Results

We made extracellular recordings from 112 orientation-tuned complex cells in the primary visual cortex of 22 macaque monkeys. Within this population of neurons, 43 cells had both cross-orientation suppression and surround suppression measurements collected. We had an additional 52 cells with surround suppression stimuli only and 17 cells with cross-orientation suppression stimuli only. We report the responses of only complex cells because for simple cells the modulation of response as the gratings drifted prevented us from easily determining response latencies. It is reported that surround suppression operates similarly for simple and complex cells (Cavanaugh et al., 2002b; Mizobe et al., 2001; Rose, 1977; Nelson and Frost, 1978; Walker et al., 2000; Dreher, 1972; Kato et al., 1978; Levitt and Lund, 2002). The characteristics of cross-orientation suppression are also reported to be similar for simple and complex cells, although simple cells tend to show stronger suppression (Bonds, 1989; Morrone et al., 1982). In order to test the generality of our results, we also recorded from 21 simple cells in V1 using a modified cross-orientation stimulus with static gratings.

Figure 6-1 contains example tuning curves which are representative of cells that show cross-orientation suppression and surround suppression. After finding the optimal grating orientation and spatial and temporal frequency, we used tuning curves such as this (Fig. 6-1A) to assess the presence of surround suppression and to determine the size of the CRF (see Methods in Ch. 2). We used this size for our dynamic stimuli designed to test suppression from inside and outside the CRF.

In Figure 6-1A, the example cell shows a peak response when the grating is between one and two degrees in diameter. As the grating increases in size beyond two degrees, the cell's response is reduced. When the grating is nearly six degrees across, the cell's response is approximately 2/3 of its peak response. This response pattern is characteristic of cells that have iso-orientation surround suppression. Figure 6-1B shows a tuning curve for an example cell's response to a preferred base grating (50% contrast) as the contrast of a mask grating increases. The cell's response decreases as the mask contrast increases beyond about 6%. When the mask reaches 50% contrast, the cell's response is reduced by more than half. This tuning curve is representative of cells that show cross-orientation suppression.

Figure 6-1: A, As the size of a preferred grating stimulus is increased, this cell shows response suppression beyond about 1.5$^\circ$. When the grating is at its largest size, the cell's response is reduced by approximately 40%. B, The response to a preferred grating can also be reduced by introducing an orthogonal mask. As the contrast of the mask increases (added onto a 50% contrast base grating), this cell shows response suppression. When the mask is at 50% contrast (total contrast of mask and base is 100%), the cell's response is reduced by about 70%.

We then presented a random, dynamic stimulus to test cross-orientation or surround suppression. In the surround stimulus, we presented a full-contrast drifting grating at either the preferred orientation or an orthogonal orientation to either the CRF or surround. Figure 6-2A shows this sequence for seven periods of the stimulus. At the end of each period, each grating will randomly be chosen to either continue drifting seamlessly or change orientation by 90$^\circ$. In Figure 6-2, we have prepared a particular sequence of transitions for the purpose of illustration. In the actual stimulus presented to cells, the order of the stimuli is random. For example, in the first transition in Figure 6-2A, the CRF stimulus changed from orthogonal to preferred while the surround stimulus remained orthogonal. We termed this transition ``onset'', because it typically produced a strong excitatory response. In the next transition, labeled ``suppression'', the surround stimulus changes from orthogonal to preferred while the CRF stimulus remains preferred. When the surround stimulus changes back to orthogonal in the next transition (``release''), the suppression is removed. When the CRF stimulus changes to orthogonal (``offset''), the response typically drops off quickly. We were also interested in a fifth transition (``onset and suppression''), because it sets the timing of the excitatory CRF response against suppression from the surround. It is worth noting that we were not always able to determine the response latency for all four transitions for every cell. This accounts for the differing number of cells in the figures that follow.

Figure 6-2: A, The stimulus for testing surround suppression was comprised of a grating with a size found to be optimal for the CRF and an annular surround grating configured to nearly fill the screen. Each of these gratings could be in either a preferred or an orthogonal orientation. After each cycle of drift, the center and surround gratings would each randomly be chosen to either continue drifting seamlessly or change to an orthogonal orientation. We labeled the transitions in which the center grating became preferred or orthogonal (while the surround remained orthogonal) as ``onset'' and ``offset''. Similar transitions for the surround grating were labeled ``suppression'' and ``release''. We called the transition of both gratings from orthogonal to preferred ``onset and suppression''. B, We designed the cross-orientation stimulus to be analogous to the surround stimulus. The size of the grating stimuli were chosen in the same way as for the surround stimulus. Each grating was at 50% contrast, and could be in either a preferred or an orthogonal orientation. In this case, after each cycle of drift, the base and mask grating were randomly chosen to either be present or absent. This created four possible stimuli - a blank screen, a preferred grating or an orthogonal grating at 50% contrast, or a plaid at 100% contrast. We labeled the transitions in which the preferred grating was added or removed while the orthogonal grating was absent as ``onset'' and ``offset''. If the preferred base grating was present, the addition or removal of the orthogonal mask were labeled ``suppression'' and ``release''. Changing from a blank to the plaid stimulus was termed ``onset and suppression''.

In the cross-orientation stimulus, we presented either a blank or a grating at half-contrast within the CRF at either a preferred or an orthogonal orientation. Figure 6-2B shows this sequence for seven periods of the stimulus. In these transitions, we used the same labels as in the surround stimulus (Fig. 6-2A), except the suppressive stimulus is now an orthogonal mask grating. When the two gratings are present together the resulting stimulus is a full-contrast plaid. For both the cross-orientation and surround stimuli, the two gratings (base and mask for the former; CRF and surround for the latter) each had two possible orientations (preferred and orthogonal), which produced four possible stimuli (and therefore 16 possible transitions). After each cycle of drift, a new one of the four possible stimuli was chosen at random.