The efficiency of visual transparency

This thesis examines the phenomenon of visual transparency in a novel application of the efficiency approach. Transparency provides a useful stimulus to probe the visual mechanisms that underlie the visual surface representation, introduced in Chapter One. Previous research has found that there is a cost in processing visual transparency defined purely by motion or stereo cues. This has been interpreted in terms of visual mechanisms constraining the recovery of transparency. However, the cost for transparency may reflect the increased complexity of the stimuli. To address this issue I computed the efficiency for motion and stereo defined transparency tasks by comparing human performance with that of the ideal observer. The efficiency approach has two key advantages over traditional psychophysical measures: 1) it provides a performance measure normalised relative to the available information, 2) it is an absolute measure and can be compared directly across diverse tasks. I provide a review of the efficiency approach in Chapter Two. In Chapter Three, I present a study of the efficiency for speed discrimination of transparent random dot stimuli and comparable non-transparent random dot stimuli, as a function of the speed ratio and the dot density of the stimuli. In Chapter Four, I present a study of the efficiency for depth discrimination of transparent and non-transparent random dot stereograms, across a range of disparity ratios and dot densities. In Chapter Five, I present an extension of the efficiency approach to the motor domain, for the smooth pursuit of high-density transparent and non-transparent random-dot stimuli. Finally, in Chapter Six I provide physiologically plausible accounts of the findings