A Simplified, Low-Level Account of the Bistable Perception Yielded by Objects Drifting toward and Past One Another

Abstract

It is conjectured that the relative frequency of streaming vs bouncing percepts yielded by two objects drifting toward and past one another reflects the relative ‘strength’ of the underlying ‘continuity’ and ‘discontinuity’ motion vectors in the stimulus. The frequency of the two perceptual states should then be predictable from the relative effective contrast of these components which, in turn, is governed by the relative speed sensitivity and spatiotemporal integration range of the underlying sensors. Three distinct experiments were run to test this hypothesis. Percentage bouncing (%B) was assessed with two equal variance Gaussian blobs of equal contrasts as a function of their trajectory length at a given speed (experiment 1), and as a function of their relative physical contrast at a given speed and trajectory length (experiment 2; this latter manipulation was meant to control the relative input to the two competing motion sensors). Apparent contrast of a single drifting blob was independently measured as a function of its speed and trajectory length (experiment 3). The %B obtained in experiment 1 was inferred from experiments 2 and 3 by means of a transformation relating relative apparent contrast (at a given trajectory length) and %B (for that same trajectory) as a function of the relative physical contrast of the continuity and discontinuity motion components. This transformation translates the effect of trajectory length (and speed) in terms of an objective contrast effect. Measured and predicted %B in experiment 1 (ie as a function of trajectory length at a given speed) are in rather good agreement. The phenomenology of simple (and perhaps more complex) Ternus-type stimuli can thus be derived from the sensitivity/apparent contrast and the spatiotemporal integration characteristics of the underlying motion sensors.