To study the visual processing of periodic and aperiodic patterns, we have analyzed neuronal responses in areas V1 and V2 of the visual cortex of alert monkeys during behaviorally induced fixation of gaze. Receptive field eccentricities ranged between 0.5° and 4°. We found cells that responded vigorously to gratings, but weakly or not at all to bars and edges. In some cells the aperiodic stimuli even reduced the activity below the spontaneous level. The distribution of a bar-grating response index indicated a discrete population of 'grating cells' characterized by more than 10-fold superiority of gratings. We estimated that these cells have a frequency of 4% in V1 and 1.6% in V2, and that about 4 million grating cells of V1 subserve the central 4° of vision. The converse, cells that responded to isolated bars but not to gratings of any periodicity, was also observed. The grating cells of V1 were mostly (23 of 26) found in layers 2, 3, and 4B. They preferred spatial frequencies between 2.6 and 19 cycles/degree (median, 9.3), with tuning widths at half-amplitude between 0.4 and 1.4 octaves (median, 1.0). Their tunings were narrower, and their preferred frequencies higher, than those of other cells on average. Grating cells were also narrowly tuned for orientation. Those of V2 were similarly selective. The responses of grating cells depended critically on the number of cycles of the gratings. With square waves of optimum periodicity responses required a minimum of 2-6 grating cycles and leveled off at 4-14 (median, 7.5). The corresponding receptive field widths were 0.34-2.4° (median, 0.78°) for V1 and 0.72-2.4° (median, 1.4°) for V2. Grating cells typically gave unmodulated responses to drifting gratings, were unselective for direction of motion, and were strongly activated also by stationary gratings. Half of those of V1 were monocular, the others binocular, some showing strong binocular facilitation and disparity sensitivity. Length summation was usually monotonic, but strong end-inhibition was also observed. In contrast to other cells, grating cells were not activated by harmonic components. Spatial-frequency response curves for sine-wave, square-wave, and line gratings were similar. Square-wave gratings of one-third the preferred frequency failed to excite the cells, while the isolated 3f component (f = the fundamental of the square wave) of these gratings evoked strong responses. In spite of the nonlinear features, grating cells had low contrast thresholds. The detection performance of individual cells reached the performance of monkeys known from behavioral studies. Since grating cells were not activated by the proper spatial- frequency components of edges and bars, and failed to signal the higher harmonics of gratings, which are essential for perception, we conclude that these cells do not serve a spatial-frequency analysis of the stimulus. They seem to be specialized for the detection of periodic patterns and may play a role in perception of texture. Responses to checkerboard patterns and to gratings with jittered periodicity supported this assumption. We argue that some studies designed to investigate spatial-frequency channels in human vision may have tested the performance of grating cells.
|Original language||English (US)|
|Number of pages||19|
|Journal||Journal of Neuroscience|
|State||Published - 1992|
ASJC Scopus subject areas