Impaired context-learning (value signal normalization) in the aging brain

点击翻译按钮获取中文摘要

ABSTRACT Aging is associated with increased susceptibility to fraud and impairments in financial decision making. To be an effective financial decision maker, in addition to learning how to gain reward, one must also learn how to avoid losses. Powerful brain systems reinforce learning when the outcome is rewarding. However, it is not clear whether reinforcement-learning brain systems contribute to the problem of learning to avoid losses, as when learning to avoid losses, a “no-change” outcome may be the best option available but is not in itself inherently rewarding. Our recent findings indicate that, over time, younger adults learn to adjust based on context such that in loss contexts, having a no loss outcome activates reward centers in the brain and trigger reinforcement-learning processes that allow them to learn as effectively about avoiding losses as gaining rewards. In contrast, our preliminary behavioral findings indicate that older adults show selective impairments in learning to avoid losses. This age-related impairment could either be due to impairments in the ability to learn about the decision context or to a more general ‘positivity effect’ in which they attend relatively less to negative than positive outcomes. In the current proposal, we develop and test computational models to identify the contribution of different types of learning mechanisms to these age differences. In particular, we will compare the degree to which impairments in learning about the context versus a bias to attend more to positive than negative feedback can account for older adults’ learning patterns. We then will examine the neural correlates of the best fitting models for the young and older participants. In addition, in a representative sample of Americans, we will examine how individual differences in context learning and positivity bias relate to their financial decision-making abilities and susceptibility to scams.