Human Working Memory

Abstract:  by computational efficiency concerns, and embodies the general principle of similarity-based interference (Baddeley & Logie; Cowan; Schneider; and O'Reilly, Braver & Cohen --- all in this volume). (4) Soar emphasises the role of learning in WM phenomena, even on tasks which experimentally are regarded as concerning just "performance." The moral is that WM cannot coherently be studied independently of long-term memory and while ignoring learning. (5) Soar stresses the recognitional usage of information in long-term memory acquired as a by-product of earlier task performance. It therefore has close links to other approaches which emphasise the involvement of long-term memory in WM, and may be able to offer a computational process model for "long-term WM" (Ericsson & Delaney, this volume). 1. Introduction In the Spring of 1987, Allen Newell delivered the William James lectures at Harvard University (Newell, 1990). He used the occasion to argue the case for striving towards theoretical un