Abstract
Combination of signals from the two eyes is the gateway to stereo vision. To gain insight into the stages of binocular signal processing, we studied binocular summation for luminance-modulated horizontal gratings (L or LM; 1st order) and contrast-modulated gratings (CM; 2nd order). To create LM, a sine-wave luminance grating (L) was added to dynamic 2-D bandpass noise (centre frequency 6 c/deg, rms contrast 0.2), while for CM the noise contrast was spatially modulated by the signal. In a 2AFC detection task we measured contrast thresholds (L, LM) or modulation thresholds (CM) for the signal grating (spatial frequency 0.75 c/deg, duration 0.2 s) shown to one eye, to both eyes in-phase, or to both eyes in anti-phase. For LM and CM, the noise was always binocular, even when the signal was monocular. The noise was the same in both eyes, or was uncorrelated, or anti-correlated. As in previous work, mean binocular thresholds for luminance gratings (L) were about 5 dB better than monocular thresholds - close to perfect linear summation (6 dB). The same was true for LM and CM: the binocular advantage was 5–6 dB even when the noise was uncorrelated or anti-correlated between the eyes. The latter results imply that binocular combination for CM arises from summation of envelope responses, and not directly from summation of the carrier patterns. In all cases antiphase signals showed no summation, but in most cases also showed little or no cancellation. This is consistent with half-wave rectification of both 1st-order and 2nd-order responses before binocular summation: only like-signed signals are combined. These results support an extended FRF (filter-rectify-filter) model: FRFRB, in which two filter-rectify stages precede binocular summation (B). This architecture is strikingly similar to that proposed for cat cortex area 18 (Tanaka & Ohzawa, 2006, Journal of Neuroscience).
Supported by BBSRC grant (BB/H00159X/1) to MAG, EPSRCgrant (EP/F026269/1) to AJS.