The role of theories of cognitive aging: comment on Salthouse

Cognitive models of aging ought, minimally, to simulate age effects for a broad range of tasks, and their underlying assumptions ought to be constrained by existing data on cognitive mechanisms. Salthouse's model (and other parallel distributed-processing network models) meets the first criterion better than it meets the second. Another weakness is in its inability to account for goal-directed behavior. Thus, models like that of Salthouse seem best suited to the description of automatic information processes such as recognition. I conclude that such models are best evaluated in competition with standard symbolic processing models. I introduce simulation of chess position recall by older and younger chess players and compare it with Salthouse's model.