Matters temporal.

Peter Dayan
Trends in Cognitive Sciences, 5 , 105-106.


Abstract

Current evidence suggests that neural Hebbian learning in cortical and hippocampal syna\-pses is fundamentally predictive rather than conventionally correlational. Exactly what sort of predictions are acquired, and in what neural architectures, is the focus of substantial work. A recent paper by Rao & Sejnowski (2001) has suggested an interesting interpretation in terms of a popular predictive algorithm which has roots in psychology, computer science and engineering.
pdf

back to:   top     publications