Deviations from Bayes' Theorem During Belief Updating in Younger and Older Adults: Evidence From Behaviour and Neural Activity

Abstract

Updating prior information with new information in accordance with Bayesian principles is a difficult task. Younger adult decision makers deviate from Bayes’ theorem by either overweighting prior information (i.e., using a conservatism heuristic) or overweighting new information (i.e., using a representativeness heuristic) on decision tasks without feedback. Similar to younger adults, older adults make decisions that require belief updating. Given agerelated decrements in cognitive control, older adults may be at a disadvantage compared with younger adults when updating beliefs. Prior research shows no age differences when making decisions under risk, however older adults perform worse than younger adults when making decisions under ambiguity. Currently it is unknown how older adults use heuristics when updating beliefs about risk and ambiguous information compared with younger adults. The primary aim of this dissertation was to examine age-related differences in the use of heuristics during belief updating, as well as the cognitive processes and neural correlates that underpin behaviour. In three experiments, younger and older adults completed a belief updating task with and without feedback using an urn-ball paradigm. The main results showed that both younger and older adults committed the representativeness error more than the conservatism error, with no age differences observed when updating beliefs without feedback but with younger adults updating beliefs more accurately than older adults with feedback. Further, age differences in the neural correlates that underlie belief updating showed evidence that older adults recruit additional resources in frontal regions of the brain to facilitate performance compared with younger adults. Event-related potentials showed evidence of cognitive control in response to conflicting information in both age groups, but a diminished neural response to feedback in older compared with younger adults. Additionally, while younger adults were not influenced by ambiguous information, older adults avoided committing the representativeness error only when new information was ambiguous. Last, individual differences in numeracy and cognitive reflection, but not thinking disposition, modulated belief updating performance. Together, the results show that younger and older adults can learn to update beliefs with feedback but with younger adults learning to a greater degree than older adults, especially when information is ambiguous.