Process Explanations of Age-Related Changes in Memory Specificity

Adult aging is associated with a decline in memory for the specific details of past experiences, but the reasons for these declines remain to be fully elucidated. Here, we consider the potential of computational process models of memory to reveal mechanisms underlying age-related changes in memory specificity. We define empirical benchmarks in the most widely used tasks for studying age differences in memory -- recognition and free recall. We then systematically review two broad classes of process models -- global matching and retrieved context models -- that are well-suited to explaining these benchmarks. We discuss and at times illustrate through simulations how age differences in memory specificity can arise from different mechanisms in these models, such as reductions in encoding fidelity, cue-selectivity of retrieval, contextual reinstatement, or retrieval monitoring. These models hold great promise for uncovering the latent mechanisms of age-related memory change, but several explanatory shortcomings must be addressed for a more comprehensive theory to emerge. Because process models often incorporate simplifying shortcuts, they risk advancing descriptive rather than mechanistic explanations for age differences in memory. Furthermore, models that have identified multiple processes of age-related memory impairment have left unaddressed the ways in which these processes affect each other. By illuminating these and other challenges in current process models, our review aims to stimulate future advances that can dramatically reshape our understanding of the processes of age-related changes in memory specificity.