utilitarian factors in a probabilistic decision-making task
Equation 3.1 would give the complete distribution over the probabilities for some
3.4 and 3.5, parameter c: subsequently dropped from the analysis due to non-
3.4.4 Performance referenced to the ecological observer model: Developmental differences in the role of performance feedback
Participant decisions were referenced to those of an ideal observer, and this showed that both adults and adolescents selected the correct stimulus at a level significantly above chance. However, adults selected the correct stimulus on a greater percentage of trials than did adolescents, which suggests that adults were more sensitive to utilitarian performance feedback (wins and losses) than were adolescents.
It has previously been suggested that the ability to take into account performance feedback continues to develop during adolescence (Huizinga et al., 2007; Byrnes et al., 1999; Crone & van der Molen, 2004) and that, in particular, the time-course of developing sensitivity to performance feedback may differ depending on whether the feedback is negative or positive (Crone et al., 2004, 2008; Huizinga et al., 2006; van Duijenvoorde et al., 2008). The current study investigated whether the age group difference in performance was due to a differential response to positive (win) and negative (loss) performance feedback, by creating a computational model of participant behaviour (the ecological observer model) containing separate parameters encapsulating the behavioural influence of positive (‘you win 10 p’) and negative (‘you lose’) performance feedback on behaviour.
Results of this analysis showed that the adolescent and adult groups differed in their behavioural sensitivity to performance feedback. Specifically, the adult group showed a stronger tendency than did the adolescent group to select a face once it had been associated with a reward. In both age groups, receiving negative feedback
(‘you lose’) in response to choosing a face decreased the probability that the face would subsequently be chosen. Thus, adult decisions reflected a composite of both negative and positive performance feedback whereas adolescents showed a bias to respond to negative feedback only.
In the ecological observer model, the effect of performance feedback depended on the belief level of participants when the feedback was given (see Methods). If the evidence based on previous trials indicated that participants (as described by the ecological model) were nearly certain that a particular face was correct, then receiving positive feedback in response to the alternative face had very little effect on belief and subsequent behaviour. The effect of performance feedback also depended on current position within each 26-trial block. When participants had received very little performance feedback, at the beginning of a block, each new instance of feedback had a large effect. Later on in the block, each new instance of feedback had proportionally less effect (see Figures 2 and 5 in Averbeck & Duchaine, 2009). Therefore, the developmental result showing greater sensitivity to positive performance feedback in the adult than in the adolescent group may reflect a more mature ability to keep track of and update the history of rewards associated with each stimulus.
3.4.4.1 Adolescent response to positive vs. negative feedback and behavioural strategies
Results from the current study indicate that adolescents aged 12-13 exhibit an unbalanced response to positive vs. negative performance feedback. This finding is in general agreement with studies using rule-switch tasks, which have shown that children and adolescents exhibit greater behavioural asymmetry in their response to positive and negative performance feedback than do adults (Crone et al., 2004, 2008; Huizinga et al., 2006; van Duijenvoorde et al., 2008). However, data from the current study indicate that adults are more sensitive to positive feedback than are adolescents and that both age groups are equally sensitive to negative feedback, whereas studies using rule-switch tasks tend to report the opposite finding that children and adolescents respond better to positive than to negative feedback. For example, while performance in rule-switch tasks is consistently worse following negative (‘switch’) than positive (‘maintain’) feedback across age, performance is disproportionately worse following negative feedback in children aged 8-9 (Crone et
al., 2008; van Duijenvoorde et al., 2008). However, the tasks used are quite different. In the next paragraph, a possible explanation will be suggested for the divergent findings.
It has been suggested that a developmental shift in strategy is responsible for the age-related changes in behaviour during decision-making tasks. In a study by Huizinga et al. (2007), it was shown that, whereas pre-adolescents (aged 6-9 years) were behaviourally sensitive to the frequency of loss only, adolescents aged 10-15 took into account both the frequency and magnitude of losses. Between mid- adolescence (13-15 years) and young adulthood (18-25 years), the tendency to take into account both the magnitude and frequency of losses continued to increase, and in addition there was evidence for an increase between adolescence and adulthood in the adoption of the maximally unbiased strategy of indifferent choosing between two options that are equally beneficial. Therefore, developmentally immature decision- making may be characterized by a particular set of biases, and focus on a subset of stimulus attributes. The pattern of adolescent behaviour in the Face Decision Task, whereby adolescents take into account negative but not positive feedback, is consistent with a simple behavioural strategy of switching choices after incurring a loss. In contrast, adult participants are able to integrate both wins and losses, showing an increased tendency to switch choices following a loss, but also a tendency to maintain the current choice following a win. The divergent findings between studies may be due to a difference in the strategy employed to accomplish distinct behavioural tasks (context-dependence of strategies). A future study could investigate the factors that cause adolescents to adopt a particular strategy in favour of another, as this could be relevant to adolescent decision-making in naturalistic contexts.
3.5 Conclusion
The current study was carried out to investigate decision-making in adults (aged 21- 43) and adolescents (aged 12-13) using a probabilistic Face Decision Task in which happy and angry face stimuli were stochastically associated with reward. A computational method of analysis was implemented, which showed that the decisions of both adolescents and adults can be characterized as a composite of both
social and utilitarian factors. Across age, participants showed a similar social bias to choose the happy face; however, the response to utilitarian performance feedback differed across age. Whereas adults integrated both positive and negative performance feedback (wins and losses) to either adjust or maintain ongoing selections, adolescents failed to respond to positive feedback, switching choices after incurring a loss. This suggests that age-related change in the ability to integrate positive and negative stochastic feedback may contribute to the development of decision-making abilities between adolescence and adulthood.
3.5.1 The next chapter
Adolescence is characterized by changes in decision-making, including in the ability to take into account performance feedback. In this chapter, effects of facial emotion cues on the decisions of adults and adolescents were shown. In general, adolescence is said to be characterized by changes in emotion and social understanding, including increased emotional self-awareness and self-consciousness. There is evidence that that pubertal development rather than chronological age per se contributes to increased social awareness and self-conscious affect (Simmons et al., 1973; Rosenberg & Simmons, 1975). Furthermore, it has been shown that puberty hormones organise the structure and function of the brain, including parts of the brain involved in emotion and social behaviour (Romeo, 2003; Cahill, 2006; Ahmed et al., 2008; Schulz et al., 2009; Peper et al., 2009; Neufang et al., 2009). Therefore, in the next chapter, a study is described that investigated pubertally-associated development in self-reported emotions. Participants aged 9-16 years were divided into groups on the basis of a puberty development questionnaire, and performed an emotional self-report task. In the task, participants were asked to imagine their emotional responses to a variety of social and basic emotional scenarios. Patterns of emotional responding were investigated, to evaluate the hypothesis that the onset of puberty marks an increase in the awareness of complex, or ‘mixed,’ emotions.