Binocular contrast discrimination needs monocular multiplicative noise

Abstract

We report the surprising result that although binocular vs monocular viewing of a sinusoidal grating has approximately the same perceived contrast, contrast discrimination is ͩ0;2 better for binocular viewing. We argue that the most plausible class of models with this property has a gain control stage with ocular crosstalk whereby the effective contrast of each eye is halved (see Ding and Sperling, 2006), followed by multiplicative noise, and then the summation of the two eyes' signals and noise going to the decision stage. To pin down the model, we performed three experiments using identical stimuli to measure perceived phase, perceived contrast, and contrast discrimination of a cyclopean sinewave. On each trial, horizontal sinewaves (0.68 cpd) were presented independently to the two eyes using a mirror stereoscope. Perceived phase was measured using a paradigm adapted from Ding and Sperling (2006). (The perceived phase is sensitive to the ratio of right and left eye contrasts and is important for pinning down the gain control parameters.) Perceived contrast was measured by matching the contrast of a binocular grating to a monocular standard. Contrast discrimination was measured with both pedestal and test sinewaves either in the same eye (monocular), in different eyes (dichoptic), or in both eyes (binocular). A modified Ding-Sperling model (Ding, Klein and Levi, in press) plus a contrast discrimination mechanism were used to fit all three data sets simultaneously. We tested four different model configurations: multiplicative noise or a non-linear contrast transducer could be either inserted before or after the combination site. Only multiplicative noise inserted before the combination site was able to account for the data. Meeting abstract presented at VSS 2013