Spectral properties of V4 neurons in the macaque

Spectral properties of 129 cells in the V4 area of 5 macaque monkeys were studied quantitatively with narrow-band and broad-band colored lights. The large majority of cells exhibited some degree of wavelength sensitivity within their receptive fields. The half-bandwidth of the primary peak in the spectral-response curve was less than 50 nm for 72% of the cells; the mean half-bandwidth of these cells, 27 nm, is similar to that found for color-opponent ganglion cells and cells in the parvocellular dorsal lateral geniculate nucleus (dLGN). Contrast-response functions indicated that the narrow spectral tuning of these cells derived from cone opponent interactions. From comparison of receptive-field sizes, we suggest that a typical V4 neuron sums inputs that ultimately derive from several thousand ganglion or parvocellular dLGN cells. In spite of their wavelength sensitivity, most V4 cells had properties that would not fit some simple criteria for classification as "color selective." First, few cells showed overt signs of color opponency, namely, on-inhibition or off-excitation to spectrally opponent wavelengths. Second, about 30% of the cells in V4 had spectral-response curves with 2 peaks. (The wavelength distribution of these second peaks was almost identical to that of primary peaks, and combinations of peak wavelengths were fairly random.) Third, most cells responded to white light; overall, the response to white light was about 60% of that to the best narrow-band or broad-band colored light. Similarly, most V4 cells gave at least a small response to all or nearly all of the different broad-band colored lights we presented. Therefore, a given V4 cell is very likely to respond to most of the colored or white surfaces in natural scenes. These combinations of response properties probably explain the widely divergent percentages of "color" cells reported in previous studies of V4. The most unusual spectral property we found in V4 was a large, spectrally sensitive surround outside the "classical receptive field" of most cells. Although stimulation of the surround by itself did not cause any response, surround stimulation could completely suppress the response to even the optimally colored stimulus in the receptive field. In general, the optimal wavelengths for receptive-field excitation and surround suppression were the same or nearly so. Thus, "color contrast" may be computed in V4. In some cases, contrasting wavelengths in the surround caused moderate enhancement of response to a receptive-field stimulus.(ABSTRACT TRUNCATED AT 400 WORDS)