Neural chaos scheme of perceptual conflicts

Abstract

Multistable perception is perception in which two (or more) interpretations of the same ambiguous image alternate while an obserber looks at them. Perception undergoes involuntary and random-like change. The question arises whether the apparent randomness of alternation is real (that is, due to a stochastic process) or whether any underlying deterministic structure to it exists. Upon this motivation, we have examined the spatially coherent temporal behaviors of multistable perception model based on the chaotic neural network from the viewpoint of bottom-up high dimensional approach. In this paper, we focus on dynamic processes in a simple (minimal) system which consists of three neurons, aiming at further understanding of the deterministic mechanism.