Detection and identification of monocular, binocular, and dichoptic stimuli are mediated by binocular sum and difference channels

How are signals from the two eyes combined? We asked whether the mechanisms that limit detectability of simple binocular and dichoptic stimuli also set the limits for their identification. For example, at low contrasts, can we (a) identify monocular versus binocular stimulation and/or (b) identify stimuli that are the same in both eyes (e.g., both light discs or both dark) versus stimuli with opposite polarity (light disc in one eye, dark disc in the other). For the same- versus opposite-polarity tasks, mean proportions of correct trials for detection and for identification were almost identical. This is the classic signature of separate mechanisms for the two stimuli in question. For the monocular versus binocular task, however, identification (one eye or two?) was notably worse than detection, but these very different outcomes do not demand fundamentally different explanations. We developed a model with binocular sum and difference channels and formulated the identification task in a two-dimensional decision space whose coordinates were the sum and difference channel responses. This space was ideally suited to the same versus opposite polarity tasks, having orthogonal response axes (90° apart) for these stimuli. But monocular discs stimulated both channels, with greater overlap of monocular and binocular response distributions, hence greater perceptual confusion and poorer identification. When bias and uncertainty were also accounted for, the model fit to identification data was excellent. We conclude that the same binocular sum and difference channels are used in stimulus detection and in perceptually encoding the degree of difference between inputs to the two eyes.