Wednesday March 18, 11:30-12:30

SAC: A Unified Activation Based Account of Implicit and Explicit Memory Phenomena

Lynne M. Reder

(Department of Psychology, Carnegie Mellon University, United States)

In this talk, I will describe an activation based model of memory called SAC (Source of Activation Confusion) that accommodates a wide variety of memory results. This localist network has semantic and perceptual features as basic elements, but builds up higher level units,corresponding to patterns of past experience. The model has been used to account for a variety of findings in the domains of implicit and explicit memory, both from the literature and from new results that were obtained as part of tests of novel predictions of the theory. The experiments include (1) continuous, rapid, feeling of knowing experiment, in which arithmetic problems are presented multiple times and subjects try to decide rapidly with each presentation whether they already know the answer and can quickly retrieve it, or must calculate the answer; (2) continuous remember/know judgments varying pre-experimental and experimental word frequency, where the pattern of remember vs. know judgments varies in systematic ways depending upon both pre-experimental word frequency and experimental word frequency(i.e., numbers of repetitions in the experiment); (3) implicit memory tasks using fragment completion and word identification, where we vary the same factors used in the continuous recognition experiment, but also vary the delay between encoding and the implicit memory test. These experimental data are fit to simulated data derived from a computational model. The fits are impressive, especially considering that the level of precision is high (the grainsize of the analyses is fine) and considering that most of the model parameters were estimated from previous model fits, leaving few degrees of freedom for the many data points.