A temporal learning rule in recurrent systems supports high spatio-temporal stochastic interactions

摘要

The maximization of spatio-temporal stochastic interactions (called TIM) has been proposed as an information-theoretic organizing principle in neural systems which supports a high cooperativity among cells and complex correlation patterns. The present work shows that temporal learning rules induce a high (though not always maximal) stochastic interaction in Markov chains and probabilistic neural networks.