Neurons in striate cortex limit the spatial and temporal resolution for detecting disparity modulation.

Stereopsis is the process of seeing depth constructed from
binocular disparity. The human ability to perceive modulation of
disparity over space (Tyler, 1974; Prince and Rogers, 1998; Banks
et al., 2004a) and time (Norcia and Tyler, 1984) is surprisingly
poor, compared with the ability to detect spatial and temporal
modulation of luminance contrast. In order to examine the
physiological basis of this poor spatial and temporal resolution of
stereopsis, I quantified responses to disparity modulation in
disparity selective V1 neurons from four awake behaving monkeys. To
study the physiological basis of the spatial resolution of
stereopsis, I characterized the three-dimensional structure of 55
V1 receptive fields (RF) using random dot stereograms in which
disparity varied as a sinusoidal function of vertical position
(“corrugations”). At low spatial frequencies, this produced a
modulation in neuronal firing at the temporal frequency of the
stimulus. As the spatial frequency increased, the modulation
reduced. The mean response rate changed little, and was close to
that produced by a uniform stimulus at the mean disparity of the
corrugation. In 48/55 (91%) of the neurons, the modulation strength
was a lowpass function of spatial frequency. These results suggest
that the neurons have fronto-parallel planar receptive fields, no
disparity-based surround inhibition and no selectivity for
disparity gradients. This scheme predicts a relationship between RF
size and the high frequency cutoff. Comparison with independent
measurements of RF size was compatible with this. All of this
behavior closely matches the binocular energy model, which
functionally corresponds to cross-correlation: the disparity
modulated activity of the binocular neuron measures the correlation
between the filtered monocular images. To examine the physiological
basis of the temporal resolution of stereopsis, I measured for 59
neurons the temporal frequency tuning with random dot stereograms
in which disparity varied as a sinusoidal function of time.
Temporal frequency tuning in response to disparity modulation was
not correlated with temporal frequency tuning in response to
contrast modulation, and had lower temporal frequency high cutoffs
on average. The temporal frequency high cut for disparity
modulation was negatively correlated with the response latency, the
speed of the response onset and the temporal integration time
(slope of the line relating response phase and temporal frequency).
Binocular cross-correlation of the monocular images after bandpass
filtering can explain all these results. Average peak temporal
frequency in response to disparity modulation was 2Hz, similar to
the values I found in four human observers (1.5-3Hz). The mean
cutoff spatial frequency, 0.5 cpd, was similar to equivalent
measures of decline in human psychophysical sensitivity for such
depth corrugations as a function of frequency (Tyler, 1974; Prince
and Rogers, 1998; Banks et al., 2004a). This suggests that the
human temporal and spatial resolution for stereopsis is limited by
selectivity of V1 neurons. For both, space and time, the lower
resolution for disparity modulation than for contrast modulation
can be explained by a single mechanism, binocular cross-correlation
of the monocular images. The findings also represent a significant
step towards understanding the process by which neurons solve the
stereo correspondence problem (Julesz, 1971).