demonstrators tended to over-respond, making more responses than the number o f food rewards available. Cam made 110% more responses than the 30 required to retrieve all the food items for demonstration sessions to animals from Group RIGHT, while Pep over responded by 51% in sessions to animals from Group LEFT. Cam's greater over-responding is reflected in his short session times (the shortest possible is 540 sec if all food items are retrieved within 20 sec of their delivery), and generally shorter times than the more cautious
Pep. However, both animals always finished their sessions in around 10 to 15 minutes. In contrast, there was marked variability in the demonstration-test interval, resulting from differences in the time taken to remove the demonstrator and move the observer into the test cage. The shortest delay was only 3 minutes, while the longest delay was 20 minutes. Importantly, both demonstrators reliably opened the food cover in the trained direction. In only one session. Pep's second demonstration to Paq (Group LEFT), were any untrained responses made. In this session, only 2 responses were made in the incorrect direction, and neither resulted in food recovery.
Observers
It was predicted that observers from Group LEFT would make proportionately more left responses in order to retrieve food items than observers from Group RIGHT. Unfortunately this prediction could not be tested because only two subjects {Car and Pap), both from Group RIGHT, retrieved any food items from the apparatus. Both animals fulfilled the session criterion of recovering 20 food items, and a discrimination ratio was calculated for each animal by dividing the number of left responses by the total number of responses made in the test session. These discrimination ratios revealed that both subjects had a strong bias
to respond to the left, the opposite direction to that which was demonstrated {Car, .80; Pap,
1.00).
It is inq)ossible to determine, on the basis of these data, whether Car and Pap responded in the opposite direction to that which was demonstrated because they were influenced by what they had observed, or whether such an effect (which might have been described as egocentric observational autoshaping, see Chapter 2, Section 3.3.2) would be typical of other monkeys, had they responded. Experiment 10 was not a fair bidirectional control test of this sample of capuchin monkeys' capacity to imitate because the reliability and the representativeness of this mismatch cannot be evaluated on the basis of the data from these two subjects alone. Furthermore, without animals from Group LEFT having tended to move the manipulandum to the right, an overall preference for left responses cannot be ruled out as an explanation for the directional mismatch of the two successful animals.
It might be argued that the disappointingly low number of animals that retrieved food items suggests that these animals were not capable of imitation, particularly if it could be shown that they were motivated to recover the food items and had observed the demonstrations. There is some support for this position. First, there are reasons to suggest that the observers were likely to have been motivated to retrieve the food items used in this experiment. The mixture o f peanut pieces and candy that was mainly used through testing was sufficient to maintain a high level of responding from the demonstrators. Furthermore, the observers were normally very highly motivated to eat bananas - the food type used on the first trial of the second test session.
Second, there is some evidence to suggest that these subjects had tended to pay attention to their demonstration sessions. It was possible to estimate the extent to which the observers had an opportunity to observe demonstrated responses from the videorecordings o f the demonstration sessions by using a measure of the relative frequency o f these responses in which the observer was close to the apparatus. Proximity rates were the percentage of demonstrator responses recorded onto video tape in which each observer had been within camera shot (an area of approximately 1 m^ around the apparatus) one second prior to each
Figure 7.2: Mean proximity rates for each group over both tests from Experiment 10. Group LEFT observed a left pushing demonstrator, and Group RIGHT observed a right pushing demonstrator. Proximity Rates were the percentage o f recorded demonstrator responses in which the observer had been in the proximity of the apparatus. The dashed line at 20% indicates an estimation of the proximity rate expected by chance (see text for further details). 100' 80' ( T ▼ LEFT 10
.
RIGHT T e s t 1 T e s t 2response*^. Mean proximity rates for each group over both tests are presented in Figure 7.2. Figure 7.2 also indicates, for illustrative purposes, an estimate of the proximity rates that observers might be expected to have had by chance. Assuming that the observers would spend the demonstration session on the floor of the observation cage, this was estimated to be the ratio o f the area in camera shot (1 m^) to the area of the floor of the cage (5 m^). This is likely to a conservative estimate of chance proximity because these monkeys generally spend more than 20% of their time away from the floor (Vitale et al., 1991). The relatively high proximity rates for this sample, particularly for the first test, would appear to indicate that food recovery demonstrations were salient events for these observers.
The data presented in Figure 7.2 also suggests that Group RIGHT tended to have a higher proximity rate than Group LEFT, and that proximity was less on the second test for both groups. These impressions were examined by conducting an ANOVA on data transformed by raising the proximity rates to the power of 2.7 to stabilise heterogeneity of variance. Group RIGHT tended to spend a higher proportion o f demonstrated responses in the vicinity of the apparatus than Group LEFT, F (l,6 ) = 10.64, p = .017, and animals' proximity rates were lower on the second test relative to the first, F (l,6 ) = 6.72, p = 0.041. The interaction between these factors was not statistically significant (F<1). The overall decrease in the proportion o f demonstrator responses for which the observer was in the proximity of the apparatus from the first test to the second would appear to suggest that repeated testing may result in a decrease in the salience of the demonstrators' behaviour (or operations of the apparatus). This result suggests that further testing of these monkeys using the current procedure may be unlikely to yield an observational effect such as imitation.
While the case for an inability to imitate is strengthened by the observers' apparent motivation to recover food items, and interest in demonstrations, there may be other reasons which may have prevented 8 o f the 10 subjects from retrieving food from the apparatus. All the
^^Not every demonstration session was captured in its entirety. However, a sufficiently large proportion of the demonstrated responses were filmed to render this measure reliable as an indicator of what happened throughout each session (98%, first test; 99%, second test). Proximity was scored for the second before each response to avoid potential over-estimation of opportunities to observe responding. Casual observation of the videos had revealed that monkeys sometimes approached the apparatus very rapidly after forceful demonstrator responses which resulted in an audible bang.
monkeys may have been unable to exploit an ability to imitate during testing because they had only a transient memory for the demonstrators actions; this memory may have been only weakly encoded, and thus decayed during a lengthy demonstration-test interval. Another factor which may have prevented monkeys from using any capacity that they might have to imitate was that the apparatus had a different appearance from viewpoints in the observation and test cages. This difference may have been sufficient to prevent the animals from recognising that the object that they were exposed to on test was the same object from which they had previously seen a demonstrator remove food. Finally, because the apparatus was a novel object for these animals, they may have been sufficiently afraid o f it during the test phase to subdue any motivation to attempt to retrieve food items. A pre-observation phase in which the non-manipulable elements of the apparatus could have been pre-exposed to the observers may have attenuated neophobia directed to this object.
E
x p e r i m e n t1 1 : G
r o u p t e s t i n gIn Experiment 11 observers were retested using a procedure which differs from that in Experiment 10. This modified procedure was intended to overcome some o f the potential factors that may have prevented subjects from retrieving food in Experiment 10, and may have made that experiment an insensitive bidirectional control test. Only the animals that had failed to retrieve food were given further demonstrations of food recovery from the same demonstrator. However, this time all the non-retrievers from a domestic group were in the test cage with their demonstrator during these sessions, and remained there for a social test session until one of the animals had successfully retrieved a number of food items. Social testing might be a more sensitive procedure than individual testing for any o f the following reasons: First, by not requiring observers to be moved to a different cage at the end of a demonstration session, demonstration-test intervals might be shortened, ininiinising the memory requirement for the task. Second, by allowing observers to witness their demonstrator manipulating the food-recovery apparatus from the same side as they would confront it at test (i.e., from the test cage), observers may be more likely to recognise that they had an opportunity to manipulate the same object as had their demonstrator. Third, by accompanying observers with other members of their social group, subjects may be generally
more manipulative while in the test cage (social facilitation), perhaps through the reduction of isolation induced anxiety (Clayton, 1978).
Method
The method was identical to that for Experiment 10, except in the following respects.
Subjects
The subjects were from the same two domestic groups of capuchin monkeys used in Experiment 10. Cam (Group RIGHT) and Pep (Group LEFT) again served as demonstrators. The remaining animals, with the exception of the two animals that retrieved food in Experiment 10, again served as observers. Group sizes were now 3 (Group RIGHT) and 5 (Group LEFT).
Procedure
Before the demonstrator was introduced into the test cage, all the unsuccessful observers were moved into that cage, and the apparatus was baited. A test session began when the demonstrator was removed from the test cage, immediately after he had retrieved the 30th food item of the demonstration session. Test sessions lasted for 10 minutes, during which all the observers had free access to the apparatus without being impeded by the presence of the demonstrator. If, at the end of a test session, an observer had managed to retrieve 20 food items, that animal was excluded from the observing group for any subsequent social testing. This step was taken to prevent that animal from monopolising access to the apparatus during subsequent tests. The procedure for social testing was repeated until only a single observer was present. If the lone observer did not recover any food items in its test session, the procedure was repeated one final time with members of its social group present in the observation cage in an attempt to alleviate anxiety caused by testing in isolation.