A Cyclic Sequential Sampling Model of Bistable Auditory Perception

Abstract

We develop a cyclic sequential sampling model of bistable perception, based on the pioneering work of Vickers (1972). The model has two key parameters: a drift rate that measures the information in favor of one percept over the other; and a boundary separation that measures the evidence required by an observer to establish a percept. We implement the model within a graphical Bayesian framework, and apply it to data from several participants measuring their bistable perception for ambiguous auditory stimuli. We show that the model fits the distribution of latencies between perceptual reversals well, that the inferred drift rate parameter changes systematically as the auditory stimulus is manipulated to favor one percept over the other, and that the boundary separation parameter changes over participants to measure individual differences in their bistable perception.


Back to Saturday Papers