Prediction in the Aging Brain: Merging Cognitive, Neurological, and Evolutionary Perspectives

Abstract

Abstract Although the aging brain is typically characterized by declines in a variety of cognitive functions, there has been growing attention to cognitive functions that may stabilize or improve with age. We integrate evidence from behavioral, computational, and neurological domains under the hypothesis that over the life span the brain becomes more effective at predicting (i.e., utilizing knowledge) compared to learning. Moving beyond mere description of the empirical literature—with the aim of arriving at a deeper understanding of cognitive aging—we provide potential explanations for a learning-to-prediction shift based on evolutionary models and principles of senescence and plasticity. The proposed explanations explore whether the occurrence of a learning-to-prediction shift can be explained by (changes in) the fitness effects of learning and prediction over the life span. Prediction may optimize (a) the allocation of limited resources across the life span, and/or (b) late-life knowledge transfer (social learning). Alternatively, late-life prediction may reflect a slower decline in prediction compared to learning. By discussing these hypotheses, we aim to provide a foundation for an integrative neurocognitive–evolutionary perspective on aging and to stimulate further theoretical and empirical work.