Abstract
Introduction. During binocular rivalry, transition phases often take up about half of the observation time. We have characterised transition durations in terms of their dependence on stimulus strength and their distributions, providing constraints on models complementary to those posed by dominance durations (VSS 2005). Trying to reconcile existing models with data on transition phases, we find that realistic behaviour requires an interaction between deterministic factors (adaptation, mutual inhibition) and noise, leading to our present question: What is the relative importance of deterministic and stochastic forces underlying the change from one percept to the other?
Results. We present a model of transitions in binocular rivalry, which relates distributions of transition durations to the relative strength of deterministic and stochastic components. It starts from the accepted idea that binocular rivalry can be described as a non-linear system with two stable states (attractors), in which transitions are initiated when adaptation reduces the stability of one attractor. The model treats transitions as a random walk (the stochastic component) in a flow field (the deterministic component), from the destabilised attractor to the other one. Transition durations for various contrast conditions differed from traditional Gamma distributions and were better described by our model's two-parameter distribution. We found consistent effects of stimulus contrast on fit parameters, indicating a stronger relative influence of noise at lower contrast. Further work includes testing the effect of specifically manipulating the noise content of the stimulus during transitions.