Classical and instrumental learning by neural networks

Author(s): Grossberg, S. |

Year: 1974

Citation: Progress in theoretical biology. New York: Academic Press, pp. 51-141.

Abstract: This article reviews results chosen from the theory of embedding fields. Embedding field theory discusses mechanisms of pattern discrimination and learning in a psychophysiological setting. It is derived from psychological postulates that correspond to familiar behavioral facts. The theory tries to isolate facts which embody fundamental principles of neural design, and which therefore imply and illuminated many less evident facts and predictions. The postulates reveal their implications by being translated into rigorous mathematical expressions. On various occasions, the precision of the this mathematical language has uncovered unsuspected physical properties of the postulates, or corrected erroneous conclusions of prior heuristic thinking. In particular, the mathematics can be given a natural anatomical and physiological interpretation. The neural networks hereby derived can thus be rigorously analyzed both behaviorally and neurally.

Topics: Biological Learning, Mathematical Foundations of Neural Networks, Models: Other,

PDF download




Cross References