Subject Age/Sex Success Subject Age/Sex Success Subject Age/Sex Success
KL 3M 62.5 BB 3M 100* DC 3M 40 JW 3M - CG 3M 50 MH 3M 50 RR 3F - KE 3F 50 JJ 3M 60 EC 3F 75 RG 3F 50 AB 3F 70 JP 4M 30 JT 4M 80 MO 4M 44.44 CB 4M - CM 4M 30 RC 4M 90* JL 4M - BL 4M 80 L 4M 40 JT 4F 50 SA 4F 30 HP 4F 50 CG 4F - SD 4F 66.67 CA 4F 66.67 C 4F - ER 4F 10* ML 4F 70
Median 0 Median 50 Median 55
Table 12.2 Percentage o f successful trials. The symbol represents a significant result based on a binomial test (* p < 0.05). For results less than 50% correct, significance indicates a performance significantly below chance.
12.3.4 Patterns o f success
There was a positive correlation between increasing trial number and group performance. However, this relationship was not significant for any of the experimental groups, and hence the children did not show a progressive learning effect (Spearman’s rho\
controls = 0.39, p = 0.253; C-only = 0.44, p = 0.199; I+C == 0.451, p = 0.191; See Fig. 12.2).
100 -
I
Y = 5.15X + 28.33 80 - 60 - 8 40 - C È £ ■ Controls 0 1 2 3 4 5 6 7 8 9 10Figure 12.2 Percentage o f children from each group who chose the correct side of too insertion on each trial. The line represents the linear best-fit, and suggests that there is a positive relationship between trial number and group performance. However this relationship is not significant for any o f the groups.
Trial number 100 - 100 -
I
Y = 3.55X 4-38.59 a Y = 3.34X4- 39.18 2? S I c 60 - 40 - 60 - ♦ C-only 0 1 2 3 4 5 6 7 8 9 K 0 1 2 3 4 5 6 7 8 9 1012.3.5 Side biases
A greater number of subjects in the C-only gi'oup showed a significant tendency to insert the tool into the same side of the apparatus. However, when analysed as a group there was no significant difference in the side bias of children from the C-only and I+C groups
(median C-only = 50, median I+C = 52.22; Mann-Whitney U-test for unmatched samples, z
= -0.53, Ni = 10, N2 = 10, p = 0.593, two-tailed; see Table 12.3). Interestingly, each child
who showed evidence of a side bias had a preference for the left side of the tube, which was the con*ect side of tool insertion for the first demonstration and trial. In addition, the majority of the remaining subjects tended to insert the tool more frequently into the left side of the tube. Tliis suggests that children tended to reproduce the same behaviour that they observed during the first successful trial, rather than taking account of the causal properties of the task that determined the correct side of insertion on subsequent trials.
CONTROLS C-ONLY I + C
Subject Age/Sex % Left Subject Age/Sex % Left Subject Age/Sex % Left
KL 3M 37.5 BB 3M 50 DC 3M 40 JW 3M - CG 3M 50 MH 3M 100* 1 RR 3F - KE 3F 100* JJ 3M 90 EC 3F 50 RG 3F 100* AB 3F 20 JP 4M SO JT 4M 50 MO 4M 44.44 : CB 4M - CM 4M 100* RC 4M 60 JL 4M - BL 4M 50 L 4M 70 JT 4F 100* SA 4F 60 HP 4F 80 CG 4F - SD 4F 30 CA 4F 44.44 .1 C 4F - ER 4F 40 ML 4F 40 Median 50 Median 52.22 ;
Table 12.3 The percentage o f trials in which the tool was inserted into the left-hand side o f the tube. Percentage o f left insertions is used as a measure o f side bias.
12.4 Discussion of Experiment 6
The results suggest that like chimpanzees, 3- and 4-year-old children did not significantly benefit from the observation of errors in the context of this study. Subjects who obseiwed both success and failure were unable to utilise this additional infonnation to extract the causal relationships involved in task solution, and perfoim significantly better than children who observed only solutions. However, children in the experimental gi'oups who observed the tool being inserted into the apparatus seemed to learn the causal relevance of tool-reward contact. Children in the control group saw the tool being manipulated in the same direction and to the same extent as the other groups. The major difference was that in the control demonstiation the tool was moved along the top of the apparatus and therefore had no significant effect on the reward located inside the tube. Observing that the tool could potentially be used to effect the position of the reward helped children to learn about contact. However they did not learn that the coiTect side of tool insertion was related to the position of the reward relative to the trap.
Children from each group showed a tendency to insert the tool more frequently into the left-hand side of the tube. Indeed, a number of children showed a significant left side bias (see Table 12.3). The left side was the conect side of insertion for each group during the first demonstration and trial. It therefore seems that children were more influenced to reproduce the behaviour that they observed during the first successful trial, than take account of the causal properties of the task that deteimined the correct side on subsequent trials. In addition, a number of the children in the control group were seen to reproduce the irrelevant actions and in some cases vocalisations of the demonstrator.
The results of this study contrast with the findings of Want & Harris (2001), who found that 3- year-old children were able to benefit horn observing mistakes. However, there are a number of important ways in which the cunent study differs from theirs. Firstly, subjects received four demonstrations throughout the block of ten trials that were counter balanced for the correct side of tool insertion. The increased number of demonstrations meant that subjects had a greater opportunity to learn by social, rather than individual learning. Secondly, the position of the reward was presented in a randomised sequence tlnoughout the
block of ten trials. It was therefore not possible to solve the task by learning to reverse responses on alternate trials. Finally, the trap was located in the middle of the tube so that the apparatus did not have to be rotated between trials.
The methodology of the present experiment was designed to test children under the same level of control that would be expected for work with any non-human species. When the same strict controls were applied, the results were not radically different from the chimpanzee subjects in the previous chapter. Both chimpanzees and children were able to learn the causal relevance of contact. However, control subjects who could not learn this mle behaved differently in each case. Chimpanzee control subjects responded by ignoring the demonstrations, and trying to reach the reward with their fingers. However, children in the control group responded by reproducing the inelevant actions and vocalisations of the demonstrator. This finding is in accord with several studies that have demonstrated that young children have a tendency to reproduce observed actions at the expense of efficiency (Nagell et al., 1993; Whiten et al., 2003; Whiten et al., hi press; see Chapter 4).
However, it is possible that the differences in experimental design meant that although the task was conceptually difficult for children of tliis age, it may be solved by older children. The following chapter will discuss whether the ability to solve the trap-tube task is related to the age of the participants.
Sum m a r y of Ch a pter 13
Experiment 7 was designed to determine whether cliildren’s ability to benefit from the observation of eiTors is related to age. The previous experiment showed that like chimpanzees, 3- to 4-year-old children do not benefit from the obseiwation of errors in the context of the trap-tube task. This experiment was therefore conducted with 5- to 6-year-old children, to investigate whether older children would perform differently.
The results show that there was no difference in the perfomiance of children from any of the three experimental groups, and that all groups perfonned significantly better than would be expected by chance. It seems that by the age of 5- to 6-years-old, children understand the causal principles involved in the trap-tube task, and hence do not benefit from observing eiTors.