A small difference between the images on the left and right retina, called 'binocular disparity', produces different magnitudes of phase shifts depending on the spatial frequency (SF). More specifically, the magnitude of phase shift may be proportional to SF for a constant binocular disparity. While many binocular neurons in the primary visual cortex (V1) are selective to binocular disparity, it's been unclear whether information from different SF channels is integrated in V1 single cells such that these cells can be tuned to the same disparity across a broad range of SF.
To ask this question, single neurons were recorded extracellularly in the V1 of anesthetized and paralyzed adult cats while sine-wave gratings were presented dichoptically. The SF and phase parameters were selected randomly in each eye for every stimulus frame. After recording, tunings to binocular phase difference were calculated for all SF combinations between the two eyes to estimate the strength of binocular interaction in the frequency domain.
We found that some complex cells show binocular interaction only for a limited range of SFs if the SF for one eye is fixed and that for the other eye varied, but exhibit binocular interaction for a much broader range of SFs if SFs for the two eyes were varied together. This characteristic cannot be explained by a single disparity detector. Therefore, the binocular receptive fields of a subset of single complex cells in V1 may be constructed by pooling multiple disparity detectors tuned to different SFs. This raises a question whether these multiple disparity detectors are tuned to the same disparity. To examine this issue, binocular receptive fields were reconstructed in a different restricted SF range. They were tuned to the same disparity regardless of SF bands used to reconstruct binocular receptive fields. This suggests that disparity detectors for different SF channels are pooled in these neurons to encode the same disparity across a broad range of SF.