ABSTRACT
Schacter (1987) reviews three different approaches that explain dissociations between 'explicit' and 'implicit' memory task performance: the activation view, the processing view, and the multiple memory system view. These theoretical distinctions have maximum explanatory power only when implemented in an operational process-level model of memory performance. As suggested by Schacter, the three 'views' or 'loci of effect' may not be mutually exclusive. This presentation starts with the most parsimonious assumption, a single memory system, modeled using Murdock's (1982) distributed memory model TODAM. We assess the degree to which this single system can account for a selection of both implicit memory phenomena (i.e., priming effects) and explicit memory performance and suggest necessary extensions or modifications of the model. Most implicit memory phenomena involve redintegration of partial cues (e.g., word fragment completion), a task which is handled easily and naturally by distributed memory models. Throughout, we discuss how the assumptions about encoding, storage, and retrieval processes made by TODAM to account for the data can be interpreted in light of current theoretical explanations of explicit/implicit memory differences.
