Stimulus strength and dominance duration in perceptual bi-stability. Part I: the unperceived stimulus affects the very first dominance epoch

Abstract

Levelt (1967) found an intriguing relation between the strength of binocular rivalry stimuli and their dominance durations. When one eye's stimulus is strengthened (eg by increasing contrast), the mean dominance duration of that eye does not change; instead, the mean duration of the other (unaffected) eye decreases. This observation imposes significant constraints on models of rivalry. Here we propose that it is particularly useful for testing different accounts of the role of adaptation in rivalry. Adaptation is commonly thought to be involved in bi-stable alternations. But the relative role of a decay in strength of the dominant (perceived) stimulus versus a recovery from adaptation of the suppressed (unperceived) stimulus is not known. We addressed this by examining the duration of the very first dominance epoch. Methods: contrast was manipulated for each eye independently in a full-factorial design. Trials were sorted based on which eye was dominant first, and the mean duration of the first epoch was calculated for each eye. Results: Levelt's observation holds already in the first dominance epoch; the mean duration of the first stimulus seen depended on the contrast of the (as yet unperceived) stimulus projected to the other eye. In the next abstract (Part II, Hupé and Rubin) we present evidence that Levelt's result can be generalized to bi-stable motion displays, and there, too, we find that the effect of the unperceived percept on the dominance duration of the other percept is present already in the first dominance epoch. Conclusions: these results cast a doubt on models which make use of recovery from adaptation of the suppressed stimulus to explain Levelt's result. In the very first dominance epoch, the suppressed stimulus is not yet adapted: it is not only unperceived at that moment — it has not been perceived (dominant) at all yet. These observations impose new, stronger constraints on models of bi-stable perception.