Until recently, cognitive psychologists had interpreted memory changes in old age in terms of “process” models for information flow. A meticulously detailed exemplar is the “working memory” model first proposed by Baddeley and Hitch (1974) and since extensively developed (Baddeley, Grant, Wight, & Thompson, 1975; Baddeley & Lieberman, 1980). Information from the sense organs is “encoded” into characteristic “representations” in one or more modality-based subsystems (the visual “scratch pad” and the auditory “articulatory loop”). Besides its own characteristic representation code, each of these subsystems has characteristic temporal holding characteristics, which may depend on rate limitations to a dynamic process (the refreshment cycle time of the articulatory loop) or on capacity limitations of unknown provenance (the capacity of the scratch pad). Each subsystem also has its characteristic place in an information-routing diagram for the total “memory system.”

In this framework, the study of cognitive aging has become an investigation of the differential vulnerability of subsystems and of their linkages. Thus, among other excellent recent reviews, Erber (1982) discusses changes in the efficiency of “sensory memory,” “primary memory,” “secondary memory,” and dynamic read-in and read-out processes (encoding and retrieval). Kausler (1982) uses a similar “bottom-up” hierarchical description of changes in hypothetical “primary memory” and “episodic memory” systems. Craik (1976) suggests a framework for interpreting age changes in memory in terms of a hierarchical scheme of progressively “deeper” processing stages.