Learning and Memory
John R. Anderson, 2000

First Impressions
This book is a catalog of classic and recent results in cognitive psychology, particularly in the domains of learning and memory in humans and other animals, interspersed with arguments for Anderson's cognitive model, ACT. In the original version of ACT, started in the late 1970s or early 1980s, the acronym meant "Adaptive Control of Thought". The three letters have been used in various ways since, most recently in the title of Anderson's 1998
Atomic Components of Thought .

One of the strong points in Anderson's approach to psychological modelling is his insistence on neural plausibility. By that, I mean that his models must not only work in several specific senses, but that they must also be plausibly implementable in a reasonably natural neural network configuration. Most of the ongoing work with these models at Carnegie Mellon is done with a "digital" implementation, this is to say, a traditional computational model, implemented in LISP. However, Anderson and his group have also produced a neural network version of the system and have shown it to be operationally equivalent to the LISP version, although it runs much, much slower on a traditional computer. As is made clear throughout the discussions in this book, any new ideas proposed for implementation in the ACT model must meet the neural plausibilty test, as well as satisfactorily duplicating the behavior of subjects in the psychological experiments being modelled.


Top of Page | LAM Summary | Sort by Topic | Sort by Title | Sort by Author