| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1Howard Hughes Medical Institute and 2Systems Neurobiology Laboratories, Salk Institute for Biological Studies, La Jolla, California; and 3Max-Planck-Institut für Biologische Kybernetik, Tübingen, Germany
Submitted 9 April 2004; accepted in final form 21 February 2005
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
). These responses are often modulated by the copresence of stimuli surrounding the CRF (Allman et al. 1985; Blakemore and Tobin 1972; Das and Gilbert 1999; DeAngelis et al. 1994; Kapadia et al. 1995; Knierim and van Essen 1992; Lamme 1995; Levitt and Lund 1997; Li and Li 1994; Maffei and Fiorentini 1976; Nelson and Frost 1978; Zipser et al. 1996). These modulatory effects parallel many well-known examples of the influence of context on visual perception (Albright and Stoner 2002).
One major focus of studies of visual context has been textural cues that distinguish different regions of an image. Differential orientation of texture elements (see Fig. 1) is a potent cue for perceptual segmentation (Bergen and Julesz 1983; Caputo 1996; Foster and Westland 1995; Li et al. 2000) and has a marked influence over orientation judgments (Mareschal et al. 2001; Sekuler 1965; Wehrhahn et al. 1996; Westheimer et al. 1976). Similarly, responses to oriented CRF stimuli have been found to be stronger in the presence of textured surrounds that promote perceptual segmentation (orthogonal lines) than in the presence of surrounds that do not promote segmentation (parallel lines) (Cavanaugh et al. 2002; Fries et al. 1977; Gilbert and Wiesel 1990; Knierim and van Essen 1992; Lamme 1995; Levitt and Lund 1997; Li et al. 2000; Nelson and Frost 1978; Zipser et al. 1996).
|
) and to improve orientation discrimination (Kooi et al. 1994; Wehrhahn et al. 1996). Although previous neurophysiological studies have investigated the effects of luminance contrast amplitude of textured surrounds on CRF responses in V1 (e.g., Hegde and Felleman 2003), polarity effects have received scant attention.
Another interesting but rarely studied issue is the extent to which contextual cues have interactive effects. This issue has considerable behavioral significance, because dissimilar surfaces usually differ in more than one aspect. One study representative of this genre found that chromatic differences distinguishing a visual motion signal had a marked positive influence on perceptual and neuronal responses to motion (Croner and Albright 1997, 1999). Luminance contrast polarity differences have a similar effect on perceptual sensitivity to motion (van der Smagt and van de Grind 1999). Simply stated, these studies reveal that the effects of one cue may depend on the state of another. We posited that such a rule might hold for the contextual effects of orientation and contrast polarity on the representation of orientation.
We tested our hypothesis by recording responses of V1 neurons in alert rhesus monkeys and by assessing orientation sensitivity of humans. In accordance with previous reports, we found that extra-CRF stimuli (masks) resulted in suppression of responses to a CRF stimulus (target), but that either differential orientation or polarity between target and mask reduced the magnitude of suppression. Similarly, masks impaired perceptual judgments of orientation, but the impairment was smaller if there was a difference between target and mask orientations or polarities. Combined orientation and polarity differences yielded effects that were no different from the effects of either cue alone. Despite the observed similarities between the averaged neuronal response rates for the different cue conditions, however, we found that the masking effects elicited by these different conditions occurred with different latencies. These findings refine our understanding of nonclassical receptive fields.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
SUBJECTS. Two adult rhesus monkeys (Macaca mulatta, one male and one female) were used in this study. Subjects had normal color vision and no significant refractive error. Experimental protocols were approved by the Salk Institute Animal Care and Use Committee, and conform to U.S. Department of Agriculture regulations and to the National Institute of Health guidelines for humane care and use of laboratory animals.
SURGICAL PREPARATION.
Procedures for surgery and wound maintenance have been described in detail elsewhere (Dobkins and Albright 1994). In short, a head post and a recording cylinder were affixed to the skull using stainless steel rails, screws, and dental acrylic. Chambers were positioned above area V1 (
11 mm lateral and 16 mm posterior to the midsagittal and interaural planes) to allow for a dorso-ventral electrode trajectory. A search coil for measuring eye position was surgically implanted in one eye using a variation of the method of Judge et al. (1980). After surgical recovery and attainment of criterion performance on the visual fixation task, a craniotomy was performed to allow for electrode passage into area V1. All surgical procedures were conducted under sterile conditions using isoflurane anesthesia.
VISUAL STIMULI. All visual stimuli were generated using a high-resolution graphics display controller (1,280 x 1,024 pixels, 8 bits/pixel) operating in a Pentium class computer. Stimuli were displayed on a 21-in analog RGB video monitor (Sony GDM-2000TC; 76 Hz, noninterlaced). A second computer managed the stimulus computer and behavioral data acquisition and also monitored eye position. Visual stimuli were viewed from a distance of 60 cm and were presented for 500 ms on each trial.
The stimuli used for our experiments were constructed of two basic components: target and mask. The target component consisted of a single line that was of the optimal orientation (assessed by means described below) and presented in the CRF of the neuron under study. Line width was 0.1° and length was adjusted such that the line remained within the CRF boundaries (but never exceeded 0.7°). The target was presented on a gray background (10.2 cd/m2) and was either brighter (86.2 cd/m2) or darker (1.26 cd/m2) than the background. Target luminances were chosen such that the absolute value of Michelson contrast between target and background was 78%, regardless of contrast polarity.
The mask component consisted of a 7 x 7 rectilinear grid of 0.7 x 0.1° lines, the exact positions of which were randomly assigned on each trial (such that individual lines never overlapped one another) about grid coordinates in a 6.5° square patch. All lines in a given mask were of the same orientation, which was either parallel or perpendicular to the orientation of the target stimulus. As for target lines, mask lines were presented on a gray background and, for a given mask, all were either brighter or darker than the background. Luminances and luminance contrasts (with respect to background) of the mask lines were drawn from the same values as those used for target lines. A circular region (2–3°) in the center of each mask contained no mask lines and was identical to the gray background. This region was centered on the CRF of the neuron under study to ensure that all mask lines remained well outside the CRF. This central region (partially) occluded some of the more central mask elements, thus changing the element density near the target compared with further away. However, it was superimposed on all mask types in the same manner to ensure that element density could not affect neuronal responses to the various target-mask combinations differentially.
Target and mask components were combined to yield a total of 14 stimulus conditions, 10 of which are shown in Fig. 1. Two stimulus conditions simply consisted of optimally oriented target lines of each contrast polarity (bright, dark) presented without a mask. Eight additional stimulus conditions resulted from the conjunctions of three main independent variables, each of which had two values: 1) contrast polarity of the target with respect to background (bright, dark), 2) contrast polarity of the target with respect to polarity of the mask (same, different), and 3) orientation of the target with respect to orientation of the mask (same, different). Thus for each target contrast polarity (columns in Fig. 1), the five principal conditions were 1) target-only (to): mask absent; 2) no-cue (nc): mask present, orientation and polarity same as target; 3) orientation-cue (oc): mask present, orientation different from target, polarity same; 4) polarity-cue (pc): mask present, polarity different from target, orientation same; and 5) double-cue (dc): mask present, orientation and polarity different from target.
Four conditions (not shown in Fig. 1) consisted of the four possible masks presented alone [bright/same-orientation (as CRF preference), dark/same-orientation, bright/different-orientation, dark/different-orientation]. The neuronal effects of these four mask-only conditions were used to screen data for further analyses.
Finally, for a subset of recordings, four additional control conditions were added to assess the effects of mask luminance independently of orientation. For these controls, the mask consisted of a uniform gray 6.5° square patch, which matched the average luminance of either the bright or the dark mask.
BEHAVIORAL TASK.
Monkeys were seated in a standard primate chair (Crist Instruments) with the head post rigidly supported by the chair frame. The behavioral task simply required subjects to fixate a small (0.12°) centrally located spot for the duration of each trial. Each trial began with the appearance of this fixation spot on the video display. After ocular fixation was achieved and held for 500 ms, the stimulus appeared for 500 ms, after which the fixation spot remained visible for another 100 ms. Trials in which eye position was maintained within 0.5° of the fixation spot were concluded with a small (0.15 ml) juice reward. A trial was aborted immediately if eye position deviated from the fixation window at any time, and a 1-s time-out ensued. Eye position was sampled at 500 Hz using the magnetic scleral search coil technique (Robinson 1963).
ELECTROPHYSIOLOGICAL RECORDING.
The activity of single units or small multi-units in V1 was recorded with tungsten microelectrodes (FHC, 3-M
base impedance), which were driven into cortex using a hydraulic micropositioner (David Kopf, model 650). Neurophysiological signals were amplified using standard equipment (Bak MDA-3 amplifier) and spikes detected via an on-line template matching algorithm (Alpha Omega). Digitized spike times (as well as eye position signals) were recorded on a computer using the CORTEX program (Laboratory of Neuropsychology, NIMH), which was also used to control stimulus presentation.
RECEPTIVE FIELD MAPPING AND INITIAL ASSESSMENT OF ORIENTATION SELECTIVITY.
Most cells studied were situated in the superficial layers of area V1 (revealed by electrode depth and pattern of activity encountered along each penetration) within the dorsolateral operculum of the left occipital lobe. All had CRFs located within the central 5° of the lower-right visual field [mean eccentricity: 2.9° ± 0.86° (SD)]. CRF locations and boundaries were determined using the minimum response field method (Barlow et al. 1967), which involved moving a bright or dark bar over a gray background while amplified neuronal signals were played on an audio monitor. For a subset of cells, basic properties of the CRF were subsequently assessed using an automated sequence of dark and bright flashes: Small dark and bright squares (0.10–0.22° square) were flashed for 100 ms at random locations in a square grid (1–2.0° centered on the hand-mapped coordinates) followed by a 100-ms pause. For each presentation of a square, the exact eye position (within the fixation window) was determined and the grid shifted, so that the grid center remained at the same coordinates with respect to the eye-position. As noted previously (Albright and Desimone 1987), the minimum response field and automated methods yielded very similar estimates of CRF boundaries. As an extra safeguard against intrusion of extra-CRF stimuli into the mapped CRF, four additional conditions containing only a mask were included. Neuronal responses to these mask-only conditions were used to determine eligibility for further analysis. After the CRF was mapped, orientation tuning was assessed using stimuli consisting of small lines (0.7 x 0.1°, bright or dark) oriented from 0 (vertical) to 160° in 20° steps.
ASSESSMENT OF MASKING EFFECTS. Once the CRF location, dimensions, and orientation selectivity were determined, we examined the effects of a variety of surround masking conditions on the neuronal response to an optimally oriented CRF target. With few exceptions (e.g., loss of unit isolation or loss of behavioral control), the responses of each neuron were recorded using a minimum of 10 presentations of each stimulus condition. The measure of neuronal response used for comparison of the effects of different experimental conditions was the mean spike rate computed within a window extending 40–540 ms after stimulus onset.
CRITERIA FOR INCLUSION IN DATA ANALYSES. In order for a neuron to be included in the masking analyses, the data had to satisfy three requirements. First the cell had to be tuned significantly around the orientation that was used for the masking experiments. A Rayleigh test of randomness was thus applied to all orientation tuning data. Only those cells reaching the criterion of P < 0.05 were included in the next analysis. The second requirement was that mask-only conditions yield no significant change in neuronal activity. More specifically, we included only those cells for which mask-only conditions did not yield a response (assessed during the period 40–100 ms after stimulus onset) that was significantly different from spontaneous activity (assessed –20 to +40 ms relative to stimulus onset; Wilcoxon signed rank test; P > 0.05). This 40-ms cut-off was chosen because that is the average response latency of V1 neurons in our sample when presented with a CRF stimulus. The third requirement was simply that data be obtained for a minimum of six presentations of each relevant stimulus condition.
MASKING INDICES. We computed a series of indices to quantify the effects of surround stimuli (mask) on suppression of responses to CRF stimuli (target). These indices were computed separately for bright target and dark target conditions. Each index consisted of a contrast measure computed between responses to specific stimulus conditions. The terms used to compute these indices were thus the response rates for the five conditions in each column of Fig. 1: Rto = neuronal response elicited by target-only condition; Rnc = neuronal response elicited by no-cue condition; Roc = neuronal response elicited by orientation-cue condition; Rc = neuronal response elicited by polarity-cue condition; and Rdc = neuronal response elicited by double-cue (orientation + polarity) condition.
GENERAL MASKING INDEX. The general masking index (GMI) reflects the degree to which masks suppress the neuronal response to the target, independent of the type of mask used. The index was computed as the contrast between the target-only response (Rto) and the average of the responses obtained under the four mask conditions (Rnc, Roc, Rpc, and Rdc): GMI = [Rto –(Rnc + Roc + Rpc + Rdc)/4]/[Rto + (Rnc + Roc + Rpc + Rdc)/4].
SPECIFIC MASKING INDICES. For each neuron we computed four specific masking indices, each of which reflects the contrast between the target-only response and the response elicited by one of the four masking conditions: No-Cue Masking Index (MInc) = (Rto – Rnc)/(Rto + Rnc); Orientation-Cue Masking Index (MIoc) = (Rto –Roc)/(Rto + Roc); Polarity-Cue Masking Index (MIpc) = (Rto –Rpc)/(Rto + Rpc); and Double-Cue Masking Index (MIdc) = (Rto –Rdc)/(Rto + Rdc).
We used these specific masking indices to test hypotheses about the effects of segmentation cues on relief from masking suppression. Specifically, we statistically compared (ANOVA with multiple posthoc comparisons) the values of all pairs of indices across the neuronal population. Thus a comparison of MInc versus MIoc revealed the degree to which the orientation cue yielded relief from masking suppression caused by a no-cue mask. Similarly, the MInc versus MIpc comparison revealed the degree to which the contrast polarity cue yielded relief from masking suppression. The MIoc versus MIpc comparison revealed the degree to which masking relief afforded by orientation differed from that afforded by contrast polarity. The MInc versus MIdc comparison revealed the degree to which the double-cue mask yielded masking relief. Finally, the MIoc versus MIdc and MIpc versus MIdc comparisons enabled us to assess the relative effects of single- versus double-cue masks on relief from suppression.
ANALYSIS OF RESPONSE TIMING IN AVERAGED PERISTIMULUS HISTOGRAMS. We evaluated the timing of certain key events in the neuronal responses to different stimulus conditions in an effort to identify the sources of inputs to the recorded neurons. In doing so, we endeavored to determine the times at which PSTHs (averaged across neurons) differed from one another and from baseline. To establish that curves were significantly different, we binned (10-ms bin width) the spikes in each averaged PSTH and performed paired t-test on the number of spikes in each bin for all desired PSTH comparisons. The times of occurrence of differences between curves (P < 0.01) were thus established at 10-ms resolution. However, to facilitate viewing of persistent differences between averaged PSTHs, which would otherwise be hampered by visual noise, we plotted the PSTHs at 20-ms resolution and used a Gaussian kernel (20-ms SD) to smooth the curves.
We stress that a rigorous quantitative analysis of event timing was only applied to PSTHs averaged across the neuronal population. Low firing rates, and consequent lack of statistical power, prevented use of these methods for PSTHs from individual neurons.
Human psychophysical experiments
SUBJECTS. Three subjects (1 female and 2 males) participated in these experiments. Two of the subjects were naïve as to the purpose of the experiment. All subjects had normal or corrected-to-normal acuity.
VISUAL STIMULI.
Stimuli were generated using a standard VGA graphics display controller (GeForce 2 MX; 640 x 480 pixels, 8 bits/pixel) operating in a Pentium class computer, and using software developed by Dr. G. Westheimer, (University of California Berkeley). Stimuli were displayed on a 21-in analog RGB monitor (Iiyama Vision Master 500; 76 Hz, noninterlaced). The voltage/luminance relationship was linearized independently for each of the three guns in the display (Watson et al. 1986). Stimuli were viewed from a distance of 6 m in a dark room (<0.5 cd/m2).
As was the case for our neurophysiological experiments, the stimuli used for human psychophysics were constructed of two basic components: target and mask. These components were themselves similar to those used for neurophysiology, but had the following exact specifications. Background luminance of the monitor was 10.4 cd/m2. Target and mask lines were either brighter (86 cd/m2) or darker (1.25 cd/m2) than the background, yielding Michelson contrasts of 78.4 and –78.5%, respectively. Target line length was 0.8°, which is 0.1° longer than the longest target lines used by us for single V1 neurons and approximately the size of multi-unit or aggregate single-unit RFs at the target location in monkey V1 (e.g., Hubel and Wiesel 1974; Van Essen et al. 1984). The mask was 5° square and composed of 7 x 7 masking elements, each 0.7° long. These masking elements were either vertical or horizontal and the position of each was randomly jittered about a mean grid coordinate (avoiding overlap). A 1 x 2° rectangular window was placed in the center to exclude the masking pattern in the immediate vicinity of the target. Stimulus presentation time was 80 ms.
Target and mask components were combined to yield a total of 10 stimulus conditions, which are analogous to the 10 conditions used for neurophysiology that are shown in Fig. 1 (the primary difference being that masking lines for psychophysics were always either horizontal or vertical). The four mask-only conditions used for neurophysiology were omitted from the psychophysical experiments.
BEHAVIORAL TASK.
Subjects were required to judge small differences in the orientation of a target presented at 2.9° eccentricity in the presence or absence of a variety of masking stimuli (see Fig. 1), while fixating a small central fixation spot. Subjects initiated each behavioral trial by a button press. After a 1-s delay, the stimulus appeared for 80 ms. Subjects viewed stimuli with both eyes and were required to report (via button press) whether the target line appeared to be tilted clockwise or counterclockwise with respect to vertical. Subjects were well able to perform such a task with vertical serving as an implicit reference (see Li et al. 2000), which prevented the need for additional explicit reference stimuli that might have influenced the percept of the target. No feedback was given. The method of constant stimuli was used to determine orientation discrimination thresholds. Seven target orientations were employed, which included true vertical and three orientations each to the left and right of vertical (spanning a range adjusted as required for each subject to obtain threshold measurements). Conditions were presented in a randomly interleaved fashion. Data presented herein were collected following a training phase that was continued until psychophysical thresholds became stable. Each data point is based on >270 responses collected over at least 2 days.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
We recorded from a total of 239 V1 neurons in two monkeys. Most of these cells were presumed to be of the complex type because they were recorded from the superficial layers in V1 (Alonso 2002). Moreover, the CRFs that were mapped by the automated procedure did not show any sign of disparate ON and OFF regions. One hundred fifty of these neurons (28 multi-units and 122 single units) met our criteria for full analysis. A total of 124 cells were tested with a bright target; 69 were tested with a dark target. Forty-three cells were tested with both target polarities. (The initial choice of testing with a bright or dark target was quasi-random. As it turns out, the number of cells that met all of our criteria for inclusion in the data analysis differed for bright and dark targets.)
Representative neurons: forms of response suppression and stimulus dependence
Our stimulus conditions enabled us to examine directly 1) the extent to which placement of an oriented texture mask in the CRF surround influences the response to a CRF target, 2) the dependence of any response masking on mask orientation relative to target orientation, 3) the dependence of any response masking on mask polarity relative to target polarity, and 4) the existence of any additive or interactive effects of relative orientation and polarity.
Bright target conditions
The cell displayed in Fig. 2, which was tested with bright target conditions (Fig. 1), provides a representative portrait of the effects of the above manipulations. Figure 2A depicts the neuronal response as a function of target orientation for the target-only condition, which reveals that the cell was narrowly tuned with a preference for lines oriented 20° clockwise from vertical.
|
; Li et al. 2000). Third, turning the tables, consider the effects of polarity for a surround mask that was of the same orientation as the target. A mask with the same polarity as the target yielded suppression (Rnc bar in Fig. 2B) that was significantly greater than that resulting from a mask with a different polarity (Rpc bar; P < 0.05). This novel finding constitutes a contrast polarity analog of the established orientation-masking phenomenon.
Finally, our experiment allowed us to examine the conjoint effects of orientation and contrast polarity cues. Figure 2B reveals that the degree of response suppression for a double-cue mask (Rdc bar) was no different from the effects of either cue alone (Roc and Rpc bars; P > 0.1 for each comparison). Thus, although masking responses remained suppressed relative to that elicited by the target alone (Rto bar), we saw no evidence for an additional suppression relief that would be consistent with additive effects of the two cues. On the contrary, the pattern of suppressive effects seen in Fig. 2B bears the signature of a highly interactive system: an orientation difference resulted in reduced mask suppression when mask and target were of the same luminance contrast polarity (Rnc and Roc bars), but not when mask and target were different polarities (Rpc and Rdc bars). Similarly, a polarity difference resulted in reduced mask suppression when mask and target were of the same orientation (Rnc and Rpc bars), but not when mask and target were different orientations (Roc and Rdc bars).
Figure 2C shows the PSTH for the target-only condition along with PSTHs for the two same-polarity masks. As expected, the cell responded strongly to the onset and offset of the target in the absence of a mask (Rto). The magnitude of the sustained component of the response to a target was significantly smaller when a same-polarity mask was present (Roc), and even more so when the mask and target were of the same orientation (Rnc). Figure 2D shows the PSTH for the target-only condition with those for the two different-polarity masks. The magnitude of the sustained component of the response to a target was smaller when a different-polarity mask was present. However, the responses to same- (Rpc) and different-orientation (Rdc) masks were virtually indistinguishable throughout the entire epoch in which data were recorded.
Dark target conditions
The cell displayed in Fig. 3, which was tested with dark target conditions, exhibited behavior similar to that seen under bright target conditions (Fig. 2) and thus makes a case for generality across target contrast polarity. Figure 3A depicts the neuronal response as a function of target orientation for the target-only condition, which reveals that the cell was broadly tuned with a preference for lines oriented 80° counterclockwise from vertical. The data presented in Fig. 3B reveal that suppression by a surround mask [repeated measures ANOVA (F = 3.81, P < 0.01)] was strongest when its line elements were of the same contrast polarity and orientation as the target (Rnc bar); the same-polarity mask elicited weaker suppression when target and mask were oriented differently (Roc bar; P < 0.01). Similarly, the suppression caused by a same-orientation mask was strongest when target and mask were of the same contrast polarity (Rnc bar) and weakest when target and mask were of different polarities (Rpc bar; P < 0.01). As was true for the bright target conditions, double-cue masks (Rdc bar) yielded a degree of suppression that was indistinguishable from that caused by either single-cue mask (Roc and Rpc bars; P > 0.5).
|
Population data
The two neurons highlighted above exhibited suppression of the response to a target in the CRF when a mask was presented outside the CRF. The magnitude of this suppression was dependent on the relative orientation of target and mask as well as the relative contrast polarity of target and mask, such that suppression was relieved when target and mask differed along either dimension. The effects of the two cues were roughly equivalent, but the combination of cues yielded effects that were no different from either cue alone. We believe this pattern of effects to be representative of the neuronal population studied, and we have computed a number of measures to quantify these effects more broadly (see METHODS).
Bright target conditions
POPULATION AVERAGES. The average relative influences of orientation and contrast polarity cues are documented in Fig. 4A, which shows firing rates averaged across all neurons (n = 124) tested with the five indicated bright target conditions. For each neuron, firing rate for the target-only condition (Rto) was arbitrarily set to 100. Responses obtained under the four indicated mask conditions (Rnc, Roc, Rpc, Rdc) were normalized to the target-only response. These population averages appear remarkably similar to data from the representative neuron shown in Fig. 2B: responses to the target were significantly suppressed (relative to target-only condition) by the presence of a simultaneously presented mask of any type [repeated measures ANOVA (F = 25.03, P < 0.0001)] with LSD posthoc comparisons, P < 0.0001]. On average, however, mask suppression recovered significantly relative to that seen for the no-cue condition if target and mask lines differed either along the dimension of orientation (P < 0.0001) or contrast polarity (P < 0.05). The suppression caused by a double-cue mask was no greater than that caused by either cue alone (P > 0.4).
|
SPECIFIC MASKING INDICES REVEAL RECOVERY FROM RESPONSE SUPPRESSION. We computed four specific masking indices for each neuron (MInc, MIoc, MIpc, MIdc), which capture the observed differences between responses elicited by the target-only condition (Rto) and those elicited by each of the four masking conditions (Rnc, Roc, Rpc, Rdc). These indices, which were computed as contrast measures (see METHODS), were compared with one another to evaluate the relative degrees of suppression caused by different mask types on a cell-by-cell basis. The means of the distributions for each index are indicated in Table 1A, along with the results of tests of significance (repeated measures ANOVA, LSD posthoc comparisons) for each paired comparison of distributions (6 pairs total).
|
; Li et al. 2000). Similarly consistent with the cell exhibited in Fig. 2, the mean of the distribution of MIpc is significantly smaller than that for MInc. This difference supports, at the population level, our contention that the effects of cue differences on recovery from masking suppression generalize to the domain of contrast polarity. The distributions of MIoc and MIpc are indistinguishable, which confirms that the two cues lead to similar degrees of recovery from masking suppression for individual V1 neurons. As expected from the effects of solitary cue differences, the mean of the distribution of MIdc is significantly smaller than that for MInc. A more telling evaluation of the effects of the double-cue mask comes from comparisons between MIdc and each of the two single-cue indices. As anticipated from the example neuron shown above (Fig. 2A), both comparisons failed to reveal a difference between the effects of single- and double-cue masks at the population level. Thus despite the fact that recovery from masking suppression was rarely complete under single-cue mask conditions, there was generally no additional relief afforded by the simultaneous presence of two segmentation cues. Dark target conditions
Population averages and response indices obtained for neurons tested with a dark target (n = 69) are shown in Fig. 5, following the same format used in Fig. 4 for bright-target population data.
|
The GMI for dark target conditions is plotted in Fig. 5B. The plot reveals that the GMI was positive for the vast majority (84%, 58 of 69) of cells studied (mean = 0.14; t = 7.40, P < 0.0001). The GMI for the neuron highlighted in Fig. 3 is indicated by an arrow along the abscissa.
The means of the specific masking index distributions are indicated in Table 1B, along with the results of tests of significance for each paired comparison of distributions (6 pairs total). As for bright target conditions, the means of the distributions of orientation-cue, polarity-cue, and double-cue indices are significantly smaller than that for the no-cue index. The orientation- and polarity-cue indices do not differ from one another, nor do they differ from the double-cue index.
Bright targets versus dark targets: within-neuron comparisons
The analyses described above reveal that many of the response characteristics and suppression effects seen for the population of cells tested with a dark target were similar, on average, to those for cells tested with a bright target. Examination of the subset of cells (n = 43) that was tested with both target polarities revealed that this pattern of effects was nearly identical on a cell-by-cell basis. Specifically, when bright and dark target responses for each cell were normalized to those elicited by the respective target-only conditions, the four masking condition responses (Rnc, Roc, Rpc, Rdc) were not significantly different, on average, for bright and dark targets (paired t-test; P > 0.1, T < 1.4, for each of the 4 comparisons).
Is response suppression influenced by mask luminance?
The suppression effects showed herein and in previous studies (e.g., Knierim and van Essen 1992; Li et al. 2000) were elicited by the presence of textured patterns that either raised or lowered the space-averaged luminance in the non-CRF. These luminance changes predictably altered the luminance contrast between the CRF and non-CRF, and we were compelled to consider the possibility that such contrast differences contributed to the observed response modulations. We believe this possibility unlikely, a priori, simply because the magnitude of suppression—when it was observed—was highly dependent on mask orientation. Furthermore, Sqautrito et al. (1990) sought and failed to find any effect of uniform surround luminance changes on responsivity and orientation tuning in monkey V1. Finally, if there were a contribution to suppression that stemmed from luminance contrast, it would arguably be irrelevant to many of our response comparisons, which largely focus on the relative degrees of suppression elicited by masks that differ from one another but have identical space-averaged luminance.
In any event, to fully evaluate a possible role for luminance differences between CRFs and non-CRFs, we included two luminance control conditions. In these conditions the spatial region normally occupied by the mask was of a uniform luminance that was elevated or lowered to equal the space-averaged luminance of either the bright or dark masks (i.e., 16 or 9.5 cd/m2). Forty-four V1 neurons were tested under these conditions as well as using the standard stimulus set (Fig. 1). The mean GMI for this subpopulation of neurons did not differ from that for the larger population (t-test, P > 0.1, T < 1.25), yet—consistent with the findings of Squatrito et al. (1990)—responses to target stimuli were not suppressed by either of the uniform luminance changes applied to the non-CRF (paired t-test, P > 0.1, T < 1.4 for either comparison).
Timing of neuronal response modulation
Differences in the temporal dynamics of responses elicited by CRF stimuli under different contextual conditions have been used previously to draw inferences about the sources of modulatory signals and the underlying mechanisms (e.g., Bair et al. 2003; Knierim and van Essen 1992; Lamme 1995; Zipser et al. 1996). To enable such inferences here, we performed a temporal analysis of spike rates for different stimulus conditions averaged across cells. To facilitate measurement of the timing of average spike rate modulations, we restricted this analysis to those cells that exhibited significant context-induced response suppression [i.e., the response to the target-only condition (Rto) was significantly greater than the response to the no-cue mask condition (Rnc)]. PSTHs were normalized to the mean target-only spike rate for each cell and averaged across the population. Data obtained using bright targets (n = 45 cells with Rto > Rnc) are plotted in Fig. 6 (data from dark target conditions are similar).
|
40 ms after stimulus onset, regardless of the presence, absence, or type of mask (see Fig. 6B). The target-only response (Rto) rose rapidly to a transient peak occurring 70 ms after stimulus onset, and declined very gradually over the course of stimulus presentation. The responses to all masking conditions (Rnc, Roc, Rpc, Rdc) exhibited the same initial transient, but declined rapidly to 50% of the initial peak. The single- (Roc, Rpc) and double-cue (Rdc) response curves stabilized at a level that was intermediate between the target-only and no-cue responses (see Fig. 6A; as expected from mean response rates). [Note that for this analysis the temporal resolution was 10-ms per bin (see METHODS) and the timing information provided is conditional on this level of approximation. In Fig. 6, the data are plotted in 20-ms bins (and smoothed in Fig. 6A) for clarity.]
There was, however, a revealing pattern by which responses to the different masking conditions diverged. The no-cue response (Rnc) began to diverge significantly (see METHODS for statistical criterion) from the target-only response (Rto) immediately prior to the transient peak (i.e., approximately 60 ms following stimulus onset). Interestingly, at this point of divergence, the polarity-cue (Rpc) and double-cue (Rdc) responses paired-off with the target-only response, whereas the orientation-cue response (Roc) paired-off with the no-cue response (Rnc). The next significant event was separation of polarity-cue (Rpc) and double-cue (Rdc) responses from the target-only response, which occurred roughly coincident with the transient peak (80 ms following stimulus onset). Thus, at this early stage, the orientation-cue response remained indiscriminable from the no-cue response, although those masks containing a polarity cue (single or double-cue) already yielded a differential effect. Finally, at approximately 140 ms following stimulus onset, the response to the orientation-cue condition (Roc) rose to a level above that for the no-cue condition. The magnitude of all responses declined gradually over the course of stimulus presentation but the ordinal position of the responses (target-only largest, single- and double-cue intermediate, no-cue smallest) remained unchanged (Fig. 6A).
Human psychophysical experiments
We conducted a psychophysical experiment with human subjects to determine whether stimulus conditions resulting in suppression of neuronal responses to targets would lead to corresponding changes in the perception of oriented stimuli. The task chosen for this purpose required subjects to make fine judgments about the orientation of a target line (see METHODS). These orientation judgments are presumed to depend on a high fidelity neuronal representation of target orientation (Mareschal et al. 2001; Wehrhahn et al. 1996). Stimuli were centered at the mean RF eccentricity for our sample of V1 neurons (2.9°).
Figure 7A contains orientation discrimination data obtained using the same set of five bright target conditions employed for our neurophysiological experiments (see Fig. 1). Sensitivity (1/angular threshold for detection of an orientation difference) to the target-only condition (Fig. 7A, 1st bar) provided a reference point for comparison of the effects of different mask conditions. The only condition that yielded any significant decrease in sensitivity was that in which the mask was of the same orientation and contrast polarity as the target, i.e., the no-cue mask condition (2nd bar). Despite the continued presence of a mask, we found that sensitivity was completely restored if the mask and target lines were of different orientations (3rd bar), which corroborates previous reports (Wehrhahn et al. 1996). We also found that sensitivity was completely restored when mask and target lines were of different contrast polarities (4th bar). The effects of the two cues on orientation discrimination were not cumulative, however; the recovery from suppression that was associated with a double-cue mask (5th bar) was no different from that associated with either cue alone (3rd and 4th bars). The effects of single- and double-cue masks on orientation discrimination for dark targets (Fig. 7B) were nearly identical to those seen for bright targets.
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Feature salience, discriminability, and neuronal response modulation
Image features that are physically distinct from their surroundings are more easily detected than those that are not (e.g., Beck 1972; Bergen and Julesz 1983; Treisman and Gelade 1980). Abundant evidence also exists to conclude that specific attributes of image features distinguished in this manner are also more easily discriminable than are attributes of features that blend into their surroundings (e.g., Caputo 1996; Mareschal et al. 2001; Wehrhahn et al. 1996). The psychophysical experiments in this study address the degree to which the perception of orientation is influenced by distinguishing contextual cues. Consistent with previous reports (Wehrhahn et al. 1996; Westheimer et al. 1976), we found line orientation judgments to be impaired when the target line was imbedded in a surround composed of identical lines. This interference disappeared when the target line was distinguished from surrounding features, either by orientation or luminance contrast polarity differences.
A neural correlate of masking was initially reported for area V1 in primates by Knierim and Van Essen (1992), who found that sensitivity to an oriented CRF stimulus was dependent on the relationship between the stimulus and the surround—an indistinct CRF stimulus, as defined by the lack of an orientation difference between the stimulus and surrounding features, yielded response suppression relative to a target-only condition. A distinct CRF stimulus, in contrast, yielded relief from suppression. Numerous subsequent studies have confirmed these effects (e.g., Cavanaugh et al. 2002; Lamme 1995; Levitt and Lund 1997; Li et al. 2000; Zipser et al. 1996). These results reveal that similar effects occur for luminance contrast polarity.
This class of contextual influences from the non-CRF can be viewed as scaled-up multi-attribute versions of the contrast detection properties seen in center-surround CRFs and cortical simple cells (Albright 1994; Cavanaugh et al. 2002; Mareschal et al. 2001; Sceniak et al. 2001). Unlike CRF contrast detection, however, the non-CRF effects are strictly modulatory: surrounding features in the non-CRF do not themselves elicit a change in neuronal spike rate. Rather, the gain of response to CRF stimulation is determined by the relationship between the CRF stimulus and surrounding features.
Cue interaction: sensitivity to one cue is modulated by the presence of another
In this study we assessed neuronal and perceptual sensitivity to an oriented target stimulus as we varied two dimensions of nontarget features (orientation and contrast polarity). Remarkably, we found that a polarity cue improved sensitivity to the target when an orientation cue was absent, but not when it was present, and vice versa. Similar cue interactions were observed by Croner and Albright (1999) and van der Smagt and van de Grind (1999), who found that color and contrast polarity cues for segmentation facilitated motion processing when motion signals were weak, but had little impact when they were strong. Findings of this nature reinforce the view that multivariate contextual interactions are an essential element of visual processing (see Albright and Stoner 2002 for review), they suggest that notions of selective information channeling in visual cortex (e.g., Livingstone and Hubel 1987) are inadequate, and they stress the value of experimental approaches that include such contextual manipulations.
Form-cue invariant contextual effects?
There are a number of object qualities, such as size, location, form, and motion, that possess no inherent relationships with the diverse cue differences that coincide with an object's boundaries in space. It follows that visual system functions reliant on such object-based qualities must be sensitive to multiple cues for object definition, but they must also disregard cue diversity (Stoner and Albright 1993). In practice, the characteristic signature of a "form-cue invariant" (Albright 1992) process is a logical OR operation, in which cues elicit equivalent noncumulative effects. The masking effects seen in these present experiments qualify as form-cue invariant, at least for the range of cues tested. [Note that this does not necessarily apply to all cues. van der Smagt et al. (2002) have found linear summation for color and orientation in a different, but similar, experiment.] These conclusions regarding form-cue invariant contextual effects must be qualified by noting that, although the different cues used have similar effects on mean response rates, there are clearly differential effects on response dynamics—at least for orientation and polarity cues (Fig. 6). We have used these temporal effects to guide our thoughts on mechanisms (see below); it remains to be seen whether they have any functional significance.
Comparison of neuronal and perceptual effects of context
Our experiments afforded an unprecedented opportunity to examine the relative effects of different masking conditions on the neuronal and perceptual representations of orientation. We assessed these representations by different means, however. It is worth considering the extent to which these differences bear on our conclusions.
Our neurophysiological and psychophysical measurements differed in that the former consisted of response magnitudes and the latter were angular discrimination thresholds. Although we do not yet understand the precise causal relationship between response magnitude and discriminability, it is certainly plausible that orientation-dependent responses become less discriminable when overall response rates decline—by either altering the slope of the shoulders of the orientation-tuning curve or reducing the signal-to-noise ratio. In any event, the remarkable qualitative similarity between neuronal and perceptual patterns of masking and masking relief argues for a causal link.
There are, however, quantitative differences between these effects: 1) the inhibition caused by a mask without additional cues appears almost twice as strong in human observers compared with the individual V1 neurons, and 2) the masking relief brought about by the presence of an orientation and/or contrast polarity cue is only partial (relative to target-only response) for individual V1 neurons, whereas it is complete for human observers. At present, there are no certain conclusions one can draw from these quantitative differences. However, we offer the following speculations.
To begin with, recall that the neuronal measure is one of response magnitude, whereas the perceptual measure reflects discriminability. To evaluate the relationship between the two measures, we began by normalizing each relative to the effects of the simplest stimulus used, i.e., the target-only condition (Fig. 7C, leftmost pair of bars), and then causally linking the two measures at that reference point. Rigorous interpretations of neuronal versus perceptual response deviations from this reference point are inherently problematic because we have no independent means to verify the scaling relationship between the two measures. Thus one can only assume for now that the observed 74% of target-only neuronal response elicited by the no-cue mask is what naturally gives rise to the observed 60% of target-only perceptual response to the same stimulus. Given this mapping, the complete relief from masking that we have observed perceptually in the presence of orientation and/or polarity cues may be accounted for by supposing that neuronal activity is pooled to minimize the influence of neuronal noise (Croner and Albright 1999; Shadlen et al. 1996). According to this hypothesis, the partial masking recovery seen for individual neurons yields complete recovery at the neuronal population level and thus complete perceptual relief from masking. Our hypothesis also supposes a perceptual response "ceiling," which corresponds to the target-only level of performance, such that any greater neuronal activity does not lead to a greater perceptual response. The concept of such a ceiling is supported by the additional paradoxical observation that the target-only neuronal response gives rise to the same level of perceptual performance as does the much smaller neuronal responses elicited by single-cue or double-cue mask conditions Fig. 7C.
It may also be the case that differential allocation of attention contributes to the aforesaid quantitative differences between perceptual and neuronal data. Specifically, in the psychophysical experiments, attention was most likely directed at the target, whereas in the neuronal experiments, it was most likely directed at the fixation spot. This seems an unlikely explanation for the larger inhibitory effect of the no-cue condition observed psychophysically (relative to the neuronal effect size), however, as allocation of attention generally improves psychophysical performance instead of impairing it (e.g., Posner 1980). Moreover, the effect of differential allocation of attention on neuronal responses in area V1 is still a subject of much debate (e.g., Moran and Desimone 1985; Roelfsema et al. 2004). In any event, it is worth noting that attention was (roughly speaking) a constant within each set of experiments. Our goal in this study was to compare results across the different stimulus conditions within each set of experiments (i.e., across stimulus conditions for which attention was unchanging). The interesting outcome was that the results of these inter-condition comparisons were relatively similar for the neuronal and psychophysical experiments, despite the fact that attention was invoked differently.
Mechanisms of masking and segmentation: where do contextual signals originate?
Contextual influences of the sort seen in this study clearly require that visual information be integrated across space in a feature-specific manner. The relevant signals are believed to arise either via feedback connections from more central visual areas or via long-range horizontal connections within a given visual area (Allman et al. 1985; Lamme et al. 1998; Stettler et al. 2002). The intra-areal hypothesis is supported in part by evidence indicating that the long-range horizontal connections in area V1 tend to connect cortical domains of similar orientation and convey orientation-specific signals (Bosking et al. 1997; Gilbert and Wiesel 1990; Kisvarday et al. 1997; Malach et al. 1993; Stettler et al. 2002), and their influence can be inhibitory (McGuire et al. 1991). In addition, it is well known that luminance increments and decrements are processed by separate channels that originate with ON-center and OFF-center retinal ganglion cells (Famiglietti and Kolb 1976; Schiller 1992; Wiesel and Hubel 1966) and remain largely segregated up to area V1 (Horton and Sherk 1984; Hubel and Livingstone 1990; Schiller 1984; Schiller et al. 1986). This early segregation of contrast polarity signals is thus also compatible with a contextual signal that is intrinsic to area V1.
In previous studies, the temporal dynamics of non-CRF suppression and relief from suppression have revealed consistent patterns, which have also been used as evidence for the origin of contextual signals. Our data are similar, for example, to those of Knierim and Van Essen (1992) as well as Li et al. (2000), who found that target-only responses diverged on average from no-cue responses by the time of the initial response peak. Interestingly, we found that the timing of response suppression and recovery differed depending on whether the contextual cue was orientation or contrast polarity (see Fig. 6). We believe that this pattern can be accounted for conceptually by a model based on intrinsic horizontal connections in area V1 that manifest both same-orientation inhibition and different-polarity excitation. In principle, both are contextual signals that will elevate CRF responsivity in the presence of center-surround contrast, but the differential sources of these hypothesized signals allows for differential timing and magnitudes of their respective effects on CRF responses. Alternatively, the surround modulatory signals may reflect feedback projections from higher cortical areas (Bair et al. 2003), such as V2, which convey orientation and polarity-specific information.
|
ACKNOWLEDGMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Present address of M. J. van der Smagt: Psychonomics Dept., Helmholtz Instituut, Universiteit Utrecht, Heidelberglaan 2, 3584 CS Utrecht, The Netherlands. (E-mail: M.J.vanderSmagt{at}fss.uu.nl).
|
FOOTNOTES |
|---|
Address for reprint requests and other correspondence: T. D. Albright, The Salk Institute/HHMI, 10010 N. Torrey Pines Rd., La Jolla CA 92037 (E-mail: tom{at}salk.edu)
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONACKNOWLEDGMENTSREFERENCES |
|---|
Albright TD. Why do things look as they do? Trends Neurosci 17: 175–177, 1994.[CrossRef][Web of Science][Medline]
Albright TD and Desimone R. Local precision of visuotopic organization in the middle temporal area (MT) of the macaque. Exp Brain Res 65: 582–592, 1987.[Web of Science][Medline]
Albright TD and Stoner GR. Contextual influences on visual processing. Annu Rev Neurosci 25: 339–379, 2002.[CrossRef][Web of Science][Medline]
Allman J, Miezin F, and McGuinness E. Stimulus specific responses from beyond the classical receptive field: neurophysiological mechanisms for local-global comparisons in visual neurons. Annu Rev Neurosci 8: 407–430, 1985.[CrossRef][Web of Science][Medline]
Alonso JM. Neural connections and receptive field properties in the primary visual cortex. Neuroscientist 8: 443–456, 2002.
Bair W, Cavanaugh JR, and Movshon JA. Time course and time-distance relationships for surround suppression in macaque V1 neurons. J Neurosci 23: 7690–7701, 2003.
Barlow HB, Blakemore C, and Pettigrew JD. The neural mechanism of binocular depth discrimination. J Physiol 193: 327–342, 1967.
Beck J. Similarity grouping and peripheral discriminability under uncertainty. Am J Psychol 85: 1–19, 1972.[CrossRef][Web of Science][Medline]
Bergen JR and Julesz B. Parallel versus serial processing in rapid pattern discrimination. Nature 303: 696–698, 1983.[CrossRef][Medline]
Blakemore C and Tobin EA. Lateral inhibition between orientation detectors in the cat's visual cortex. Exp Brain Res 15: 439–440, 1972.[Web of Science][Medline]
Bosking WH, Zhang Y, Schofield B, and Fitzpatrick D. Orientation selectivity and the arrangement of horizontal connections in tree shrew striate cortex. J Neurosci 17: 2112–2127, 1997.
Caputo G. The role of the background: texture segregation and figure-ground segmentation. Vision Res 36: 2815–2826, 1996.[CrossRef][Web of Science][Medline]
Cavanaugh JR, Bair W, and Movshon JA. Nature and interaction of signals from the receptive field center and surround in macaque V1 neurons. J Neurophysiol 88: 2530–2546, 2002.
Croner LJ and Albright TD. Image segmentation enhances discrimination of motion in visual noise. Vision Res 37: 1415–1427, 1997.[CrossRef][Web of Science][Medline]
Croner LJ and Albright TD. Segmentation by color influences responses of motion-sensitive neurons in the cortical middle temporal visual area. J Neurosci 19: 3935–3951, 1999.
Das A and Gilbert CD. Topography of contextual modulations mediated by short-range interactions in primary visual cortex. Nature 399: 655–661, 1999.[CrossRef][Medline]
DeAngelis GC, Freeman RD, and Ohzawa I. Length and width tuning of neurons in the cat's primary visual cortex. J Neurophysiol 71: 347–374, 1994.
Dobkins KR and Albright TD. What happens if it changes color when it moves?: the nature of chromatic input to macaque visual area MT. J Neurosci 14: 4854–4870, 1994.[Abstract]
Famiglietti EV Jr and Kolb H. Structural basis for ON-and OFF-center responses in retinal ganglion cells. Science 194: 193–195, 1976.
Foster DH and Westland S. Orientation contrast vs orientation in line-target detection. Vision Res 35: 733–738, 1995.[CrossRef][Web of Science][Medline]
Fries W, Albus K, and Creutzfeldt OD. Effects of interacting visual patterns on single cell responses in cats striate cortex. Vision Res 17: 1001–1008, 1977.[CrossRef][Web of Science][Medline]
Gilbert CD and Wiesel TN. The influence of contextual stimuli on the orientation selectivity of cells in primary visual cortex of the cat. Vision Res 30: 1689–1701, 1990.[CrossRef][Web of Science][Medline]
Hegde J and Felleman DJ. How selective are V1 cells for pop-out stimuli? J Neurosci 23: 9968–9980, 2003.
Horton JC and Sherk H. Receptive field properties in the cat's lateral geniculate nucleus in the absence of on-center retinal input. J Neurosci 4: 374–380, 1984.[Abstract]
Hubel DH and Livingstone MS. Color and contrast sensitivity in the lateral geniculate body and primary visual cortex of the macaque monkey. J Neurosci 10: 2223–2237, 1990.[Abstract]
Hubel DH and Wiesel TN. Receptive fields and functional architecture of monkey striate cortex. J Physiol 195: 215–243, 1968.
Hubel DH and Wiesel TN. Uniformity of monkey striate cortex: A parallel relationship between field size, scatter, and magnification factor. J Comp Neurol 158: 295–305, 1974.[CrossRef][Web of Science][Medline]
Judge SJ, Richmond BJ, and Chu FC. Implantation of magnetic search coils for measurement of eye position: an improved method. Vision Res 20: 535–538, 1980.[CrossRef][Web of Science][Medline]
Kapadia MK, Ito M, Gilbert CD, and Westheimer G. Improvement in visual sensitivity by changes in local context: parallel studies in human observers and in V1 of alert monkeys. Neuron 15: 843–856, 1995.[CrossRef][Web of Science][Medline]
Kisvarday ZF, Toth E, Rausch M, and Eysel UT. Orientation-specific relationship between populations of excitatory and inhibitory lateral connections in the visual cortex of the cat. Cereb Cortex 7: 605–618, 1997.
Knierim JJ, and van Essen DC. Neuronal responses to static texture patterns in area V1 of the alert macaque monkey. J Neurophysiol 67: 961–980, 1992.
Kooi FL, Toet A, Tripathy SP, and Levi DM. The effect of similarity and duration on spatial interaction in peripheral vision. Spat Vis 8: 255–279, 1994.[Web of Science][Medline]
Lamme VAF. The neurophysiology of figure-ground segregation in primary visual cortex. J Neurosci 15: 1605–1615, 1995.[Abstract]
Lamme VAF, Super H, and Spekreijse H. Feedforward, horizontal, and feedback processing in the visual cortex. Curr Opin Neurobiol 8: 529–535, 1998.[CrossRef][Web of Science][Medline]
Levitt JB and Lund JS. Contrast dependence of contextual effects in primate visual cortex. Nature 387: 73–76, 1997.[CrossRef][Medline]
Li CY and Li W. Extensive integration field beyond the classical receptive field of cat's striate cortical neurons—classification and tuning properties. Vision Res 34: 2337–2355, 1994.[CrossRef][Web of Science][Medline]
Li W, Thier P, and Wehrhahn C. Contextual influence on orientation discrimination of humans and responses of neurons in V1 of alert monkeys. J Neurophysiol 83: 941–954, 2000.
Livingstone MS and Hubel DH. Psychophysical evidence for separate channels for the perception of form, color, movement, and depth. J Neurosci 7: 3416–3468, 1987.[Abstract]
Maffei L and Fiorentini A. The unresponsive regions of visual cortical receptive fields. Vision Res 16: 1131–1139, 1976.[CrossRef][Web of Science][Medline]
Malach R, Amir Y, Harel M, and Grinvald A. Relationship between intrinsic connections and functional architecture revealed by optical imaging and in vivo targeted biocytin injections in primate striate cortex. Proc Natl Acad Sci USA 90: 10469–10473, 1993.
Mareschal I, Sceniak MP, and Shapley RM. Contextual influences on orientation discrimination: binding local and global cues. Vision Res 41: 1915–1930, 2001.[CrossRef][Web of Science][Medline]
McGuire BA, Gilbert CD, Rivlin PK, and Wiesel TN. Targets of horizontal connections in macaque primary visual cortex. J Comp Neurol 305: 370–392, 1991.[CrossRef][Web of Science][Medline]
Moran J and Desimone R. Selective attention gates visual processing in the extrastriate cortex. Science 229: 782–784, 1985.
Nelson JI and Frost BJ. Orientation-selective inhibition from beyond the classic visual receptive field. Brain Res 139: 359–365, 1978.[CrossRef][Web of Science][Medline]
Posner MI. Orienting of attention. J Exp Psychol 32: 3–25, 1980.
Robinson DA. A method of measuring eye movement using a scleral search coil in a magnetic field. IEEE Trans Bio-Med Electronics 10: 137–145, 1963.
Roelfsema PR, Lamme VA, and Spekreijse H. Synchrony and covariation of firing rates in the primary visual cortex during contour grouping. Nat Neurosci 7: 982–991, 2004.[CrossRef][Web of Science][Medline]
Sceniak MP, Hawken MJ, and Shapley R. Visual spatial characterization of macaque V1 neurons. J Neurophysiol 85: 1873–1887, 2001.
Schiller PH. The connections of the retinal on and off pathways to the lateral geniculate nucleus of the monkey. Vision Res 24: 923–932, 1984.[CrossRef][Web of Science][Medline]
Schiller PH. The ON and OFF channels of the visual system. Trends Neurosci 15: 86–92, 1992.[CrossRef][Web of Science][Medline]
Schiller PH, Sandell JH, and Maunsell JH. Functions of the ON and OFF channels of the visual system. Nature 322: 824–825, 1986.[CrossRef][Medline]
Sekuler RW. Spatial and temporal determinants of visual backward masking. J Exp Psychol 70: 401–406, 1965.[CrossRef][Web of Science][Medline]
Shadlen MN, Britten KH, Newsome WT, and Movshon JA. A computational analysis of the relationship between neuronal and behavioral responses to visual motion. J Neurosci 16: 1486–1510, 1996.
Squatrito S, Trotter Y, and Poggio GF. Influences of uniform and textured backgrounds on the impulse activity of neurons in area V1 of the alert macaque. Brain Res 536: 261–270, 1990.[CrossRef][Web of Science][Medline]
Stettler DD, Das A, Bennett J, and Gilbert CD. Lateral connectivity and contextual interactions in macaque primary visual cortex. Neuron 36: 739–750, 2002.[CrossRef][Web of Science][Medline]
Stoner GR and Albright TD. Image segmentation cues in motion processing: implications for modularity in vision. J Cogn Neurosci 5: 129–149, 1993.
Treisman AM and Gelade G. A feature-integration theory of attention. Cogn Psychol 12: 97–136, 1980.[CrossRef][Web of Science][Medline]
van der Smagt MJ and van de Grind WA. Integration and segregation of local motion signals: the role of contrast polarity. Vision Res 39: 811–822, 1999.[CrossRef][Web of Science][Medline]
van der Smagt MJ, Wehrhahn C, and Albright TD. Contextual masking effects of color and orientation on Macque V1 responses. Soc Neurosci Abstr 457.454, 2002.
Van Essen DC, Newsome WT, and Maunsell JHR. The visual field representation in striate cortex of the macaque monkey: asymmetries, anisotropies, and individual variability. Vision Res 24: 429–448, 1984.[CrossRef][Web of Science][Medline]
Watson AB, Nielsen KRK, Poirson A, Fitzhugh A, Bilson A, Nguyen K, and Ahumada J. Use of a raster framebuffer in vision research. Behav Res Meth Instru Comp 18: 587–594, 1986.
Wehrhahn C and Dresp B. Detection facilitation by collinear stimuli in humans: dependence on strength and sign of contrast. Vision Res 38: 423–428, 1998.[CrossRef][Web of Science][Medline]
Wehrhahn C, Li W, and Westheimer G. Patterns that impair discrimination of line orientation in human vision. Perception 25: 1053–1064, 1996.[Web of Science][Medline]
Westheimer G, Shimamura K, and McKee SP. Interference with line-orientation sensitivity. J Opt Soc Am 66: 332–338, 1976.[Medline]
Wiesel TN and Hubel DH. Spatial and chromatic interactions in the lateral geniculate body of the rhesus monkey. J Neurophysiol 29: 1115–1156, 1966.
Zipser K, Lamme VA, and Schiller PH. Contextual modulation in primary visual cortex. J Neurosci 16: 7376–7389, 1996.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| Visit Other APS Journals Online |
