inform theoretical analyses of observational learning that to date have been relatively unconstrained by empirical observations.

EXPERIMENT 1 Dark Agouti (DA; brown in colour) and Hooded Lister (HL; black and white in shade) rats were used. The DA and HL demonstrators were trained on an instrumental discrimination in which pulling one chain (positioned on the left side of the experimental chamber) was reinforced during the presentation of one discriminative stimulus (the left Sþ; e.g., clicker), and pulling a second chain on the right side of the chamber was reinforced during another stimulus (the right Sþ; e.g., tone). Once the demonstrators had acquired this discrimination, observers were placed in an operant chamber from which they could view a demonstrator in the adjacent, demonstration chamber, engaging in the discrimination task. One group of DA and HL rats (those in group same) observed a demonstrator of the same strain, and the other group of rats (those in group different) observed a demonstrator of a different strain. Following observation sessions, the observers received a test in which they were placed in the demonstration chamber and received presentations of the two discriminative stimuli in the absence of the demonstrator and with free access to the chains. Pilot work established that rats that had received no rewards for chain pulling failed to pull them with sufficient force to actuate the microswitch to which they were attached. It was, therefore, impossible to record the observers’ behaviour using the automated system used for recording the demonstrators’ chain pulling. The observers’ behaviour during the test was therefore video-recorded, and the number of chain pulls (or contacts) that they made in the presence of the left Sþ and the right Sþ was monitored. If rats are capable of learning an instrumental discrimination by observing a demonstrator rat, then presenting the left Sþ to the observer should result in them pulling the


1911

SAGGERSON AND HONEY

left chain, and presenting the right Sþ should result in them pulling the right chain.

Method Subjects There were 24 rats in this experiment (12 HL and 12 DA). The 4 demonstrator rats (2 HL and 2 DA) had previously served as observers in another experiment on observational learning, but the 20 observers were naı¨ve and were maintained throughout the experiment at 80% of their adlib weight (M ¼ 235 g, range ¼ 206– 256 g). The observer and demonstrator rats were housed in same-strain pairs in a vivarium, which had a 12:12-hr light:dark cycle. Experimental sessions took place during the light phase of the cycle, between 0900 and 1930. The 20 rats that served as observers were housed in pairs, for 9 weeks before the experiment began. Apparatus All training and testing occurred in four operant chambers (supplied by Campden Instruments Ltd, UK) measuring 26 25 24 cm (height width depth); two served as demonstration chambers and two as observation chambers. Both demonstration chambers had a grid floor composed of 18 stainless steel rods, which were 5 mm in diameter and were spaced 11 mm apart. Below this floor was a removable tray (4 25 24 cm; height width depth), which was filled with sawdust. Three walls of the demonstration chambers (the left and right side walls and the back wall) were made of aluminium. The front wall, which served as the door to the chamber, was made of transparent Perspex and allowed observers to view the demonstrator. The ceiling was made from opaque plastic. The wall to the left of the door contained a recessed food well into which 45-mg food pellets were delivered. Access to this food well was obscured by a transparent plastic flap (6 5 cm; height width) hinged along its uppermost edge to the top of the food well aperture. Two 8-ohm speakers attached to the ceiling of the chambers were used to present a 1,000-Hz tone at 65 dB and a

1912

10-Hz clicker at 75 dB. The main overhead light in the experimental room was turned off, but an angle-poise lamp (with a 60-W bulb) positioned above the demonstration chambers provided a low level of ambient illumination. Local illumination was provided by an overhead houselight (2.8 W, 24 V) inside each chamber. Two brass-plated chains (2-mm gauge, 1-cm width, 20-cm length) were suspended through small apertures (2-cm diameter) in the ceiling of the demonstration chamber, and each rested 2 cm from the floor of the chamber. The chains were positioned 3.5 cm inside the door and 5.9 cm from each end of the door. The “left” chain was designated as the one closest to the food well, and the “right” chain was designated as the one furthest from the food well. A downward movement of either chain of 2 mm actuated a microswitch that was recorded as a response by automated software. During observation training, an observation chamber was placed adjacent and approximately 8 cm from each demonstration chamber. The two observation chambers were identical to the demonstration chambers with the exception that they contained no chains, and the rear wall had been removed and replaced with a wall of transparent Perspex. This arrangement allowed the observer to view the demonstrator from the observation chamber and meant that the observation chamber was oriented in the same way as the demonstration chamber (i.e., with the food well on the lefthand side of the chamber). The only other respects in which the observation chambers differed from the demonstration chambers were that the food magazine, houselight, and overhead speakers were not operational. Procedure The demonstrators were first trained to collect pellets (Harlan Teklan, UK) from the food well. In the first six 20-min sessions of training, food pellets were delivered on a variable-time (VT) 40-s schedule (range: 20– 60 s). The chains were not present during these sessions. During the next six 20-min sessions, a food pellet was delivered on a VT 60-s schedule and each time a chain was pulled (i.e., on a continuous



reinforcement schedule). To encourage the rats to pull the chains during these sessions, food pellet paste (food pellets mixed with water and left until a paste formed) was smeared on the chains, and training continued until the demonstrators were reliably chain pulling. After these 6 sessions, the demonstrators received 10 sessions of training in which the presentation of reinforcement was contingent on pulling the right chain during one stimulus (the left Sþ) and the left chain during the other stimulus (the right Sþ). For one of the HL and one of the DA demonstrators, the clicker served as the left Sþ, and the tone served as the right Sþ, and for the remaining two demonstrators this arrangement was reversed. In each session, demonstrators received ten 60-s presentations of each discriminative stimulus, which were presented according to a pseudorandom sequence, with the constraint that there were no more than two presentations of a given stimulus in succession. No food could be earned during the first 20-s of each trial, but food was made available on a variable-interval (VI) schedule for the remaining 40 s of each trial. The demonstrators’ behaviour during the first 20-s reinforcer-free periods of each trial allowed an assessment to be made of the discriminative control exercised by the auditory stimuli independently of the delivery of reinforcement. The VI schedule was gradually increased from 10 s, at the outset of training, to 30 s, by the end of training. The intertrial interval (ITI) was variable around a mean of 40 s (range: 20 –60 s). Observer training and testing. Each observer received two 20-min sessions of exposure to the demonstration chamber in the two days immediately prior to observation training. Although the chain was present in the chamber, no reinforcement was given for any chain pulls, and the discriminative stimuli were not presented. During observation sessions, the demonstrator was placed in the demonstration chamber, the observer was then placed in the observation chamber, and the demonstrator received a discrimination training session that was identical to those that it had received at the end of training.

For approximately half of the observers in groups same (2 HL and 3 DA) and different (2 HL and 3 DA), the demonstrator’s left Sþ was the tone, and the right Sþ was the clicker, and for the remaining observers the demonstrator received the reverse stimulus assignments. There was one observation session on each of two consecutive days, and observers received a test 2 min after the second session. Once the demonstrator had been removed from the demonstration chamber, the observer was introduced into this chamber and received two presentations of the tone (T) and the clicker (C), which were presented in a counterbalanced order (either CTTC or TCCT). No reinforcement was presented during the test sessions, and observers from the same and different groups were tested in a counterbalanced order. The observer rats’ behaviour during these test sessions was video-recorded, and their chain contacts monitored. A chain contact was defined as any contact with the chain made with the snout, head (up to and including the ears), or the paw that resulted in some movement of the chain. One person (Anna Saggerson, A.S.) scored all of the videotapes in Experiments 1 and 2 and a second person (Rachel Gazey, R.G.) scored a subset of the videotapes in Experiment 2. A.S. was blind with respect to the group allocations of individual rats during scoring (there was no information on the video footage regarding group allocations), and R.G. was blind with respect to the nature of the experiment and to group allocations. There was a high degree of concordance between scorers (Pearson’s rs .78, ps , .01) as is confirmed in greater detail when the results of Experiment 2 are presented.

Results Demonstrator performance during training and observation sessions The demonstrators’ chain-pulling responses during the two stimuli were converted into a discrimination ratio of the following kind: rate of left chain pulling during the stimulus in which this response was reinforced (left Sþ) divided by the total rate of left chain pulling during both stimuli.


1913

SAGGERSON AND HONEY

Table 1. Experiment 1: Mean discrimination ratios for demonstrators during the observation sessions Group Same Different

Left chain (20 s/60 s)

Right chain (20 s/60 s)

.65/.70 .66/.67

.75/.74 .68/.71

Note: The ratios are presented separately for the left and right chains and for the first 20 s of the 60-s trials (in which no food was presented) and the entire 60 s.

The corresponding ratio was also calculated for responses to the right chain. Using this ratio, scores above .50 indicate that rats are responding more on reinforced trials than on nonreinforced trials. The mean discrimination ratio for left and right chain pulling during the final two sessions of training was .63 (left chain ¼ .64, right chain ¼ .62). The mean rate of nonreinforced left chain pulling during the right Sþ was 3.40 rpm, and the corresponding rate of right chain pulling during the left Sþ was 3.92 rpm. The mean discrimination ratio in the 20-s reinforcer-free periods during these sessions was .66 (left chain ¼ .70, right chain ¼ .62); the mean rate of nonreinforced left chain pulling during the right S– was 4.43 rpm, and the corresponding rate of right chain pulling during the left S– was 5.55 rpm. The same demonstrators could serve as models for observers in both

groups same and different, and the demonstrators’ discrimination ratios for the first, nonreinforced 20 s of the trials and for the entire 60-s duration of the trials are shown in Table 1. Inspection of Table 1 reveals that these ratios did not vary according to the nature of the rats that they were being viewed by. The mean rates of nonreinforced left and right chain pulls from which these ratios were derived were also similar in the two groups (same M ¼ 5.14 rpm; different M ¼ 6.01). Observer behaviour during test sessions The observers’ rates of chain contacts were also converted into discrimination ratios of the kind used for their demonstrators. In this case, therefore, a score above .50 indicates that the observer is behaving in a way that is consistent with their demonstrator. The mean discrimination ratios for contacts with the right chain in groups same and different are shown in the right-hand panel of Figure 1, and inspection of this panel reveals that rats in group different were more likely to show demonstrator-consistent behaviour than those in group same. Analysis of variance (ANOVA) confirmed that the discrimination ratios for rats in group different were greater than those for group same, F(1, 18) ¼ 5.49, p , .05, although in this experiment neither group’s scores differed

Figure 1. Experiment 1: Mean chain contact discrimination ratios (+SEMs) for the left chain (left-hand panel) and right chain (right-hand panel) in groups same and different.

1914



significantly from .50: same, t(9) ¼ –1.48, p . .10; different, t(9) ¼ 1.91, p ¼ .09, respectively. The mean rate of right chain contacts for groups same and different during both stimuli were, respectively, 4.60 rpm and 5.70 rpm (right Sþ), and 5.90 rpm and 3.80 rpm (left Sþ). ANOVA with stimulus (right Sþ or left Sþ) and group (same or different) as factors revealed no effect of group or stimulus (Fs , 1), and no interaction between these factors, F(1, 18) ¼ 3.59, p ¼ .07. The mean discrimination ratios for the left chain contacts in groups same and different are shown in the lefthand panel of Figure 1. Inspection of this panel reveals, and statistical analysis confirmed, that the discrimination ratios for the left-hand chain were similar in groups same and different (F , 1) and that neither group’s scores differed from .50: same, t(9) ¼ –0.93, p . .30; different, t(9) ¼–0.41, p . .60. The mean rate of left chain contacts for groups same and different were, respectively, 4.40 rpm and 3.50 rpm (left Sþ), and 5.80 rpm and 5.10 rpm (right Sþ). ANOVA with stimulus (left Sþ or right Sþ) and group (same or different) as factors revealed no effect of group (F , 1) or stimulus, F(1, 18) ¼ 1.48, p . .24, and no interaction between these factors (F , 1).

Discussion The results of Experiment 1 suggest that the effects of observational experience differ depending on the nature of the demonstrator: Rats are more likely to behave in a demonstrator-consistent manner when their demonstrator is from a different strain than when the demonstrator is from the same strain. It should be acknowledged, however, that this pattern of results was restricted to behaviour directed towards the right chain, and the scores of neither group were significantly different from .50. The finding that the effect of primary interest was restricted to the right chain might reflect the fact that the demonstrator’s correct and incorrect left chain pull responses tended to be both temporally and spatially coincident with common magazine approach responses. From the observer’s perspective, this could render the demonstrators’ correct and

incorrect left chain responses less distinctive from one another than the demonstrators’ correct and incorrect right chain responses (which are spatiotemporally remote from magazine responses). The finding that the discrimination ratios for rats in group different were not greater than .50 might simply reflect the fact that the discrimination that they were witnessing was rather complex or that they had received relatively little observational experience. Before considering how best to interpret the effect of demonstrator type that was apparent in Experiment 1, we conducted a second study in which we attempted to replicate this effect using a simpler discrimination (involving a single, centrally positioned response chain), upon which the demonstrators received more protracted training, and in which the observers received more extensive observational training.

EXPERIMENT 2 DA and HL demonstrators received instrumental discrimination training in which pulling a chain (located in a central position adjacent to the door) was reinforced during the presentation of one discriminative stimulus (e.g., clicker; the Sþ), and pulling this chain was not reinforced during another stimulus (e.g., tone; the S –). Once the demonstrators had acquired this discrimination, observers were placed in an operant chamber from which they could view a demonstrator in the adjacent, demonstration chamber, engaging in the discrimination task. As in Experiment 1, one group of DA and HL observers (group same) observed a demonstrator from the same strain, and the other group (group different) observed a demonstrator from a different strain. The observers received repeated cycles of observation training and testing in which they were placed in the demonstration chamber and received presentations of the two discriminative stimuli in the absence of the demonstrator and with free access to the chain. If the observers are capable of learning an instrumental discrimination by observing a proficient demonstrator of a same


1915

SAGGERSON AND HONEY

or a different strain, then presenting the Sþ to the observer should provoke more chain pulling (or more chain contacts) than should presenting the S –. If the effect observed in Experiment 1 is reliable then rats in group different should be more likely to show observational learning than those in group same.

Method Subjects and apparatus Charles River Ltd (UK) supplied the 38 rats (19 DA and 19 HL) used in this experiment. Of these, 6 rats were trained as demonstrators (3 DA, 3 HL), and 32 served as observers. All rats were maintained throughout the experiment at 80% of their ad-lib weights (M ¼ 249 g, range ¼ 179 –362 g) by restricting the amount of food they were given. Observers were housed in the vivarium for 12 weeks before testing began. The apparatus was the same as that used in Experiment 1 with the exception that there was a single chain that was positioned 2.5 cm from the front of the chamber’s door, 12.3 cm from the side walls of the chamber. Procedure Demonstrator training. Once the demonstrators had been trained to collect food pellets in the same way as in Experiment 1, the single chain was introduced into the chamber, and during the next six 20-min sessions, a food pellet was delivered on a VT 60-s schedule and each time the chain was pulled (i.e., on a continuous reinforcement schedule). After these 6 sessions, the discriminative stimuli (tone and clicker) were introduced, and training continued for a further 44 sessions, by which point the demonstrators were reliably pulling the chain during the correct discriminative stimulus. In each of these sessions, demonstrators received ten 60-s presentations of both discriminative stimuli that were presented according to a pseudorandom sequence with the constraint that there were no more than two presentations of a given stimulus in succession. The ITI was variable around a mean of 40 s (range: 20 – 60 s). As in Experiment 1, no

1916

reinforcers were presented during the first 20 s of each trial, but reinforcement was provided on a VI schedule during the remaining 40 s of each trial. The VI schedule was gradually increased from 10 s, at the outset of training, to 30 s, by the end of training. For one of the HL and one of the DA demonstrators the tone served as the Sþ, and the clicker served as the S – , and for the remaining demonstrators this arrangement was reversed. Observer training and testing. Each observer was given two 20-min sessions of exposure to the demonstration chamber in the two days before observational training. As in Experiment 1, although the chain was present in the chamber, no reinforcement was given for any chain pulls, and the discriminative stimuli were not presented. During observation sessions, the demonstrator was placed in the demonstration chamber; the observer was then placed in the observation chamber, and the demonstrator received a session of discrimination training that was identical to those that it had received at the end of training. For 6 observers in groups same (3 DA and 3 HL) and different (3 DA and 3 HL), the demonstrator’s Sþ was the tone, and S – was the clicker, and for the remaining observers the demonstrator received the reverse stimulus assignments. There was one such observation session on each of 8 days, and observers received a test immediately after the 2nd, 4th, 6th, and 8th observation sessions. During each test, the demonstrator was removed from its chamber, and after a delay of 2 min the observer was placed in this chamber. The observers received four presentations of the tone and the clicker in each test, which were presented in a counterbalanced order (either CTTCTCCT or TCCTCTTC) that was exchanged between successive tests. No reinforcement was presented during the test sessions. As in Experiment 1, the observer rats’ behaviour during these test sessions was videorecorded, and their chain contacts were monitored. There was a high degree of concordance between two independent scorers. For the sample of footage that was scored by both AS and RG



(eight 1-min epochs for 10 different observers during the test), the mean rate of chain contacts scored by AS was 2.66 rpm (Sþ ¼ 2.35 rpm, S – ¼ 2.98 rpm), and for RG it was 3.15 rpm (Sþ ¼ 2.43 rpm, S – ¼ 3.88 rpm). Pearson’s product – moment correlations revealed a significant correlation between the scores generated by AS and RG for rates of chain contacts both during the Sþ and during the S2 (r s ¼ .91 and .88, respectively; ps , .01). These raw scores were converted into discrimination ratios of the following form: rate of chain contacts during the Sþ divided by the combined rate of responding during the Sþ and S2. The mean discrimination ratio for chain contacts scored by AS was .42 and by RG was .37, and Pearson’s product – moment correlations for these ratio scores also revealed a significant correlation (r ¼ .78, p , .01).

Results and discussion Demonstrator performance during training and observation sessions The mean discrimination ratio during the final two sessions of demonstrator training was .73, and the mean rate of chain pulling during the S2 was 0.79 rpm. The mean discrimination ratio during the 20-s reinforcer-free periods during the same two sessions was .74, and the mean rate of chain pulling during the S2 was 2.47 rpm. The same demonstrators could serve as models for observers in groups same and different, and their behaviour during observation sessions is shown in Table 2 pooled across the eight observation sessions. Inspection of this table indicates Table 2. Experiment 2: Mean discrimination ratios for demonstrators during the observation sessions Group Same Different

Centre chain (20 s/60 s) .94/.92 .95/.91

Note: The ratios are presented separately for the first 20 s of the 60-s trials (in which no food was presented) and the entire 60 s.

that the demonstrators were more proficient in these eight sessions than they had been during the training sessions; but that their behaviour did not vary as a function of the nature of the observer (same or different). The mean rate of chain pulling during the entire duration of S2 presentations was 1.43 rpm when demonstrators were observed by group same and 1.18 rpm when they were observed by group different. There is no obvious explanation for the finding that the demonstrators’ performance was superior during the observation phase than it had been during training. For example, it seems unlikely that this finding reflects the fact that they have received additional training or an effect of the additional period for which they had been deprived. It does seem possible, however, that an observer’s presence might increase the reluctance of a demonstrator that has just received a food pellet (for pulling the chain during the Sþ) to return to the chain (that is close to the observer) unless the Sþ is still present. This effect of the presence of a conspecific would increase discriminative performance in the demonstrators. Observer behaviour during test sessions The mean chain contact discrimination ratios for groups same and different are shown in Figure 2. Inspection of this figure reveals that, pooled across the four tests during which the observers’ performance did not vary in any systematic fashion, the chain contact discrimination ratios for rats in group different were higher than those for rats in group same, F(1, 31) ¼ 9.86, p , .01. Further, one-sample t tests revealed that the scores for group different were significantly above .50, t(15)¼2.52, p , .05, but the scores for group same were not significantly below .50, t(15) ¼ 21.88, p ¼ .08. The mean rates of contacts during the S – were 3.34 rpm and 2.68 rpm for group same and group different, respectively (F , 1). The results of Experiment 2 replicate the central finding from Experiment 1: The effect of observational experience is influenced by the nature of the demonstrator. As in Experiment 1, observers who viewed demonstrators from a different strain were more likely to exhibit


1917

SAGGERSON AND HONEY

Figure 2. Experiment 2. Mean chain contact discrimination ratios (+SEMs) in groups same and different.

demonstrator-consistent behaviour than those who observed demonstrators from the same strain. In Experiment 2, however, the effect of demonstrator type was also accompanied by a significant observational learning effect in group different. We now consider the possible bases for the pattern of results observed in Experiments 1 and 2.

GENERAL DISCUSSION The conditions under which observation of a demonstrator performing a particular response influences the subsequent behaviour of an observer has received relatively little attention. In this study, we examined whether or not demonstrator type influences the consequences of observational experience in rats. The results of Experiments 1 and 2 indicate that whether the demonstrator is the same or different strain to the observer (and the observer’s cage mate) has an impact on the consequences of observational experience. Demonstrator-consistent behaviour was more

likely to be observed when the observers and the demonstrators came from different strains than when they came from the same strain. There is a good theoretical reason to suppose that the nature of the demonstrator might play an important role in observational learning that we now consider. One simple way in which manipulating the strain of the demonstrator, relative to that of the observer, could exert an influence on observational learning is through a process of attentional modulation. If one assumes that as a consequence of protracted social housing, observers come to ignore the sensory input provided by their cage mates, then they also might be less inclined to attend to (and learn about) a demonstrator of the same strain during observational training. That is, one could imagine that an effect akin to latent inhibition (Lubow & Moore, 1959) operates in the social domain.1 If one accepts this form of analysis for the failure to find observational learning in group same, then the obvious question that one needs to address is: What did observers in group different learn as a consequence of their observational experience? There are a number of plausible answers to this question. For example, observers might acquire an association between a representation of the discriminative stimulus and that of either the location in which the demonstrator was active or the stimulus/manipulandum that the demonstrator is interacting with (i.e., the chain). When the observer is later placed in the demonstration chamber the presentation of the discriminative stimulus might then elicit orientation and approach to that location or stimulus (see Honey, Good, & Manser, 1998a; Honey, Watt, & Good, 1998b). This analysis of what the observers learnt represents an elaboration of the notions of local enhancement and stimulus enhancement originally described by Thorpe (1963) and Spence (1937), respectively. Stimulus and local enhancement refer to the ideas that an

1

The fact that the behaviour of the different types of observer did not appear to exert a reciprocal influence on the demonstrators is perhaps a little surprising given this perspective. However, it seems plausible to suppose that the demonstators’ behaviour might be more resistant to such an influence because it was under the control of the discriminative stimuli and being maintained by reinforcement.

1918



observer might be more likely to interact with a stimulus or location when they have witnessed a demonstrator doing so. The elaboration that we are proposing requires that (the products of) local and stimulus enhancement can become linked to the discriminative stimuli that were present when these processes were operating. It seems reasonable to assume that, like asocial associative learning, such socially mediated forms of associative learning will be influenced by the familiarity of the to-be-associated cues: If the observer is not attending fully to the demonstrator then any association involving the discriminative stimulus and the sequelae of the demonstrator’s interaction with its environment will be acquired less readily. To summarize, although the specific form of analysis just summarized is speculative, the findings from Experiments 1 and 2 nevertheless highlight a more general theoretical point—namely, theories of observational learning need to incorporate a role for an observer’s prior (social) experience when attempting to explain the consequences of observational experience. This role was ascribed central importance in earlier discussions of the origins of imitation in humans (e.g., Miller & Dollard, 1941) and has equivalent significance within some more recent (associative) analyses of observational learning in animals (e.g., Heyes & Ray, 2000; LaLand & Bateson, 2001; Saggerson, George, & Honey, 2005). Original manuscript received 7 June 2005 Accepted revision received 7 February 2006 First published online 4 July 2006

REFERENCES Campbell, F. M., & Heyes, C. M. (2002). Rats smell: Odour-mediated local enhancement, in a vertical movement two-action test. Animal Behaviour, 63, 1055– 1063. Grindley, G. C. (1932). The formation of a simple habit in guinea pigs. British Journal of Psychology, 23, 127– 147. Heyes, C. M. (2001). Causes and consequences of imitation. Trends in Cognitive Sciences, 5, 253– 261.

Heyes, C. M., & Dawson, G. R. (1990). A demonstration of observational learning using a bidirectional control. Quarterly Journal of Experimental Psychology, 42B, 59 – 71. Heyes, C. M., Dawson, G. R., & Nokes, T. (1992). Imitation in rats: Initial responding and transfer evidence. Quarterly Journal of Experimental Psychology, 45B, 81 – 92. Heyes, C. M., Jaldow, E., & Dawson, G. R. (1994). Imitation in rats: Conditions of occurrence in a bidirectional control procedure. Learning and Motivation, 25, 276– 287. Heyes, C. M., & Ray, E. D. (2000). What is the significance of imitation in animals? Advances in the Study of Behaviour, 29, 215– 245. Honey, R. C., Good, M., & Manser, K. L. (1998a). Negative priming in associative learning: Evidence from a serial-habituation procedure. Journal of Experimental Psychology: Animal Behavior Processes, 24, 229– 237. Honey, R. C., Watt, A., & Good, M. (1998b). Hippocampal lesions disrupt an associative mismatch process. Journal of Neuroscience, 18, 2226– 2230. Laland, K. N., & Bateson, P. (2001). The mechanisms of imitation. Cybernetics and Systems: An International Journal, 32, 195– 224. Lubow, R. E., & Moore, A. U. (1959). Latent inhibition: The effect of nonreinforced exposure to the conditioned stimulus. Journal of Comparative and Physiological Psychology, 52, 415– 419. Miller, N. E., & Dollard, J. (1941). Social learning and imitation. New Haven, CT: Yale University Press. Mitchell, C. J., Heyes, C. M., Gardner, M. R., & Dawson, G. R. (1999). Limitations of a bidirectional control procedure for the investigation of imitation in rats: Odour cues on the manipulandum. Quarterly Journal of Experimental Psychology, 52B, 193–202. Ray, E. D., Gardner, M. R., & Heyes, C. M. (2000). Seeing how it’s done: Matching conditions for observer rats (Rattus norvegicus) in the bidirectional control. Animal Cognition, 3, 147– 157. Ray, E. D., & Heyes, C. M. (2002). Do rats in a two-action test encode movement egocentrically or allocentrically? Animal Cognition, 5, 245– 252. Reed, P., Skiera, F., Adams, L., & Heyes, C. M. (1996). Effects of isolation rearing and mirror exposure on social and asocial discrimination performance. Learning and Motivation, 27, 113– 129. Romanes, G. J. (1884). Mental evolution in animals. New York: AMS Press.


1919

SAGGERSON AND HONEY

Saggerson, A. L., George, D. N., & Honey, R. C. (2005). Imitative learning of stimulus – response and response – outcome associations in pigeons. Journal of Experimental Psychology: Animal Behavior Processes, 31, 289–300. Spence, K. W. (1937). Experimental studies of learning and higher mental processes in infra-human primates. Psychological Bulletin, 34, 806– 850.

1920

Thorndike, E. L. (1911). Animal intelligence: Experimental studies. New York: Macmillan. Thorpe, W. H. (1956). The evidence for insight learning and insight. In W. H. Thorpe (Ed.), Learning and instinct in animals (pp. 121– 137). Cambridge, MA: Harvard University Press. Thorpe, W. H. (1963). Learning and instinct in animals (2nd ed.). Cambridge, MA: Harvard University Press.
