Simple- and complex-cell response dependences on stimulation parameters

We studied the response time course and amplitude dependence on stimulation parameters in cat cortical visual neurons to determine their receptive-field spatial-summation characteristics. Response poststimulus time (PST) histograms of cortical simple cells to contrast-reversal grating stimulation generally have a single peak for each stimulus temporal cycle, though the responses appear rectified. In response to contrast-reversal grating stimulation the general PST histogram time course for complex cells is two peaks, though often these peaks are of different amplitudes. The time course of complex-cell responses, and the ratio of these two response peaks often varies with stimulation parameters. The appearance of a single response peak in simple cells is reflected in the dominance of the odd harmonic Fourier portion, whereas the half-wave rectification leads to a considerable even harmonic portion. Still, this even portion is never significantly greater than the odd portion. When complex cell PST histograms have two nearly equal peaks, Fourier transformation reveals almost only even harmonic components. When the histogram contains two peaks of unequal amplitude Fourier analysis reveals large odd and even components. An even:odd Fourier harmonic portion ratio larger than 1 may be seen as a defining characteristic of complex cells, differentiating them from simple cells. Histograms with two unequal peaks appear "mixed," containing something of the "pure" single-peaked response and something of the pure double-peaked response. The degree to which the response is mixed may be measured by the ratio of the even:odd portion amplitudes. There is a great degree of variability with stimulation parameters (both spatial phase and spatial frequency) of the time course of mixed responses as opposed to the case of responses that have two equal peaks independent of stimulation grating phase and frequency. In both simple and complex cells there is a close coincidence of the spatial frequency ranges over which the even and odd portions are substantial, though many complex cells show a periodic variation of the even:odd portions ratio. This spatial-frequency dependence differs from that of LGN Y-cells where the odd portion dominates at low spatial frequencies and the even portion at high spatial frequencies. The ratio of even-to-odd portion cut-off is close to 3:1 in all Y-cells, a characteristic we did not find in cortical simple or complex cells. We suggest, therefore, that the nonlinearity of these complex cells does not derive from that of Y-cells.(ABSTRACT TRUNCATED AT 400 WORDS)