A Distributed Information Processing Model of Visual Memory

Despite a long and vigorous history of research on human memory, very little is known about how it actually works, and it is often treated as a mystery. This paper consists mainly of arranging well-known results from psychology, biology, and computer science in a way that appears to hang together. There are still many gaps, a few of which get bridged with constructs of varying quality. Distributed information processing models are presented as a formalism for relating mental functions (here, visual memory) to neural models. Our model of visual memory is connectionist and reconstructionist, and relies heavily upon mutual inhibition, relaxation, recruitment from random connections, and a kind of self-simulation