Journal of Neurophysiology

Spatial Organization and Magnitude of Orientation Contrast Interactions in Primate V1

H. E. Jones, W. Wang, A. M. Sillito

Abstract

We have explored the spatial organization of orientation contrast effects in primate V1. Our stimuli were either concentric patches of drifting grating of varying orientation and diameter or grating patches displaced in x–y coordinates around a central patch overlying the classical receptive field (CRF). All cells in the sample exhibited response suppression to iso-oriented stimuli exceeding the CRF. Changing the outer stimulus orientation revealed five response patterns: 1) orientation alignment suppression (17% of cells)—a suppressive component tuned to the same orientation as the cell's optimal, 2) orientation contrast facilitation (63%)—responses to orientation contrast stimuli exceeded those to the center stimulus alone, 3) nonorientation specific suppression (3%), 4) mixed general suppression and alignment suppression (14%), and 5) orientation contrast suppression (14%)—cross-oriented stimuli evoked stronger suppression than iso-oriented stimuli. Thus most cells (94%) showed larger responses to orientation contrast stimuli than to iso-oriented stimuli, and over one-half showed orientation contrast facilitation. There appeared to be a spatially structured organization of the zones driving the different response patterns with respect to the CRF. Nonorientation-specific suppression and orientation contrast suppression were predominantly evoked by orientation contrast borders located within the CRF, and orientation contrast facilitation was mainly driven by surround stimuli lying outside the CRF. This led to different response patterns according to border location. Zones driving orientation contrast facilitation were not necessarily contiguous to, nor uniformly distributed around, the CRF. Our data support two processes underlying orientation contrast enhancement effects: a simple variation in the strength of surround suppression drawing on the fact that surround suppression is tuned to the same orientation as the CRF and a second process driven by orientation contrast that enhanced cells' responses to CRF stimulation by either dis-inhibition or orientation contrast facilitation. We suggest these processes may serve to enhance response levels to salient image features such as junctions and corners and may contribute to orientation pop-out.

INTRODUCTION

Primate V1 cells show strong surround suppression to moving stimuli (Sceniak et al. 1999; Sillito et al. 1995), and we recently dissected the way this is modulated by reversing the direction of the surround stimulus (Jones et al. 2001). The data suggested that direction contrast may reduce both local and lateral inhibitory influences and possibly enable lateral facilitatory influences. In this work we consider the influence of the orientation of the surround stimulus. While it has long been recognized that, for many V1 cells, surround suppression is tuned to the same orientation as the cell's excitatory response and is diminished as the orientation of a surrounding stimulus deviates from this (e.g., Blakemore and Tobin 1972; Gilbert and Wiesel 1990;Kastner et al. 1997; Knierim and Van Essen 1992; Lamme 1995; Orban et al. 1979; Sillito et al. 1995), our observations suggest that more complex interactions may apply. Previously we reported that, for some primate V1 cells, orientation contrast could induce an effect that was greater than that predicted from a simple modulation of the strength of surround suppression (Sillito et al. 1995). Other studies report findings consistent with this view (Knierim and Van Essen 1992; Li and Li 1994; Nothdurft et al. 1999), although some have not seen this effect in the cat (Sengpiel et al. 1997;Walker et al. 1999). This issue is of great interest because such orientation contrast effects may link to mechanisms underlying perceptual pop-out for this class of stimuli (Bergen and Julez 1983; Treisman and Gelade 1980), or for moving stimuli, to a process extracting the motion of corners or angled junctions in images. Our work on direction contrast showed that the location of the border between center and surround stimuli with respect to the CRF strongly influenced the response pattern observed (Jones et al. 2001). Here, we explore the influence of orientation context between a central and surrounding stimulus generated in a range of spatial configurations. The results show that the location of borders with respect to the classical receptive field (CRF) is a key variable influencing both the strength and the way V1 cell responses are affected by orientation contrast. We discuss the mechanisms that may underlie these observations.

METHODS

We recorded single-unit responses from the parafoveal primary visual cortex of anesthetized [sufentanil (4 μg/kg/h) or halothane (0.1–0.4% in 70% N2O-30% O2)] and paralyzed (0.1 mg/kg/h vecuronium bromide) monkeys (Macaca mulatta). The animals were treated according to the published guidelines on the use of animals in research (EEC Directive 86/609/EEC, National Institutes of Health Guidelines for the Use of Laboratory Animals). Full experimental details are provided elsewhere (Jones et al. 2001).

Data were collected and visual stimuli were generated using a VS system (CED, UK) in conjunction with a Picasso Image Generator (John Daughman), presented on a Tektronix 608. Prior to running the experimental protocols described below, we quantitatively checked preferred orientation and direction of motion and classed cells as S or C type. We assessed the location and extent of the CRF using a battery of tests and derived CRF size from the test giving the largest value (Jones et al. 2001). First we explored the spatial locations from which a small patch of optimally oriented drifting grating evoked responses. We also defined the area (width and length) from which a moving bar elicited excitatory responses. We then varied the diameter of an optimally oriented patch of grating centered over the receptive field. Finally, we varied the inner wall diameter of an annulus of optimally oriented grating. To minimize adaptation effects from persistent presentation of optimally oriented stimuli and to generate control data for subsequent tests, the protocols varying patch size or annulus inner wall diameter also varied stimulus orientation. A blank stimulus was included in each block to assess spontaneous activity.

We used concentric sinusoidal gratings to explore the interactions between an inner stimulus patch centered over the CRF and an outer annulus. Contrast, spatial frequency, and drift rate were identical for both, and we varied the orientation of both stimulus components in a randomized interleaved sequence. The phase of our inner and outer stimuli were locked together with reference to the center of the display, thus iso-oriented concentric stimuli appeared as a single grating. Controls for responses to inner and outer stimuli alone were repeated periodically in the test sequences. Spatial frequency ranged from 1 to 4 cpd and drift rate spanned 1–4 Hz. Grating contrast [(L maxL min)/(L max+ L min)] was 0.36 with a mean luminance of 14 cd/m2. The interface border between cross-oriented inner and outer stimuli encompasses a range of spatial frequencies and orientations that might provide a separate or additional stimulus driving the observations. The calculated energy of these components is very low relative to those of the main stimuli. Nonetheless, we checked for effects from the interface by using a narrow gap (set to the mean unmodulated contrast level) between inner and outer stimuli. The data indicated that these boundary properties were not a factor underlying our findings. Individual cells were studied for 6–16 h and observations repeated several times.

We used a range of protocols to assess the influence of the spatial configuration of the stimuli on orientation contrast effects. First, we varied the diameter and hence the location, with respect to the CRF, of the border between center and surround stimuli (range, 0.3–6°). Second, we varied the outer diameter of the surround annulus (range, 2–9°). Third, we varied the inner diameter of the annulus while the diameter of the center patch was fixed. In this case, when the center patch diameter was smaller than the inner wall of the annulus, the luminance of the resulting gap was set to the mean luminance of the contrast gratings.

We also used two square grating patches to locate facilitatory zones. One patch, containing an optimally oriented drifting grating, was centered over the CRF. The other was presented in randomized sequence at a range of locations around the field either at the optimal or at the orthogonal orientation. Its contrast was also varied between two values, either 0.36 (to match the central patch) or 0 (no second stimulus present to assess the response to the central stimulus alone). Patch sizes ranged from 0.5° to 2°.

Quantifying responses

Responses were computed from the mean firing rate averaged over the full number of stimulus presentations. Typically we presented 3–5 stimulus cycles of each stimulus condition repeated over 5–20 trials. Cells were regarded as orientation biased if the ratio of the response to the optimal versus the nonoptimal orientation was 3:2 or greater and as orientation selective if the ratio was 3:1 or greater. We calculated a patch suppression index using the formula [1 − (R plat/R opt)] × 100, where R opt andR plat denote the responses to optimal and plateau stimuli, respectively (see Jones et al. 2001).

We compared the responses evoked by iso-oriented and orientation contrast stimulus configurations. Cells were regarded as showing “orientation pop-out” if the response to an optimally oriented center stimulus in the presence of an orthogonal outer stimulus (±30°) was significantly larger than the response when both stimuli were present at the optimal orientation (P < 0.05, paired t-test). In some cases, responses to orientation contrast stimuli exceeded the response to the center stimulus alone. Cells were only classed as “orientation contrast facilitation cells” if the response to the orientation contrast configuration was significantly larger (P < 0.05) than the responses to the iso-orientation configuration and the inner stimulus alone, and if the response enhancement evoked by the orientation contrast condition (normalized with respect to the response to the center stimulus alone) exceeded 10%. Cells were classed as “nonorientation specific surround cells” if there was no significant difference between the responses to iso-orientation and orientation contrast stimulus conditions (P ≥ 0.05). Cells were classed as “orientation contrast suppression cells” if the response to orientation contrast was significantly smaller (P < 0.05) than to the iso-orientation stimulus.

We graphically represented the data collected with bipartite concentric stimuli using two-dimensional iso-response contour maps and three-dimensional surface maps. Distance between contours was defined by (R maxR min)/(1 + number of levels);R min defined the first level, and we used a spline fitting algorithm to interpolate between responses. We used the same procedure to represent the data from the two patch experiments.

To explore the location of zones driving orientation contrast facilitation, we adapted methodology previously used in area MT (Xiao et al. 1997, see also Jones et al. 2001). We calculated the strength of facilitation for each surround stimulus location according to the formula F = [(R cs/R c) − 1] × 100, where F is the enhancement elicited by a surround stimulus location, R c is the response to the center stimulus, andR cs is the response to the combination stimulus. We then computed two selectivity indices (FIs) by calculating the length of the mean vectorFI=i=1nFi·sin(αi)]2+i=1nFi·cos(αi)]2i=1nFi where Fi is the magnitude of the surround facilitation at each surround angular location, αi. The Unimodal Selectivity Index (USI) was derived from the actual αi values, whereas the Bimodal Selectivity Index (BSI) was calculated with each αi value doubled. The USI assessed the tendency for surround facilitation to be concentrated in one location, whereas the BSI reflected the tendency for facilitation to be concentrated along an axis on opposite sides of the CRF. For both, a value of 1 indicated that only a single surround position (or axis) was effective in modulating activity, whereas a value of 0 denoted an uniform distribution. We used the Rayleigh test (Batschelet 1981) and the USI and BSI values to subdivide the response patterns into three groups: uniform surround facilitation (Rayleigh test, P > 0.05), asymmetric surround facilitation (Rayleigh test, P < 0.05 and USI > BSI), and bilaterally symmetric facilitation (Rayleigh test, P < 0.05 and BSI > USI).

We identified the most effective location by calculating the mean vector angle (Batschelet 1981). Thus the optimal angle OPA=arctanΣi=1nFi·sin(αi)Σi=1nFi·cos(αi) . For bilaterally symmetric surround cells, the OPA represents the angle of the axis through the two optimal surround locations.

RESULTS

Our data draw on observations from 70 cells (44 S and 28 C cells). All exhibited surround suppression for iso-oriented stimuli [mean suppression, 69.4 ± 2.17% (SE); range, 30–100%]. We observed no significant differences between S and C cells except in two instances highlighted below.

Response patterns to orientation contrast stimuli

We observed five response patterns using concentric center-surround stimuli. These are summarized in Fig.1 and were defined on the basis of the effect of varying the orientation of a surrounding stimulus on the response to an optimally oriented central stimulus (although the data were extracted from test sequences that randomly interleaved the orientation of both stimulus components).

Fig. 1.

Response patterns to orientation contrast stimuli. A andB: 2 examples of orientation alignment suppression. For each cell, the dotted line plots the response (i/s) to varying the orientation of an optimal diameter (1°) center patch of grating. Solid line shows the effect of varying the orientation of an outer annulus of grating (inner wall diameter 1°) in the presence of the optimally oriented inner patch. Gray shading denotes ±SE. Short black line denotes spontaneous activity level. C: orientation contrast facilitation. Tuning curves plot the cell's response to varying the orientation of an optimal diameter center patch presented alone (dotted line), varying the orientation of an outer annulus presented alone (dashed line), and varying the orientation of the outer annulus in the presence of the optimally oriented inner patch (solid line). The response to the orientation contrast configuration clearly exceeded that to the center stimulus. Patch diameter, annulus inner wall diameter, and interface diameter are 1°. D: mixed general suppression with a component of alignment suppression. Stimulus conventions as in A. Patch/interface diameter, 0.75°.E: nonorientation specific suppression. Stimulus conventions as in A. Patch/interface diameter, 2°. F: orientation contrast suppression. Tuning curve plots the effect of varying the orientation of an annulus in the presence of an optimally oriented center patch. Arrowhead denotes the response to the center patch (0.75°) presented alone. The annulus elicited no suppressive influence when presented at the cell's optimal orientation, but exerted potent suppressive effects as the orientation of the outer deviated away from the optimal orientation.

ORIENTATION ALIGNMENT SUPPRESSION.

The records in Fig. 1, A and B, typify the response expected from a suppressive mechanism tuned to a cell's optimal orientation. The dotted curve shows the response to varying the orientation of an optimal sized patch of grating. The solid line plots the response to varying the orientation of the surrounding annulus in the presence of the inner held at its optimal orientation. There is a clear suppressive effect tuned to the optimal orientation. As the annulus deviated from the cell's preferred orientation, the degree of suppression diminished to zero, leaving the response at the level evoked by the inner alone. This pattern of orientation pop-out (“orientation alignment suppression”) is broadly what would be expected from previous work. Twelve cells showed this response pattern.

ORIENTATION CONTRAST FACILITATION.

Another pattern is shown in Fig. 1 C, again using an optimal diameter center stimulus. Once more, the cell showed a clear suppressive effect tuned to the cell's optimal orientation. However, in this case, when the surround stimulus orientation deviated from optimal by more than 30°, the response increased substantially above that to the center stimulus (“orientation contrast facilitation”). This might be considered to reflect the addition of a component of orientation contrast facilitation above the modulation of the surround suppression, although dis-inhibition might also play a significant role. Forty-four cells showed this response pattern, and for over one-half (23/44) the responses to orientation contrast stimuli exceeded the response to any single stimulus tested. This latter group included individual cells showing increases of 300% or more.

MIXED GENERAL SUPPRESSION AND ORIENTATION ALIGNMENT SUPPRESSION.

Some cells (10) showing suppressive effects tuned to the optimal orientation also exhibited a nonorientation tuned suppressive component as shown in Fig. 1 D (“mixed general suppression and orientation alignment suppression”).

NON-ORIENTATION SPECIFIC SUPPRESSION.

We observed two cells (both S type) that exhibited the same level of suppression to any orientation of the outer stimulus (“nonorientation specific suppression,“ see Fig. 1 E).

ORIENTATION CONTRAST SUPPRESSION.

The final pattern is shown in Fig. 1 F. Here, the suppression driven by the surround stimulus was minimal at the optimal orientation (arrowhead denotes the response to the optimally oriented inner stimulus) and maximal when the orientation differed by roughly more than 30°. Although this might be considered to reflect what is often referred to as cross-orientation inhibition, clearly it is not simply that, and we have termed it “orientation contrast suppression.” Two cells (both S type) showed only this response pattern. Another eight cells also showed orientation contrast suppressive effects; however, for these, the response pattern depended on the interface diameter between the inner and outer stimuli. Generally, these cells showed orientation contrast suppression for borders within the CRF and orientation contrast facilitation for borders outside the CRF (see below).

For the first three response patterns described above, responses to iso-oriented stimuli were significantly smaller than to orientation contrast stimuli. Thus 94% of our sample (66/70 cells) showed orientation pop-out.

MAGNITUDE OF ORIENTATION CONTRAST EFFECTS.

For each response pattern, we compared the response modulation evoked by iso-oriented and orientation contrast stimulus configurations. This is summarized in Table 1, where the values denote the percentage decrement (−) or increment (+) observed with respect to the response to the center stimulus alone. For a given cell, the values for iso-orientation and orientation contrast configurations were derived from data using the same interface diameter between the inner and outer stimuli, and in each case the data derives from the interface diameter that evoked the most potent orientation contrast response modulation. Hence, the lower level of iso-orientation surround suppression observed for orientation contrast facilitation cells followed from the fact that orientation contrast facilitation was most often evoked by interface diameters larger than the CRF. Thus at these diameters, suppression was already partially implemented in the response to the center stimulus.

View this table:
Table 1.

Magnitude of orientation contrast effects, normalized with respect to the response to the test center stimulus alone

ANGULAR DEVIATION UNDERLYING THE EFFECTS.

Although it might seem convenient to reduce the description of these effects to either cross-orientation facilitatory or inhibitory influences, our data suggested much more focused interactions linked to much smaller differences in orientation between the inner and outer stimuli (e.g., Fig. 1, C andF). This is summarized in Table2, which shows the smallest deviation in orientation between the inner and outer stimuli that elicited a significant change in response and the angular deviation that evoked the maximal response change, for each response pattern. Overall, over 50% of cells showed significant changes in output for angular deviations of 22.5° or less. Even when considering angular deviations eliciting maximal changes in output, <25% of cells required angular deviations exceeding 67.5°.

View this table:
Table 2.

Degree of angular deviation underlying orientation contrast effects

Dissection of the spatial focus driving orientation contrast interactions

INTERACTIONS OBSERVED VARYING THE INTERFACE DIAMETER FOR CONCENTRIC STIMULI.

The location of the border between the inner and outer stimuli with respect to the CRF could markedly influence the magnitude and nature of the effect observed, and a given cell could show different effects for different border locations. The cell in Fig.2, A and B, showed orientation alignment suppression for a border diameter that equated to its CRF (0.75°, Fig. 2 A). Thus as the outer stimulus deviated from the cell's optimal orientation, the suppression diminished to zero, leaving the response at the level seen with the inner alone. Testing the cell with a 2° center stimulus revealed a different response pattern (Fig. 2 B). The center stimulus produced a smaller response because of surround suppression. However, varying annulus orientation, although revealing additional suppression at the optimal orientation, actually enhanced the response above that evoked by the 2° center stimulus as its orientation deviated from optimal. Thus for this spatial configuration, the cell exhibited strong orientation contrast facilitation, although the absolute response level was lower than to the optimal, CRF-sized, center stimulus.

Fig. 2.

Changing border location influences orientation contrast effects.A and B: tuning curves plot the responses of a cell to varying the orientation of a center patch presented alone (dotted line), varying the orientation of an annulus presented alone (dashed line), and varying the annulus orientation in the presence of the optimally oriented center patch (solid line) for 2 different border locations. Patch diameter, annulus inner wall diameter, and interface diameter were 0.75° in A and 2° in B. C and D: responses of another cell for 2 different border locations. Each tuning curve shows the effect of varying annulus orientation in the presence of an optimally oriented center stimulus. Arrowhead denotes the response to the optimally oriented center stimulus presented alone. Interface diameter was 0.5° in C and 1° in D. E–G: each tuning curve plots the response of the same cell to an annulus of varying orientation in the presence of an optimally oriented inner stimulus, for 3 different interface diameters. In each record, the arrow head denotes the response to the optimally oriented inner patch presented alone. For E, the border between the inner and outer stimulus (0.3°) was located within the classical receptive field (CRF, 0.75°). For F, the border was located at the edge of the CRF (0.75°), whereas for G, it was located outside the CRF (1.5°). Scale bars denote response magnitude (i/s).H–J: histograms summarize the percentage of cells tested at locations within (H), on the edge of (I), and outside (J) the CRF, exhibiting the various patterns of orientation linked surround influences. If a cell showed different effects at different border locations, it is represented more than once in the histograms. The category for points on the edge of the CRF refer to observations where the border diameter corresponded to the CRF diameter.

Another set of interactions is shown in Fig. 2, C andD. An inner stimulus smaller than the CRF (Fig.2 C) evoked orientationcontrast suppression. Conversely, for an inner stimulus diameter exceeding the CRF (Fig.2 D), the pattern switched to orientation contrast facilitation. In four cells, we were able to track changes from orientation contrast suppression, through orientation alignment suppression to orientation contrast facilitation with increasing border diameter (e.g., Fig. 2, E–G). Overall, this suggests the presence of three mechanisms that are isolated by different center-surround border interfaces. These respectively generate orientation contrast suppression, orientation alignment suppression and orientation contrast facilitation.

Although it was not possible to identify any single spatial focus with respect to the CRF that drove a specific pattern of orientation contrast dependent effects, our data suggested a general trend. The histograms in Fig. 2, H–J, summarize the percentage of cells tested at locations within, on the edge of and outside the CRF exhibiting the different response patterns. The majority of orientation contrast facilitatory effects were seen on the edge of and outside the CRF, while orientation contrast suppressive effects were virtually confined to border diameters within the CRF. Orientation alignment suppression was seen at all locations but showed the largest incidence on the edge of the CRF. It is possible that we might have observed more instances of orientation contrast suppression if we had tested loci within the very central region of the CRF; we did not use stimuli that would test this point in these experiments (DeAngelis et al. 1992). We only observed nonorientation specific suppression with interfaces within the CRF.

Although we emphasize a broad transition of effects for center-surround interfaces moving from within to without the CRF, we observed exceptions to this rule. Some cells (see above) only exhibited orientation alignment suppression. Others (n = 7) showed only orientation contrast facilitation. Figure3 shows a series of curves for the responses of a cell to an outer stimulus alone, an inner stimulus alone, and the combination of the two for a set of border diameters spanning within (0.5°), on the edge of (1°), and outside (2° and 3°) the CRF. In all cases, the combination stimulus evoked a potent facilitation of the response to the center alone when the outer orientation deviated from the optimal by 30° or more. These effects were very strong and reproducible, as is highlighted by the small error bars for the responses denoted by the gray shadows.

Fig. 3.

Changing border location did not always influence orientation contrast responses. Tuning curves in the left column plot the cell's response to varying the orientation of an annulus centered over the receptive field. Tuning curves in the middle columnplot the response to varying the orientation of a central patch, while those in the right column plot the response to varying the orientation of the annulus in the presence of a center patch held at its optimal orientation. The 4 rows correspond to stimulus components and combinations for interfaces at diameters of 0.5°, 1°, 2°, and 3°, respectively.

INTERACTIONS USING NONCONTIGUOUS INNER AND OUTER STIMULI.

We further examined the spatial characteristics of the mechanism driving orientation contrast effects by varying the inner and outer boundaries of the outer stimulus while holding the dimensions of the inner stimulus constant (e.g., Fig. 4). The effect of varying the outer wall of the outer stimulus on the combination response is shown in Fig. 4, A–C. There was possibly a very small enhancement of the level of orientation contrast facilitation in the step from 2° to 3°, but the most notable feature was the appearance of stronger iso-orientation suppression to the largest (4°) value tested for the outer diameter. On the other hand, when the outer wall of the outer annulus was held at 4° while varying its inner diameter from 1° to 3° (generating a 0°, 1°, and 2° gap, Fig. 4, D–F) orientation contrast facilitation was also largely unchanged, and if anything, most strongly expressed for the largest gap. Together, these observations suggest that processes driving orientation contrast were expressed throughout the surrounding area of visual space. Interestingly, in both sets, maximum surround suppression for the iso-oriented condition was seen when the surround stimulus was most extensive.

Fig. 4.

Exploring the spatial characteristics of mechanisms underlying orientation contrast effects. A–F: tuning curves document the effect of varying the outer (A–C) or inner (D–F) diameter of the outer annulus on orientation contrast effects. In each record, the tuning curve depicted by the dotted line shows the effect of varying the orientation of a 1° center patch. The solid line plots the effect of varying the orientation of the outer annulus while the inner stimulus was presented at its optimal orientation. In all cases, the diameter of the center patch was fixed (1°). In A–C, the inner diameter of the annulus was also fixed at 1°, while its outer diameter was varied through 2° (A), 3° (B), and 4° (C). In D–F, annulus outer diameter was fixed at 4°, and its inner diameter was varied through 1° (D—no gap), 2° (E—1° gap), and 3° (F—2° gap). G and H: surface/contour plots depict the same cell's responses to a stimulus configuration that comprised an inner stimulus (a square patch containing an optimally oriented grating drifting in the cell's preferred direction of motion) centered over the CRF and a second stimulus (another square patch of grating) which was positioned at a range of x–y locations around the central stimulus (as depicted by the schematic diagram). In G andH, the 2 plots show the modulatory effect of the outer stimulus when it was presented at either the same orientation as the center (0/0) or when it was orthogonally oriented (0/90). Center and second patch were 1° square for G and 2° square forH. For each plot, responses are normalized to the response elicited by the center patch alone (100%). Color scale bar depicts response scale values.

We dissected the mechanisms driving these interactions in a different way (Fig. 4, G and H) by using two square patches of drifting grating, one centered over the CRF (and presented at the cell's optimal orientation) and the second displaced over a set of coordinates to map the surrounding area of visual space. The second patch was presented either at the optimal or at the orthogonal orientation. The central point of each surface plot shows the response to the inner patch alone and the other locations in the surface plot show the way the response to this center stimulus was modified by the presence of the second stimulus at these locations. Using 1° patches (G), there was no effect from the second patch at 90° except for a fractional enhancement at one extreme side of the area tested, whereas the second patch at 0° evoked suppression from all locations around the field with a maximal effect at one end. This contrasts with the effects revealed using 2° stimuli (H). Here the second stimulus at 0° elicited virtually no effect from most locations, while that at 90° evoked a strong facilitation from all locations around the field. This suggests that some minimal level of spatial integration was required before the orientation contrast facilitation was enabled. Presumably any of the concentric surrounding stimuli met this criterion but a 1° patch did not.

The cell in Fig. 5 A showed an asymmetric pattern of suppression for an iso-oriented second stimulus. With an orthogonal stimulus, it showed a strongly asymmetric area of facilitation to the top right, separated from the CRF by a clear suppressive zone. We also tested its responses while varying the location of the inner wall of a surrounding annulus (Fig. 5,E–H). Essentially there was a simple modulation of surround strength as the annulus orientation was changed and this pattern held as a 0.2° and then a 0.5° gap was introduced. However, increasing the gap to 1° (Fig. 5 H) revealed a strong orientation contrast facilitation to the −90° direction of motion of the annulus. We suggest that this facilitatory effect was drawn from the asymmetric facilitatory zone shown in Fig. 5 A, but for the other stimulus configurations with smaller gaps it was masked by the counteracting inhibitory region.

Fig. 5.

A–D: these surface/contour plots depict the responses of another 4 V1 cells to the duo-square paradigm. Graphical conventions as Fig. 4. E–H: these tuning curves document the effect of varying the inner diameter of an annulus on the responses of the cell illustrated in Fig. 5 A. Conventions as in Fig. 4,D–F. In all records, the outer diameter of both the center patch and the annulus were fixed and annulus inner diameter was varied through 0.5° (A—no gap), 0.7° (B—0.2° gap), 1° (C—0.5° gap), and 1.5° (D—1° gap).

For our orientation contrast facilitation cell group (n= 44), the data from the two patch paradigm indicated a wide variety of spatial configurations driving facilitatory effects. We analyzed the data and subdivided cells into three groups using the criteria ofXiao et al. (1997; see methods). Only 14% of cells showed spatially uniform facilitation. An example is shown in Fig. 5 B, where the orthogonal stimulus evoked strong facilitatory effects from all locations around the receptive field and up to the edge of the area tested. The remainder had heterogenous surrounds. Fifty-three percent of cells showed spatially asymmetric facilitation where facilitation was concentrated in one surround location. For the example in Fig. 5 D, the orthogonal stimulus elicited a strong facilitatory effect from the bottom left location whereas all other locations evoked suppression. Thirty-three percent of cells showed bilaterally symmetric regions. For the example in Fig. 5 C, the orthogonal stimulus evoked powerful facilitation from the ends, but not the sides, of the field. For cells exhibiting heterogenous surrounds, we checked if the regions evoking facilitation were localized to the ends, sides or corners of the field. All regions evoked orientation contrast facilitation, there was no evidence to suggest that the mechanisms underlying the effect were concentrated in either end-zones or side-bands.

Orientation contrast interactions in nonorientation tuned cells

We had expected nonorientation tuned cells to exhibit uniform surround suppression. Surprisingly they did not. Of nine cells studied, four exhibited orientation alignment suppression and five orientation contrast facilitation. Thus these cells were strongly sensitive to orientation contrast but not to orientation per se. Figure6 documents the responses of a nonorientation tuned layer 4Cβ cell (CRF size 0.5°) to varying the orientation of both the center patch and outer annulus, at three border diameters. The diagonal trough running between the axes represents all those points where the orientation of the inner and outer stimuli were identical over a complete sequence of absolute orientations. Clearly, responses were minimal when the orientation of the inner and outer stimuli were the same. Moreover, for each border diameter, responses to orientation contrast stimuli exceeded the response to the inner stimulus alone and for the 0.5 and 1° configurations, the resultant output exceeded the response to the optimal inner patch presented alone. Essentially, iso-oriented stimuli drove the mechanism generating patch suppression, resulting in a reduced response, but this was disenabled when they were at different orientations. Thus despite the cell's lack of orientation tuning, its output was extremely sensitive to orientation differences.

Fig. 6.

Nonorientation tuned cells are sensitive to orientation contrast. Surface plots show the response of a nonorientation tuned layer 4Cβ cell to varying the orientation of both a center patch and outer annulus in a randomly interleaved sequence for 3 interface diameters (A, 0.5°; B, 1°; andC, 2°). Magnitude of the cell's response is shown by the height and shading of the contour. Diagonal running frombottom left to top right represents all those points where the orientation of the inner and outer stimuli were the same, over a complete sequence of absolute orientations. The cell gave a vigorous response to all those conditions where the orientation of the inner and outer stimuli differed, and a minimal response to all the conditions where they were the same. Difference in response between the aligned and nonaligned conditions was statistically significant for all 3 interface diameters (P < 0.001, Mann-WhitneyU test). Color scale bar shows response scale values.

Apart from the lack of orientation tuning, this response pattern was characteristic of many orientation tuned cells and for smaller interface diameters a very similar pattern could be seen in the surface plots documenting the responses of some orientation tuned cells.

Multi-orientation interactions in orientation tuned cells

As we varied the orientation of the concentric inner and outer stimuli in a randomized sequence for all cells studied, we had the data to generate the same surface plots for orientation tuned cells. For the cell in Fig. 7, A–D, interface diameters within the field (0.5°, 0.6°) evoked a response pattern resembling that of the 4Cβ cell. Conversely, when the interface diameter matched (0.75°) or exceeded (1.5°) the CRF, orientation tuned responses emerged. However, these were still modulated by the orientation of the outer, so that the response was minimal when the outer was at the cell's optimal orientation and maximal when it deviated by 45° or more. The cell was sharply orientation tuned to the center stimulus for all these diameters. Thus the interplay between the inner and outer stimuli, when the interface was within the field, seemed to facilitate responses to all orientations of the inner providing the outer orientation was different and suppress them when they were the same. This effect was not the trivial consequence of low orientation selectivity to the center patch at small diameters.

Fig. 7.

Orientation contrast interactions for an orientation tuned cell.A–D: surface plots show the response of an orientation tuned cell to simultaneously varying the orientation of a center patch and outer annulus for 4 border diameters (A, 0.5°;B, 0.6°; C, 0.75°; andD, 1.5°). Conventions as in Fig. 6. There was a significant difference in response magnitude between the aligned and nonaligned conditions at all interface diameters (P< 0.001, Mann-Whitney U test). E andF: cross-oriented stimuli that elicit little or no effect when presented in isolation evoke potent excitatory effects when presented together. Bar histograms show the responses (i/s) of 2 cells to an outer annulus presented at the cell's optimal orientation, a center patch presented at the orthogonal orientation, and the combination of the two. Error bars denote ±SE.

Figure 7, E–F quantifies this enabling effect of the outer on the inner stimulus for two further orientation tuned cells. The histograms plot the response to the center stimulus at 90° to the cells' optimal orientation, the outer stimulus at the optimal orientation and to the combination of the two. In both cases the firing level driven by the inner alone was at the spontaneous rate and the response to the combination was much greater than the sum of the components to the two single stimuli.

The pattern of interaction shown in Fig. 7, A andB, for border diameters within the CRF was seen for 28% of our sample. The majority of the other cells exhibited response profiles of the type shown in Fig. 7,C and D, at all interface diameters. The above observations suggest a focal interaction within the CRF that highlights the cell's output for orientation contrast irrespective of absolute orientation and indicate that this mechanism is present in a small proportion of orientation tuned cells as well as for nonorientation tuned cells. However, for most orientation tuned cells, the influence of the orientation contrast provided by the outer stimulus is only reflected in the modulation of the response to the inner at orientations in the envelope of orientations centered around the cells' optimal (e.g., Fig. 7,C and D).

Orientation contrast facilitation and firing level

We took great care to ensure that we avoided saturating responses and used a contrast of 0.36 for all observations. However, given the limited data on the prevalence of orientation contrast facilitation from previous studies, and the failure of certain groups (Sengpiel et al. 1997; Walker et al. 1999) to observe it, we considered possible links between orientation contrast facilitation and absolute response strength. First, we checked if the magnitude of the normalized enhancement reflected the cell's initial firing rate (i/s) to the center stimulus but there was no significant correlation (Spearman R) between the two values. In Table 3, we compare the mean firing level of orientation contrast facilitation cells to three stimulus conditions (the optimal center stimulus, the center stimulus used in the test paradigm, and the response to the orientation contrast configuration) first for the entire group, and then subdivided into “supra-optimal” and “facilitation only” categories. For the combined sample, there was no significant difference between the firing rates associated with the best single stimulus and the configuration evoking orientation contrast facilitation, although both were significantly larger than the response to the inner test stimulus (P < 0.001, Wilcoxon). Interestingly, two factors appeared to contribute to the identification of the “supra-optimal” group, these cells showed significantly lower responses to the best single stimulus as well as higher firing rates to orientation contrast stimuli. There was nothing in the details of the tests or our procedures to suggest that we may have missed the most appropriate “optimal” single stimulus for these cells. Overall, we favor the view that these cells were characterized by rather stronger inhibitory processes within the central regions of the CRF and that this accounted for the lower firing rates to the best single stimulus.

View this table:
Table 3.

Firing rates for orientation contrast facilitation cells

The two square dissections of the spatial organization of the mechanism integrating the facilitatory effect of stimulus configurations generating orientation contrast provided much smaller foci for stimuli generating the contrast effects than the annulus paradigm. If the integration performed some type of spatial summation one might predict that the effects elicited by the two square paradigm would be lower and that they might decline with distance from the CRF. Neither possibility was supported by the data. There was no statistical difference between the magnitudes of the effects elicited with contiguous concentric annuli or single squares of drifting grating in effective locations (Mann-Whitney Utest; P = 0.79). Given the high degree of asymmetry revealed by the analysis of the two square data this is hardly surprising because only certain regions seemed to drive the effects (consider in this case the examples in Fig. 5) and hence spatial summation by the concentric stimuli may not exert larger effects because the regions driving the effects are circumscribed for many cells.

Direct surround effects

An important question concerns the direct effects of the surround stimulus in the absence of CRF stimulation. As the tuning curves for the outer stimulus alone in Figs. 1 and 3 show, orientation contrast facilitation did not appear to be linked to direct excitatory effects driven by the outer stimulus (although subliminal facilitation could be a factor) and clear responses to the annulus alone were nearly always restricted to the cell's optimal orientation (e.g., Fig. 3). Across our sample of “orientation contrast facilitation cells” the mean firing rate evoked by an orthogonally oriented surround stimulus presented alone was low (1.5 ± 0.5 ips) and could not in any additive sense account for the variance in response to the inner alone and the combination firing levels discussed above. Indeed the response to the combination of stimuli was significantly higher than to the sum of the responses to the components (P < 0.001, Wilcoxon paired test) for each cell in the sample.

DISCUSSION

Our data taken with moving stimuli highlight a series of processes that contribute to the influence of orientation context on primate V1 cell responses. The most common pattern of influence (94% of our sample) was an orientation contrast dependent modulation in which cells gave larger responses when the surrounding stimulus orientation differed to that over the CRF. Given that the majority of primate V1 cells show strong surround suppression (Jones et al. 2001) this is consistent with an orientation contrast dependent modulation of the surround suppression (e.g., Gilbert and Wiesel 1990; Kastner et al. 1997; Orban et al. 1979; Sengpiel et al. 1997; Walker et al. 1999). The mechanism would follow from the fact that long range horizontal connections target cells of similar orientation preference and contact local inhibitory interneurons as well as spiny cells (Gilbert and Wiesel 1989; Kisvarday and Eysel 1992; Malach et al. 1993; Ts'o et al. 1986). However, the majority of cells in our sample (63%, 44/70) showed an enhancement of the response to levels above that to the inner stimulus alone as the orientation of the surround deviated away from the cells' optimal. Furthermore, roughly one-half the cells in this group (23/44) gave responses to orientation contrast stimuli that exceeded the cells' responses to an optimal, CRF-sized, inner stimulus alone. These effects were mainly (although not exclusively) seen for orientation contrast between a center and surrounding stimulus where the diameter of the central stimulus was the same size or greater than the CRF. For some cells they appeared to draw on the fact that the surround suppression was already implemented by a central stimulus that extended beyond the CRF and derived from a modulation of this (e.g., Fig. 2, A and B). For this reason we suggest that this effect might involve dis-inhibitory interactions, mediated via connections to the local inhibitory interneurons driving the suppression from other inhibitory interneurons linked to different orientations. This is plausible from the anatomical viewpoint as indicated by, for example, papers establishing the connectivity of GABAergic basket cells in V1 (e.g., Kisvárday and Eysel 1992; Kisvárday et al. 1993, 1994; alsoDas and Gilbert 1999).

In cases where the orientation contrast drove supra-optimal responses, it is likely that a surround suppressive mechanism was already implemented within the CRF, so that the response to an “optimal stimulus” was already in part diminished by an inhibitory mechanism that had the capacity to scale further as the stimulus size increased. Several lines of evidence invoke an inhibitory mechanism within the CRF driven by an optimal stimulus (Borg-Graham et al. 1998;Creutzfeldt and Ito 1968; Douglas et al. 1991; Ferster 1986; Sillito 1975,1977) and this sits in a number of models of visual cortical mechanisms (Dragoi and Sur 2000; Jones et al. 2001; Li 1999, 2000; Somers et al. 1998) and of the processes contributing to surround suppression (Sceniak et al. 1999). From this viewpoint, the scaling of the orientation contrast facilitation would build on the strength of the iso-orientation inhibition implemented in the CRF center. This does appear to vary in primate V1 cells (Jones et al. 2001). In some cells we observed facilitatory and supra-optimal facilitatory effects for interfaces between center and surround stimuli that sat within the CRF (e.g., Fig. 3). Again we would suggest that this reflects a strong inhibitory process implemented at the very center of the CRF and a dis-inhibitory mechanism driven by the orientation contrast.

Although the observation that orientation contrast effects can deliver response magnitudes that exceed those to the central stimulus alone seems compatible with a dis-inhibitory mechanism, the effect has not been previously reported. Some studies (e.g., Li and Li 1994; Nothdurft et al. 1999) include examples that might be taken to suggest the presence of the effect but the question remains why have others not seen this? One issue may follow from the fact that we used moving stimuli for all our tests in contrast to those used in studies drawing on data from behaving primates which have often used static patterns. Second, although some previous studies have used moving stimuli, many of these were carried out in cat (e.g.,Kastner et al. 1997; Sengpiel et al. 1997; Walker et al. 1999). If dis-inhibitory mechanisms, as we suggest above, largely underpin orientation contrast facilitation, then the proportion of such effects in cat would be expected to be much lower than for primate, since the incidence of surround suppression itself in the cat is much less (N. M. Oakley, P. C. Murphy, H. E. Jones, and A. M. Sillito, unpublished data). Third, a key feature distinguishing our work from that of others is that we explored a wide range of spatial interfaces between inner and outer stimuli. Only 7 of the 44 cells showing orientation contrast facilitation showed the effect at all interface locations with respect to the CRF size. Thus had we adopted the procedures used in many previous studies where only one interface diameter was used to explore orientation contrast effects we may have failed to detect the majority of examples reported in this study. This is doubly underlined by the fact that the single interface diameter selected in many previous studies never exceeded the size of the CRF (see for example, Sengpiel et al. 1997; Walker et al. 1999 in the cat), whereas we recorded the majority of orientation contrast facilitatory effects for interface diameters exceeding the CRF (as shown in Fig. 2).

Approximately 20% of the cells showing orientation contrast facilitatory effects showed orientation contrast suppression for interfaces within the CRF and orientation contrast facilitation for interfaces outside. Although cross-orientation inhibition within the CRF has been described and considered by many previous studies (Carandini et al. 1998; Creutzfeldt and Ito 1968; DeAngelis et al. 1992; Morrone et al. 1982; Pei et al. 1994; Sillito 1975), this orientation contrast suppression was much more sharply focused either side of the optimal (with the range of deviations for maximal suppressive effects spanning from 22.5–45°). The way these processes interact may depend on the space constants of the areas driving the two processes, but the balance seems to suggest that the force of the orientation contrast dis-inhibition/facilitation mechanism is likely to draw on focuses outside the CRF, while the orientation contrast suppression draws on a focus underlying the core of the CRF. It must be noted that for these cells as the interface diameter moved outside the CRF the orientation contrast suppression switched to iso-orientation “surround suppression” as well as orientation contrast facilitation. We also observed two cells showing only orientation contrast suppression as well as the two showing suppression driven by all orientations. These might reflect specialized interneurons within the circuitry mediating some of the effects discussed above.

Another issue is highlighted by the observations for nonorientation tuned cells. Here the points made about the influence of orientation contrast for the orientation tuned cells applied for any orientation. Hence the cells provided a potent signal for orientation contrast between their CRF and surrounding space but not for orientation per se. It suggests that mechanisms integrating the surround driven orientation contrast effects could be implemented across the network in a fashion that stands above the mechanism linked to orientation tuning. Normally it would appear linked to the orientation tuned CRF of cells because it is only revealed when they fire and that firing is orientation specific. Certainly we observed some orientation tuned cells (e.g., Fig. 7) that displayed a pattern of interplay between center and surround stimuli for interface diameters within the CRF that appeared very similar to those seen for the nonorientation tuned cells. It appears that the orientation contrast between inner and surround stimuli within the CRF enabled responses that were otherwise submerged. Clearly there is a great complexity and subtlety to the interactions influenced by orientation context. Obviously if orientation tuned receptive fields are built from the convergent input from nonorientation tuned cells in primate V1 then the presence of the behavior characterizing the nonorientation tuned fields for stimulus interface diameters within the CRF is hardly surprising.

The absolute firing levels of the cells under the different stimulus conditions underlines the view that a dis-inhibitory process might contribute to the different classes of effect. First, for the entire population of cells showing facilitatory effects (including the supra-optimal), the firing levels for the optimal stimulus and for the interface diameter driving the best orientation contrast facilitation were not significantly different (mean, 33 ± 4.52 and 34 ± 3.83 i/s), although both were significantly different to the mean for the inner stimulus. Thus we suggest that globally the orientation contrast serves to reset the lower firing level associated with a particular inner stimulus to that obtained by the best single stimulus. For the cells showing supra-optimal responses it was notable that overall they gave a significantly lower response to the best CRF sized single stimulus than the cells showing facilitation only, as well as a higher response to orientation contrast. The former being consistent with the view that they reflected a higher level of suppression elicited within the CRF. The larger response to the orientation contrast configuration in these cells at least raises the possibility that some additional mechanism to a dis-inhibitory process might further enhance their response. This could be orientation contrast facilitation. There are grounds for this type of interaction from both anatomical and physiological data (Das and Gilbert 1999;Kisvárday et al. 1997; Sillito 1975; Sillito et al. 1995), and it might be unmasked by the dis-inhibitory influence. Alternatively it could reflect a more potent dis-inhibition and further work is necessary to isolate this.

The location, relative to the CRF, of the border between the concentric center and surround stimuli was not the only spatial parameter examined in these experiments. We checked the effect of varying either the outer diameter of the outer stimulus, or the inner diameter, while holding that of the inner stimulus constant (Fig. 4). From this, it was clear that locations both close to and remote from the CRF could drive the orientation contrast facilitation with similar effect and that the effect did not seem to sum with the area of the cross-oriented outer stimulus. On the other hand, the strength of the surround suppression elicited under iso-orientation conditions did seem to increase with the area of the surround stimulus. We further dissected the spatial organization driving these effects by using a discrete second patch instead of an annulus to drive the orientation context effects. There were two main conclusions from this data, first, that there was an influence of patch size on the effect seen, and second, that while for some cells the zones driving orientation contrast facilitation were located uniformly around the field (14%) in the majority of cases they were heterogenous. Indeed for many (53%) the orientation contrast facilitation was driven from predominantly only one location. We found no evidence for a link between the effects and either side bands or end-zones although in 33% of the cells the zones driving effects were bilaterally symmetric. Interestingly, the magnitude of the facilitatory effects driven by the discrete patches of grating were not less than those drawn from the surrounding concentric stimuli (seeresults). This suggests that the orientation context per se, and its location, rather than the absolute extent of the stimulus providing the context was important. These data should not from this viewpoint thus be seen to necessarily reflect processes underlying orientation pop-out. Rather we suggest they may reflect a mechanism that integrates the components of moving contours that form junctions such as corners. The facilitatory effects of orientation contrast were seen for a wide range of angular deviations extending from <22.5° to 90° and would provide a broad filter highlighting foci where the orientation of a moving contour changed. Our use of moving stimuli make it difficult to directly compare our data with those studies showing facilitatory interactions when line elements or Gabor patches are added in a sequence along an axis parallel to a cell's optimal orientation (Ito and Gilbert 1999; Kapadia et al. 1995; Polat et al. 1998), although we do see facilitatory effects from optimally oriented surround stimuli that exclude the CRF center (Jones et al. 2001). The clear conclusions from our data are that orientation context exerts a potent effect on the responses of virtually all primate V1 cells and that the effect is strongly influenced by the spatial characteristics of the stimulus configuration driving the context. Finally, all the processes described whether for nonorientation tuned cells or orientation tuned cells seem to highlight foci where there is a change in orientation.

Acknowledgments

We are indebted to D. Matin and N. Burt for skilled technical assistance.

The support of the Medical Research Council is gratefully acknowledged.

Footnotes

  • Address for reprint requests: A. M. Sillito, Dept. of Visual Science, Institute of Ophthalmology, University College London, Bath Street, London EC1V 9EL, UK (E-mail: ams.admin{at}ucl.ac.uk).

REFERENCES

View Abstract