Abstract

Visual scene perception owes greatly to surface features such as color and brightness. Yet, early visual cortical areas predominantly encode surface boundaries rather than surface interiors. Whether human early visual cortex may nevertheless carry a small signal relevant for surface perception is a topic of debate. We induced brightness changes in a physically constant surface by temporally modulating the luminance of surrounding surfaces in seven human participants. We found that fMRI activity in the V2 representation of the constant surface was in antiphase to luminance changes of surrounding surfaces (i.e., activity was in-phase with perceived brightness changes). Moreover, the amplitude of the antiphase fMRI activity in V2 predicted the strength of illusory brightness perception. We interpret our findings as evidence for a surface-related signal in early visual cortex and discuss the neural mechanisms that may underlie that signal in concurrence with its possible interaction with the properties of the fMRI signal.

INTRODUCTION

Traditional views and empirical work of early vision have emphasized the role of local contrast in the reconstruction of the visual image (Hubel & Wiesel, 1962). However, mechanisms extracting local contrast appear ill suited to encode homogenous stimulus surfaces, and despite a significant amount of work, the question by which means visual surface properties are encoded remains unresolved (Komatsu, 2006; Pessoa, Thompson, & Noe, 1998). Neurophysiological studies have demonstrated that a minority of neurons in early visual cortex responds to homogenous surface luminance (Roe, Lu, & Hung, 2005; Kinoshita & Komatsu, 2001; MacEvoy, Kim, & Paradiso, 1998; Rossi, Rittenhouse, & Paradiso, 1996), and fMRI has confirmed the existence of a signal related to surface luminance in early visual cortex (Haynes, Lotto, & Rees, 2004). Furthermore, neurophysiological studies indicate that there are neurons in anatomical subcompartments of V2 that carry a signal related to the perception of surface brightness in the Craik–O'Brian–Cornsweet (COC) illusion (Hung, Ramsden, & Roe, 2007; Roe et al., 2005), and other studies showed a correlate of brightness perception in a small number of V1 neurons using an induced brightness paradigm (Rossi & Paradiso, 1999; Rossi et al., 1996). The goal of the present study was to investigate whether the signal related to surface brightness perception that has been reported by neurophysiological and optical imaging studies in cats and monkeys could also be demonstrated with fMRI in humans.

To study the perception of visual surface brightness, we used a dynamic brightness induction paradigm similar to that used by Paradiso and colleagues, who recorded from neurons in Cat area 17 (primary visual cortex) while a gray surface patch of constant luminance was placed over their receptive fields (RFs) (Rossi et al., 1996). A minority of neurons (approximately 10%) responded in antiphase to luminance modulations in abutting inducer surfaces far outside the RFs, which suggests that these neurons helped encoding the counterphase brightness changes that humans perceive in the same stimuli. The illusion and its neural correlate were present with a fixed gray surface (induction condition) and absent with a black surface (control condition). We will refer to the area of fixed luminance as the “probing region,” as it was designed to probe mechanisms of surface perception.

The antiphase relationship between perceived brightness and inducer luminance is the hallmark of brightness induction (Rossi & Paradiso, 1999; Rossi et al., 1996). To demonstrate this property in human visual cortex with fMRI, we used a slow event-related design in which changes between high and low inducer luminance levels were separated by several seconds, thereby taking into account the slow fMRI hemodynamic signal (Logothetis, Pauls, Augath, Trinath, & Oeltermann, 2001; Boynton, Engel, Glover, & Heeger, 1996). Furthermore, we optimized the spatial dimensions of the stimulus to take into account the spatial resolution limits of the fMRI signal (Cornelissen, Wade, Vladusich, Dougherty, & Wandell, 2006; Sereno et al., 1995) and limits in the spatial extent of brightness induction (Pereverzeva & Murray, 2008). A possible brightness-related antiphase signal in the probing region could be partly annulled or masked by opposite-phase signals from abutting inducers, which would be strongest near the retinotopic projections of the inducer and probing region borders. Therefore, we hypothesized that a brightness-related antiphase fMRI signal would be strongest near the middle of the retinotopic projection of the probing region and would counteract in-phase responses to inducers. We used localizer stimuli to delineate retinotopic projections of inducers and probing region.

METHODS

Participants

Seven healthy adults (25–30 years old) with corrected-to-normal visual acuity participated after written informed consent. All participants completed the main fMRI brightness induction and control experiments with stimulus patterns that comprised two inducers, as well as associated retinotopy and eccentricity localizers, and brightness psychophysical rating task. Three of the participants completed additional fMRI experiments in which luminance changes were presented in only one inducer, as well as associated psychophysical ratings of brightness induction.

Stimulus Design

Radius of the inner inducer disk was 3°, width of the probing region was 6°, and width of the outer inducer annulus was 5°. Stimuli were presented concentrically around a fixation cross (1° width and height) placed on the inner inducer. The probing region was gray (7.9 cd/m2; induction condition) or black (1.1 cd/m2; control condition). The inducers changed between minimum and maximum luminance in a sinusoidal fashion in 1 sec (range = 1.1–164 cd/m2; cf. Rossi & Paradiso, 1999). During inducer luminance modulations, contrast at the borders with the probing region varied between −76% and 91% in the induction condition and between 0% and 99% in the control condition (Michelson indexes). Figure 1A shows the induction and control stimuli at different luminance levels of the inducers.

Figure 1. 

Stimuli and experimental design. (A) Brightness induction and control stimuli (top) and timing of events (bottom). A fixation period (gray area in timing diagram) preceded each block of luminance changes. A block started with a gray-to-dark change in inducers (white area in timing diagram), followed by dark-to-bright (red area) and bright-to-dark changes (blue area; 1-sec duration), after which the new luminance value was kept constant for a variable intertrial interval (7–9 sec). The block ended with a dark-to-gray change (not shown). Luminance changes occurred only in the inducers (concentric rings, left inset), whereas probing region luminance (single annulus inset) was kept constant throughout the measurement (straight line timing diagram). The brightness perception in the gray probing region (bottom timing diagram) follows a course inverse to the physical luminance changes in inducers. (B) Schematic of a single run (color coding as in A). Between the fourth and fifth blocks, a short “relaxation period” was used (black-and-white textured region), during which participants could briefly refrain from fixation. (C) Localizer rings map onto the center of the probing and inducer regions of the stimuli. Smaller insets illustrate localizer stimulus separately for the central inducer (left), probing region (middle), and peripheral inducer areas (right; stimuli are scaled for visualization purposes).

Figure 1. 

Stimuli and experimental design. (A) Brightness induction and control stimuli (top) and timing of events (bottom). A fixation period (gray area in timing diagram) preceded each block of luminance changes. A block started with a gray-to-dark change in inducers (white area in timing diagram), followed by dark-to-bright (red area) and bright-to-dark changes (blue area; 1-sec duration), after which the new luminance value was kept constant for a variable intertrial interval (7–9 sec). The block ended with a dark-to-gray change (not shown). Luminance changes occurred only in the inducers (concentric rings, left inset), whereas probing region luminance (single annulus inset) was kept constant throughout the measurement (straight line timing diagram). The brightness perception in the gray probing region (bottom timing diagram) follows a course inverse to the physical luminance changes in inducers. (B) Schematic of a single run (color coding as in A). Between the fourth and fifth blocks, a short “relaxation period” was used (black-and-white textured region), during which participants could briefly refrain from fixation. (C) Localizer rings map onto the center of the probing and inducer regions of the stimuli. Smaller insets illustrate localizer stimulus separately for the central inducer (left), probing region (middle), and peripheral inducer areas (right; stimuli are scaled for visualization purposes).

The size of the probing region satisfied two opposite demands: maximization of probing region size to reduce effects of fMRI spatial signal spread (De Weerd, Karni, Kastner, Ungerleider, & Jezzard, 1997; Sereno et al., 1995) and minimization of probing region size to maximize the illusion across its extent (Pereverzeva & Murray, 2008). First, the spatial spread of fMRI signal dictates that the probing region should be maximized because the signals associated with probing region and inducers were expected to be in antiphase during brightness induction. Antiphase signals attenuate each other when they spread into each other's retinotopic territory. Point spread of the fMRI signal has been estimated to be on the order of just a few millimeters (Engel, Glover, & Wandell, 1997; Sereno et al., 1995), but recent estimates suggested spread of approximately 7 mm at half-width-at-half-maximum (HWHM; Cornelissen et al., 2006). According to an estimate of V1 cortical magnification (Sereno et al., 1995), the distance from the inner to the outer border of the probing region projected on V1 is about 15 mm. In light of this estimate of fMRI signal spread and chosen stimulus dimensions, a sufficient proportion of a hypothetical counterphase signal related to brightness induction was expected to survive in the middle of the probing region, potentially even directly revealing antiphase activity (depending on the strength of that signal). Going from the middle of the probing region to its border, in-phase activity from inducers was expected to increasingly dominate the hypothetical antiphase signal in the probing region. A hypothetical antiphase signal associated with induction would also spread into the territory of the inducers, thereby reduce in-phase signal of inducers. Thus, suppression of in-phase activity from inducers is an important and integral aspect of expected results during brightness induction. We anticipated that smaller cortical magnification and other factors in extrastriate areas could render the separation of signals from probing and inducer regions more difficult. Second, although the spatial spread of fMRI signal argues in favor of making the probing region as large as possible, this manipulation also decreases the strength of the illusion, which puts a limit on the extent to which the probing region can be enlarged. Using a probing region with a cortical projection in V1 that was larger than ours (18 mm), Pereverzeva and Murray (2008) found a decline of brightness induction toward the middle of the probing region. The smaller probing region in our study (both in the visual field and on cortex) was chosen to limit the decline of the illusion toward the middle.

Localizer stimuli were centered in the three parts of the stimuli. The inner radii of central, middle, and outer localizer rings were 0°, 5.4°, and 11°, respectively, and their outer radii were 0.5°, 6.6°, and 13°, respectively. Each localizer consisted of 20 sequentially positioned black-and-white segments, which reversed contrast at a rate of 16 Hz. Localizer stimuli were presented on a fixed gray background (7.9 cd/m2).

Stimuli were projected onto a diagonally positioned mirror attached to the head coil in the scanner bore using an LCD projector controlled by a PC (screen refresh rate = 60 Hz) running Presentation software (version 9.90, Neurobehavioral Systems, Inc., Albany, CA). The onsets of luminance modulations were pulse-triggered by the T2*-weighted image acquisition.

Psychophysical Ratings of Induction

All participants rated the strength of induced brightness change in a psychophysical experiment in the scanner. Participants mimicked perceptual changes in the probing region of an experimental stimulus by setting a physical luminance value in the probing region of a testing stimulus. Participants were presented with six conditions: Three types of stimulus display (two inducers, only inner inducer, only outer inducer) combined with the two luminance levels of the probing region (gray and black). For each condition, participants completed 24 trials. Each trial comprised (1) a 1-sec starting phase, in which one or both inducers were shown at their maximum or minimum luminance; (2) a luminance modulation phase, in which one or both inducers gradually changed luminance in 1 sec from maximum to minimum (down-sweep) or minimum to maximum luminance (up-sweep); (3) a 1-sec end phase, in which one or both inducers remained at final luminance; and (4) a masking phase, in which the surfaces of both inducers (but not the probing region) were masked by a black-and-white checkerboard mask. At the start of each trial, the probing region was set to gray or black, which remained fixed throughout the four trial phases. During the masking phase, participants could manually set on a continuous scale (using two buttons) the luminance level of the probing region to the luminance level that matched perceived brightness at the end of the modulation phase. The physical luminance of the probing region and the luminance set to match induction (cd/cm2) were log-transformed (natural log). A rating index (RI) was obtained by subtracting the log-transformed starting luminance values of the probing region from the log-transformed ratings provided by the subjects. The resulting RI was negative when the luminance deviation in probing region was set in antiphase to inducer luminance modulations, and a positive sign was used for in-phase settings. Figure 2 shows the mean RIs in the six conditions (error bars represent 1 SEM) obtained in seven subjects.

Figure 2. 

Psychophysical rating results. RIs (bars) plotted as a function of the six stimulus conditions averaged over all participants (N = 7). Gray bars refer to stimulus conditions in which the probing region was gray; black bars refer to conditions in which the probing region was black. Insets above the bars are schematic illustrations of the stimuli. The white regions in the illustrations indicate the inducers of which the luminance levels were modulated (two insets on the left: two inducers; two insets in the middle: inner inducer; two insets on the right: outer inducer). The luminance of the probing region in the insets refers to the fixed luminance of the probing region during the rating trials, which then could be adapted by participants to reflect experienced induction. Error bars (SEM) and bar sizes for the black probing region were close to zero, as subjects did not perceive brightness changes during inducer luminance modulations.

Figure 2. 

Psychophysical rating results. RIs (bars) plotted as a function of the six stimulus conditions averaged over all participants (N = 7). Gray bars refer to stimulus conditions in which the probing region was gray; black bars refer to conditions in which the probing region was black. Insets above the bars are schematic illustrations of the stimuli. The white regions in the illustrations indicate the inducers of which the luminance levels were modulated (two insets on the left: two inducers; two insets in the middle: inner inducer; two insets on the right: outer inducer). The luminance of the probing region in the insets refers to the fixed luminance of the probing region during the rating trials, which then could be adapted by participants to reflect experienced induction. Error bars (SEM) and bar sizes for the black probing region were close to zero, as subjects did not perceive brightness changes during inducer luminance modulations.

Functional Imaging Experimental Design

Functional runs in induction and control condition (Figure 1A) lasted each 460 sec. Eccentricity localizer functional runs lasted 290 sec and consisted of an alternation of 11-sec baseline and 11-sec localizer stimulation blocks (three blocks for each localizer). Four participants were shown three runs in induction condition, then three runs in control condition, and finally two localizer runs. The other participants started with control runs, followed by induction and localizer runs. A single session, thus, contained eight runs and anatomical scan. In three participants, additional data were collected in separate sessions using stimulus patterns in which outer or inner inducer was removed and replaced with constant black (1.1 cd/m2).

Scanning Parameters

Structural MRI and fMRI scanning were performed on a 3T Siemens Allegra head scanner using standard procedures and bird cage head coil. Slice positioning was directed at calcarine sulcus and lower gyrus. A high-resolution T1-weighted 3-D anatomical scan was acquired for each participant (MP-RAGE; repetition time = 2250 msec, matrix size = 256 × 256, 192 slices, in-plane resolution = 1 mm2). To measure BOLD contrast, a standard T2*-weighted gradient-echo EPI was used to acquire 14 slices (repetition time = 1000 msec, echo time = 30 msec, flip angle = 90°, slice thickness = 2 mm, no slice gap, matrix size = 128 × 128, in-plane resolution = 2 mm2). Chosen parameters maximized temporal and spatial resolution, at the price of being able to cover only 28 mm from top to bottom of slice block. Resulting coverage is insufficient to cover ventral and dorsal representations of early visual cortical areas. We chose to align the top of the block of slices to the calcarine fissure so that data collection typically was limited to the contralateral upper quadrant representation in ventral V1, V2, and V3.

Preprocessing and Analysis of MRIs

Anatomical images were transformed to 3-D standardized space (Talairach & Tournoux, 1988) with a resampled voxel size of 1 mm3. Functional measurements were coregistered to prestandardized 3-D anatomical space and preprocessed using BrainVoyager QX (Brain Innovation, Maastricht, the Netherlands). Of each functional run, the first six images were discarded. Functional images were transformed to 3-D standardized space (Talairach & Tournoux, 1988) with a resampled isotopic voxel size of 2 mm3. Preprocessing steps of functional time series included slice time correction, 3-D correction for head volume movements (sinc interpolation), and temporal filtering, which included linear trend removal and a high-pass filter of three cycles per time course (Goebel, Esposito, & Formisano, 2006).

The spatially standardized anatomical images were segmented according to the gray and white matter boundaries, and the cortical sheet of each hemisphere was then tessellated into a 3-D cortical surface representation (Goebel et al., 2006; Kriegeskorte & Goebel, 2001). The cortical surfaces were then inflated to push the sulci outward and flattened to achieve a 2-D representation of the entire cortical surface of each hemisphere. The time series of the volume were resampled to the cortical surfaces and slightly spatially smoothed using a 2-mm smoothing kernel based on nearest vertex neighbors. The time series on the cortical surface were used for functional analysis.

A time series analysis was conducted using a multiple regression analysis in the context of the general linear model for fMRI (Worsley & Friston, 1995). For induction and control conditions, the general linear model consisted of a single predictor (luminance changes relative to gray level in fixation periods; effect coding, 0 = baseline) and a covariate (relaxation period; dummy coding) for each functional run. For the localizer experiment, the model comprised three predictors that modeled the sequence of blocks of eccentricity localizers (dummy coding). All predictors were convolved with a two-gamma function to correct for the hemodynamic response delay.

Retinotopic Mapping and Functional Alignment

Retinotopic mapping (polar angle mapping) was carried out using standard procedures (Linden, Kallenbach, Heinecke, Singer, & Goebel, 1999; Engel et al., 1997; Sereno et al., 1995). Polar mapping was used to delineate borders between V1, V2, and V3, which were mapped on inflated cortical renderings of individual participants. The borders were used as guidelines to draw an equipolar line from central to peripheral representations within each visual area (Figure 3B) in each hemisphere. Because vertices in the inflated brain representation preserve anatomical distances (Goebel et al., 2006), matching an appropriately drawn equipolar line with outcomes of the localizer experiment permits localizing activity distributions (beta coefficients resulting from linear regression analysis) as a function of anatomical distance in millimeters. We observed large differences in cortical magnification among participants, hemispheres, and visual areas; hence, the length of sampling differed from case to case. To average across hemispheres, we functionally aligned the anatomically selected cortical paths within each visual area across hemispheres into a common space by nonlinear morphing, similar to previously published approaches (e.g., Listgarten, Neal, Roweis, & Emili, 2005; Forshed, Schuppe-Koistinen, & Jacobsson, 2003). Alignment, analysis, and plotting of results were performed using custom-written routines in Matlab (The Mathworks, Natick, MA). By functionally aligning the cortical paths, constrained by localizer data, our data became spatially comparable over hemispheres and could be collapsed. As a result, averaged beta coefficients are shown as a function of anatomical distance. The method is described in more detail elsewhere (Jans, Been, van de Ven, Goebel, & De Weerd, submitted).

RESULTS

Psychophysics

Participants verbally reported seeing the brightness illusion most strongly and consistently when the probing region was gray and only in cases in which the outer inducer changed luminance, alone or in conjunction with the inner inducer. The psychophysical ratings obtained in a separate session in the scanner supported the participants' verbal reports (see Figure 2). Ratings differed significantly from the initial brightness of the gray probing region immediately after a change in luminance of both inducers (T[6] = −6.7, p = .0005) or immediately after a luminance change in only the outer inducer (T[6] = −4.2, p = .0055). In both cases, the rated changes in brightness were in the direction opposite to the luminance changes of the inducers. Participants did not see the brightness illusion in the gray probing region when luminance changed only in the inner inducer (T[6] = 1.9, p = .11). Participants perceived no brightness changes in the black probing region after luminance changes in any of the inducers (ps > .23). Interestingly, the ratings suggested that the brightness illusion in the gray probing region was stronger when luminance changed simultaneously in both inducers compared with only the outer inducer. Analysis of the difference in ratings between the two conditions was significant (T[6] = −2.2, p = .034, one-tailed). There was no difference in the strength of the induction effect when the data for inducer luminance increases and decreases were compared (ps > .27).

Neural Correlates of Brightness Induction on the Cortical Surface

We analyzed fMRI signal during the brightness induction and control conditions sampled along cortical paths drawn in the ventral part of early visual areas V1, V2, and V3 to which data collection was limited (see Methods). In the remainder of the article, we use V1, V2, and V3 as shorthand for the ventral part of these visual areas. Data were sampled along cortical paths through the stimulus representations in V1, V2, and V3 (Figure 3A) in 14 hemispheres from seven participants. After alignment, beta coefficients were averaged along the 14 anatomical paths drawn through the stimulus representations in each area (see Methods). The resulting distributions of beta coefficients show the correlation of fMRI signal with repeated localizer presentations or repeated inducer luminance modulations (Figure 3BD).

Figure 3. 

Cortical responses for the localizer and brightness stimuli. (A) Illustration of cortical paths along which functional data were sampled in V1, V2, and V3. The cortical paths contain a color gradient (purple to green) corresponding to the continuum from fovea to periphery. (B–F) Overview of fMRI results for localizer and brightness induction experiments. Rows are plots for V1 (top row), V2 (middle row), and V3 (bottom row). (B) Beta coefficients of the inner (purple), middle (orange), and outer (green) localizers sampled along the cortical paths (x axis is marked with the same color gradient as the cortical paths in A) plotted as a function of cortical distance (mm). Each localizer shows one peak of positive beta coefficients and a decline away from the peaks. (C) Beta coefficients of the brightness induction (blue) and control (green) stimuli sampled along the cortical paths plotted as a function of cortical distance (mm). Shaded regions represent ±1 SEM (N = 7, hemispheres averaged within participants). Display conventions as in column B. (D) Means of beta coefficients in inducers (black) and probing region (white).

Figure 3. 

Cortical responses for the localizer and brightness stimuli. (A) Illustration of cortical paths along which functional data were sampled in V1, V2, and V3. The cortical paths contain a color gradient (purple to green) corresponding to the continuum from fovea to periphery. (B–F) Overview of fMRI results for localizer and brightness induction experiments. Rows are plots for V1 (top row), V2 (middle row), and V3 (bottom row). (B) Beta coefficients of the inner (purple), middle (orange), and outer (green) localizers sampled along the cortical paths (x axis is marked with the same color gradient as the cortical paths in A) plotted as a function of cortical distance (mm). Each localizer shows one peak of positive beta coefficients and a decline away from the peaks. (C) Beta coefficients of the brightness induction (blue) and control (green) stimuli sampled along the cortical paths plotted as a function of cortical distance (mm). Shaded regions represent ±1 SEM (N = 7, hemispheres averaged within participants). Display conventions as in column B. (D) Means of beta coefficients in inducers (black) and probing region (white).

Figure 3B and C shows the beta coefficients in localizer and brightness induction experiments sampled across the cortical paths in V1, V2, and V3. Beta coefficients for the inner (purple line), middle (orange), and outer localizers (green) are shown in Figure 3B. The peaks indicate the approximate anatomical location of the localizer stimuli, and distributions of beta coefficients show a large spread away from the peaks of the localizer stimuli. Figure 3C shows the coefficients of the brightness induction (blue) and control condition (green) in the main experiment. We applied a repeated measures ANOVA to the data in Figure 3C, with within-subject factors Visual Area (V1, V2, V3), Condition (illusion, control), and Stimulus Region (probing region, inducers), to the beta coefficients. We found significant main effects for Condition (F[1, 6] = 66.6, p < .001) and Stimulus Region (F[1, 6] = 36.3, p = .001) and a significant Visual Area × Condition interaction effect (F[2, 12] = 4.5, p = .035).

In the control condition (green line), in which participants reported no brightness induction (see Figure 1), mostly in-phase fMRI signal was found in the cortical projection areas of the localizers and probing region in all three visual areas (i.e., signal correlated positively with inducer luminance modulations; see Table 1). The in-phase signal was strongest for inducers and became weaker toward the middle of the probing region (Inducer > Probing; V1: T[6] = 5.0, p = .0025; V2: T[6] = 2.9, p = .0282; V3: T[6] = 2.2, p = .069). At the probing region center, fMRI signal remained in-phase with the luminance changes (i.e., average beta coefficients were of positive sign) in all three visual areas (minimum T = 1.9; see Table 1). In the induction condition (blue line in Figure 3), in which participants reported antiphase brightness induction, beta coefficients at the probing region were significantly smaller than coefficients at the inducers in all three visual areas (V1: T[6] = 9.3, p < .001; V2: T[6] = 8.6, p < .001; V3: T[6] = 4.3, p = .005). In addition, however, three findings suggest the presence of another antiphase signal in the probing region associated with surface brightness perception. First, there was a strong decline of beta coefficients in the probing region during brightness induction compared with the control condition, as is evidenced by the divergence of green (control condition) and blue lines (induction condition; Control > Induction at probing region; V1: T[6] = 4.8, p = .003; V2: T[6] = 4.6, p = .004; V3: T[6] = 2.5, p = .045). Second, this divergence was statistically significant not only in the probing region but also in the inducer region (Control > Induction at inducers; V1: T[6] = 2.7, p = .035; V2: T[6] = 4.4, p = .004; V3: T[6] = 2.9, p = .029). Third, the average of the coefficients was negative at the cortical center of the probing region for all three visual areas (all Ts < −0.5) but was significant only for V2 (T[6] = −3.7, p = .01), but not for V1 (p = .4) or V3 (p = .6; see Table 1). Taken together, this analysis indicates that in V2, but not in V1 or V3, there is an antiphase signal in early visual cortex related to brightness induction in the probing region that, because of spatial spread, also attenuated in-phase activity of inducer regions.

Table 1. 

fMRI Signal in Inducers and Probing Region


Induction
Control
Inducers
Probing Region
Inducers
Probing Region
T
p
T
p
T
p
T
p
V1 5.2 .0021 −0.8 .4292 4.8 .0032 3.0 .0232 
V2 5.0 .0026 −3.7 .0101 5.6 .0014 1.9 .1023 
V3 4.8 .0029 −0.5 .6179 5.9 .0010 2.3 .0642 

Induction
Control
Inducers
Probing Region
Inducers
Probing Region
T
p
T
p
T
p
T
p
V1 5.2 .0021 −0.8 .4292 4.8 .0032 3.0 .0232 
V2 5.0 .0026 −3.7 .0101 5.6 .0014 1.9 .1023 
V3 4.8 .0029 −0.5 .6179 5.9 .0010 2.3 .0642 

Tests were performed on the beta coefficients sampled along the cortical paths in visual areas V1, V2, and V3 and indicate whether the positive or negative average Beta values are significantly larger or smaller than zero (note the negative t values inside the probing region in the induction condition and especially the highly negative values in V2).

Control Experiments

The observation that, in V2, antiphase activity during induction was lost during the control condition strongly suggests that activity in early visual cortex contributed to the encoding of brightness in the probing region. However, local contrast changes at the border between inducers and probing region were different in the induction and control conditions (see Methods; Figure 1A). Prior fMRI studies have shown that local changes in activity at one location in the retinotopic map can lead to opposite changes in activity in surrounding regions (Shmuel, Augath, Oeltermann, & Logothetis, 2006), possibly because of long-range lateral inhibition (Angelucci et al., 2002; Levitt & Lund, 2002). Hence, the specific local contrast changes in the induction condition may have led to antiphase activity inside the probing region, whereas, for unknown reasons, the specific local contrast changes in the control condition may not have. Under this scenario, the antiphase fMRI activity in the probing region during brightness induction might have been an artifact of local contrast conditions rather than a correlate of surface brightness induction. To exclude this possibility, a stimulus was devised, in which the outer inducer was replaced with constant black and only the luminance of the inner inducer was manipulated. In this stimulus, there is no brightness induction neither for the gray nor for the black probing region (see rating results in Figure 2), yet local contrast changes at the border between probing and inner inducer regions were identical to those in the original induction and control stimuli. If local contrast changes indeed determined the fMRI response irrespective of experienced surface brightness, then antiphase activity would be expected for the gray, but not the black, probing region, mimicking results with two inducers. However, if the perception of brightness induction were the determining factor for antiphase activity in the probing region, then antiphase activity would be expected neither for the gray nor the black probing region. Our results in V2 support the latter hypothesis, as the responses to the inducer next to the black and gray probing regions were indistinguishable (Figure 4A; p > .6, at a cortical distance of 27.8 mm). This confirms that differences in local contrast changes between brightness induction and control stimuli are insufficient to explain the antiphase activity observed during brightness induction (with a gray probing region and two inducers). Note that the data in Figure 4A (from the condition in which only the inner inducer changed luminance) suggest that activity related to the inner inducer spread farther into the probing region than in the main experiment (in which both inducers changed luminance). This may be related to a release from inhibitory interaction between inducers at opposing sides of the probing region, but dedicated experiments are required to test that hypothesis.

Figure 4. 

fMRI signal in brightness induction. (A) Beta coefficients of luminance modulations of the inner inducer-only condition shown as a function of V2 cortical distance. Responses to the stimulus with the gray (black) probing region are shown in red (blue). Shaded regions represent ±1 SEM. Activity did not differ between conditions at any cortical position. (B–D) RI as a function of amplitude index from selected probing region voxels (see Figure 3) plotted for three stimulus conditions with the gray probing region in V1 (B), V2 (C), and V3 (D) during luminance up and down modulations. Colored symbols refer to participant data for induction conditions (see legend inset).

Figure 4. 

fMRI signal in brightness induction. (A) Beta coefficients of luminance modulations of the inner inducer-only condition shown as a function of V2 cortical distance. Responses to the stimulus with the gray (black) probing region are shown in red (blue). Shaded regions represent ±1 SEM. Activity did not differ between conditions at any cortical position. (B–D) RI as a function of amplitude index from selected probing region voxels (see Figure 3) plotted for three stimulus conditions with the gray probing region in V1 (B), V2 (C), and V3 (D) during luminance up and down modulations. Colored symbols refer to participant data for induction conditions (see legend inset).

Association between Ratings and Neural Correlate of Brightness Illusion

Finally, to investigate the perceptual relevance of antiphase activity in the probing region, we tested whether the strength of antiphase activity predicted the strength of brightness induction. We correlated the ratings of the participants and conditions for which we obtained functional imaging data. Results are shown in the scatterplots of Figure 4BD, with the correlation value, 95% confidence interval (CI), and associated p value listed at the top of each panel. Ratings correlated significantly with the amplitude index across the conditions only in area V2 (r = 0.5, 95% CI = [0.09, 0.64], p < .001), but not in V1 (r = 0.13, 95% CI = [−0.17, 0.48], p = .48) or in V3 (r = 0.35, 95% CI = [−0.06, 0.56], p = .08). Thus, fMRI activity in the probing region's representation in V2 predicted the perceived strength of induced surface brightness in the different stimulus conditions, but not in V1 and V3.

DISCUSSION

Using fMRI, we investigated whether early visual cortex shows a functional neural correlate for brightness induction. We found an fMRI correlate of the antiphase property of brightness induction in early human visual area V2. Furthermore, this correlate predicted the strength of induced surface brightness modulations in a physically constant probing region. Our data converge with findings from single-cell recording and hemodynamic measurement studies in animals that demonstrated a correlate of brightness induction in V2 (Hung et al., 2007; Roe et al., 2005; Hung, Ramsden, Chen, & Roe, 2001) and suggest that brightness induction reveals a basic property of surface brightness perception related to other types of contextual visual surface perception (Komatsu, 2006; Pessoa et al., 1998). Furthermore, we demonstrated that our results are unlikely to reflect unspecific spread of activity generated by contrast changes at the border between inducers and probing region. Instead our data suggest the existence of a small signal related to the perception of surfaces, in addition to a large signal related to the encoding of boundaries. Our results are unlikely to be associated with variations in pupil dilation as a result of luminance manipulations. Previous studies did not find differences in the retinotopic distribution of visual cortical activity after changes in pupil dilation (Haynes et al., 2004; Rooney & Cooper, 1988).

The finding of a correlate of brightness induction that was stronger in V2 than in V1 is in agreement with a number of other studies that have used other paradigms to manipulate brightness of neighboring stimulus surfaces of identical luminance. For example, several brightness perception studies used a static brightness induction display (COC stimulus). Neuronal recording and optical imaging experiments in monkeys that were presented with a COC display demonstrated that the activity of V2 interstripe neurons reflected the illusory differences in COC-induced surface brightness, whereas V1 blob neurons did not (Hung et al., 2007; Roe et al., 2005). A recent human fMRI study confirmed that extrastriate cortex may contribute more to surface brightness perception in COC stimuli than V1 (Perna, Tosetti, Montanaro, & Morrone, 2005). Furthermore, neurophysiological studies of texture filling-in in a Troxler fading paradigm also demonstrated contributions of extrastriate, but not striate, areas (De Weerd, Gattass, Desimone, & Ungerleider, 1995). The stronger contribution of V2 compared with V1 also has been described in the domain of contour perception (Ramsden, Hung, & Roe, 2001; Peterhans & von der Heydt, 1989; von der Heydt & Peterhans, 1989; von der Heydt, Peterhans, & Baumgartner, 1984). The evidence suggests that V2 contributes importantly to contextual processes that contribute to contour and surface perception.

However, several animal neurophysiological studies (Kinoshita & Komatsu, 2001; Rossi & Paradiso, 1999; Rossi et al., 1996) have shown a contribution of V1 to brightness induction. In addition, a few recent fMRI studies showed a brightness induction correlate in human V1 using a dynamic (Pereverzeva & Murray, 2008) as well as a static brightness induction stimulus (Boyaci, Fang, Murray, & Kersten, 2007). Our results do not exclude the possibility that V1 may also contribute to the perception of brightness (cf. Peters, Jans, van de Ven, De Weerd, & Goebel, 2010; Roe et al., 2005; Ramsden et al., 2001), although this effect may be smaller compared with V2, and our paradigm or stimulus may not have been sensitive enough to pick up the effect in V1. Some fMRI studies, in fact, did not find a brightness correlate in neither V1 nor extrastriate cortex using brightness induction (Cornelissen et al., 2006; Perna et al., 2005) as experimental paradigms.

Experimental design and its interaction with known limitations in temporal and spatial resolution (Bandettini, 2002; Menon & Kim, 1999; Kim, Richter, & Ugurbil, 1997) may be a factor that explains the different results in different studies of brightness induction. In the fMRI studies by Pereverzeva and Murray (2008) and Cornelissen et al. (2006), changes in inducer luminance were presented at a relatively high temporal rate of 1 Hz. Although this presentation rate could be resolved in neurophysiological measurements, it may have reduced the sensitivity to pick up small stimulus-surface-related signals in the presence of large boundary-related signals (Heeger & Ress, 2002; Logothetis et al., 2001; Boynton et al., 1996). The problem of lack of sensitivity might have been worsened by putative spatial or lateral spread of signals, which can be expected to be in counterphase. Nevertheless, it is difficult to determine why, despite very similar designs, the study from Pereverzeva and Murray (2008) and Cornelissen and colleagues (2006) led to different conclusions.

Furthermore, spatial resolution limitations of fMRI likely have determined in important ways the fMRI correlate of brightness induction we have observed. Estimates of the spatial spread of the fMRI signal vary from 2 to 7 mm HWHM (Cornelissen et al., 2006; Engel et al., 1997; Sereno et al., 1995). Our own data are rather in agreement with the estimates from Cornelissen et al. (2006). Several factors may contribute to the fMRI signal spread, including RF sizes in the cortical area under study (Dumoulin & Wandell, 2008; Smith, Singh, Williams, & Greenlee, 2001), neuronal signal spread because of lateral connectivity and feedback loops (Angelucci et al., 2002; Levitt & Lund, 2002; Stettler, Das, Bennett, & Gilbert, 2002; Levitt, Yoshioka, & Lund, 1994), as well as hemodynamic factors (Buxton, Wong, & Frank, 1998; Malonek & Grinvald, 1996). Because of this, fMRI signals generated in close spatial proximity will spread into each other's territory. Therefore, in our fMRI study, antiphase activity associated with brightness induction not only competes with in-phase activity related to luminance modulations of inducers but also with signals that encode the boundary of the probing region. This mix of signals related to physical aspects of the inducers is unrelated to the fMRI signal associated with brightness induction and could contribute to the masking of the signal related to brightness induction. If this were true, then decreased fMRI point spread should be associated with increased magnitude of antiphase activity in the probing region during brightness induction. We tested this hypothesis by computing in each participant the point spread of signals (in mm HWHM) from the inner and outer inducers into the probing region in the control condition and comparing these with the beta coefficients in the probing region during brightness induction. We found that larger brightness induction responses (i.e., negative values farther away from 0) correlated with smaller signal spread from the inducers into the cortical representation of the probing region (r = 0.59, p = .042). Another sign of the interaction between spread and induction is the finding that in-phase responses to inducer luminance modulations overall were smaller in the induction condition than in the control condition. This may be because of the antiphase activity present in the probing region during induction spreading into the territory of inducers, thereby attenuating in phase activity from inducers.

The precise neural circuitry that underlies brightness induction is currently unknown. Brightness induction may result from contextual interactions among neurons that are sensitive to physically homogenous stimulation, localized in V1 interblobs and V2 interstripes (Goebel & De Weerd, 2009; Hung et al., 2007; Roe et al., 2005), under the assumption that these interactions only take place among surfaces that sufficiently drive stimulated neurons (i.e., not when one of the surfaces is black). Alternatively, or in addition, contrast-polarity-sensitive neurons (Zhou, Friedman, & von der Heydt, 2000) may contribute to induction or to filling-in effects, in which a signal at boundaries is interpolated into surfaces (Grossberg & Hong, 2006; Grossberg, 2003; Neumann, Pessoa, & Hansen, 2001). In addition, it has been argued that linear filtering explains a significant portion or all of the induction effects. For example, Blakeslee and McCourt (2004) have suggested that many brightness effects can be explained by assuming a multiscale spatial filtering in the human visual system combined with contrast normalization. In this proposal, neurons in V1 interblobs and V2 interstripes might play a role in low spatial frequency filtering rather than being a substrate for spreading of surface information. This is related to Cornelissen et al.'s (2006) suggestion that surface-related information is derived from spatial filtering, without the need for spreading activity. However, multiscale filtering does not explain the full spectrum of known brightness illusions. For example, the static COC illusion cannot be explained only by multiscale filtering without also assuming a spreading process (Todorovic, 1987), and the same holds for a number of other visual surface illusions (Grossberg, Kuhlmann, & Mingolla, 2007; Grossberg & Hong, 2006; Grossberg & Yazdanbakhsh, 2005). In general, computational models using multiscale filtering have not been able to explain the same range of brightness percepts as models that do include an explicit surface filling-in mechanisms (Grossberg & Hong, 2006). In a recent large-scale computational model, we predicted our dynamic brightness illusion effect in early visual cortex using principles of lateral connectivity as a means for spreading of surface information across the cortex (Peters et al., 2010).

Although participants were required to carefully fixate, which by itself may be an attention-demanding task (Martinez-Conde, Macknik, Troncoso, & Dyar, 2006), the possibility exists that at least some attention was covertly allocated to the peripherally presented stimuli, which might have influenced the described effects. However, neural signals related to surface perception have been observed in prior studies in the absence of attention (Weil, Watkins, & Rees, 2008; Boyaci et al., 2007; Meng, Remus, & Tong, 2005) and even under anesthesia (Roe et al., 2005; Rossi & Paradiso, 1999). Therefore, we consider it unlikely that the reported data reflect predominantly attention effects.

In general, research into the neural correlates of surface perception has yielded conflicting data and views. For example, a neurophysiological study on surface texture in monkey visual cortex provided evidence for surface-related signals in V2 and V3 (De Weerd et al., 1995), but a recent fMRI study did not confirm that evidence (Weil et al., 2008). An fMRI study on color filling-in reported a correlate in human V1 (Sasaki & Watanabe, 2004), but a neurophysiological study in monkey V1 and V2 did not (Friedman, Zhou, & von der Heydt, 2003). Similar differences between studies remain in the domain of surface brightness perception, and further research is required to resolve them. The present study, however, supports the idea that there is a signal related to surface brightness in extrastriate cortex, although much more work is required to understand the mechanisms underlying that signal.

Acknowledgments

We thank Michael Capalbo and Francesco Gentile for acquiring part of the retinotopic measurements, Sven Gijsen for providing technical assistance, and Jochen Weber for providing software assistance. This project was financially supported by grants from the Netherlands Organization of Scientific Research (NWO) to V. V. (grant 451-07-014) and P. D. W. (grants 453-04-002 and 400-04-036).

Reprint requests should be sent to Vincent van de Ven, Department of Cognitive Neuroscience, Faculty of Psychology and Neuroscience, Maastricht University, P.O. Box 616, 6200 MD Maastricht, the Netherlands, or via e-mail: v.vandeven@maastrichtuniversity.nl.

REFERENCES

REFERENCES
Angelucci
,
A.
,
Levitt
,
J. B.
,
Walton
,
E. J.
,
Hupe
,
J. M.
,
Bullier
,
J.
, &
Lund
,
J. S.
(
2002
).
Circuits for local and global signal integration in primary visual cortex.
Journal of Neuroscience
,
22
,
8633
8646
.
Bandettini
,
P.
(
2002
).
fMRI: The spatial, temporal, and interpretive limits of functional MRI.
In K. Davis, D. Charney, J. Coyle, & C. Nemeroff (Eds.),
Neuropsychopharmacology: The fifth generation of progress: An official publication of the American College of Neuropsychopharmacology
(pp.
343
356
).
Philadelphia, PA
:
Lippincott Williams & Wilkins
.
Blakeslee
,
B.
, &
McCourt
,
M. E.
(
2004
).
A unified theory of brightness contrast and assimilation incorporating oriented multiscale spatial filtering and contrast normalization.
Vision Research
,
44
,
2483
2503
.
Boyaci
,
H.
,
Fang
,
F.
,
Murray
,
S. O.
, &
Kersten
,
D.
(
2007
).
Responses to lightness variations in early human visual cortex.
Current Biology
,
17
,
989
993
.
Boynton
,
G. M.
,
Engel
,
S. A.
,
Glover
,
G. H.
, &
Heeger
,
D. J.
(
1996
).
Linear systems analysis of functional magnetic resonance imaging in human V1.
Journal of Neuroscience
,
16
,
4207
4221
.
Buxton
,
R. B.
,
Wong
,
E. C.
, &
Frank
,
L. R.
(
1998
).
Dynamics of blood flow and oxygenation changes during brain activation: The balloon model.
Magnetic Resonance in Medicine
,
39
,
855
864
.
Cornelissen
,
F. W.
,
Wade
,
A. R.
,
Vladusich
,
T.
,
Dougherty
,
R. F.
, &
Wandell
,
B. A.
(
2006
).
No functional magnetic resonance imaging evidence for brightness and color filling-in in early human visual cortex.
Journal of Neuroscience
,
26
,
3634
3641
.
De Weerd
,
P.
,
Gattass
,
R.
,
Desimone
,
R.
, &
Ungerleider
,
L. G.
(
1995
).
Responses of cells in monkey visual cortex during perceptual filling-in of an artificial scotoma.
Nature
,
377
,
731
734
.
De Weerd
,
P.
,
Karni
,
A.
,
Kastner
,
S.
,
Ungerleider
,
L. G.
, &
Jezzard
,
P.
(
1997
).
An investigation of fMRI resolution in visual cortex. Presented at the Third International Conference on Functional Mapping of the Human Brain.
Neuroimage
,
5
,
S45
.
Dumoulin
,
S. O.
, &
Wandell
,
B. A.
(
2008
).
Population receptive field estimates in human visual cortex.
Neuroimage
,
39
,
647
660
.
Engel
,
S. A.
,
Glover
,
G. H.
, &
Wandell
,
B. A.
(
1997
).
Retinotopic organization in human visual cortex and the spatial precision of functional MRI.
Cerebral Cortex
,
7
,
181
192
.
Forshed
,
J.
,
Schuppe-Koistinen
,
I.
, &
Jacobsson
,
S. P.
(
2003
).
Peak alignment of NMR signals by means of a genetic algorithm.
Analytica Chimica Acta
,
487
,
189
199
.
Friedman
,
H. S.
,
Zhou
,
H.
, &
von der Heydt
,
R.
(
2003
).
The coding of uniform colour figures in monkey visual cortex.
Journal of Physiology
,
548
,
593
613
.
Goebel
,
R.
, &
De Weerd
,
P.
(
2009
).
Perceptual filling-in: From experimental data to neural network modeling.
In M. Gazzaniga (Ed.),
The cognitive neurosciences
(pp.
435
454
).
Cambridge, MA
:
MIT Press
.
Goebel
,
R.
,
Esposito
,
F.
, &
Formisano
,
E.
(
2006
).
Analysis of functional image analysis contest (FIAC) data with brainvoyager QX: From single-subject to cortically aligned group general linear model analysis and self-organizing group independent component analysis.
Human Brain Mapping
,
27
,
392
401
.
Grossberg
,
S.
(
2003
).
Filling-in the forms: Surface and boundary interactions in visual cortex.
In P. De Weerd & L. Pessoa (Eds.),
Filling-in: From perceptual completion to cortical reorganization
(pp.
13
37
).
Oxford
:
Oxford University Press
.
Grossberg
,
S.
, &
Hong
,
S.
(
2006
).
A neural model of surface perception: Lightness, anchoring, and filling-in.
Spatial Vision
,
19
,
263
321
.
Grossberg
,
S.
,
Kuhlmann
,
L.
, &
Mingolla
,
E.
(
2007
).
A neural model of 3D shape-from-texture: Multiple-scale filtering, boundary grouping, and surface filling-in.
Vision Research
,
47
,
634
672
.
Grossberg
,
S.
, &
Yazdanbakhsh
,
A.
(
2005
).
Laminar cortical dynamics of 3D surface perception: Stratification, transparency, and neon color spreading.
Vision Research
,
45
,
1725
1743
.
Haynes
,
J. D.
,
Lotto
,
R. B.
, &
Rees
,
G.
(
2004
).
Responses of human visual cortex to uniform surfaces.
Proceedings of the National Academy of Sciences, U.S.A.
,
101
,
4286
4291
.
Heeger
,
D. J.
, &
Ress
,
D.
(
2002
).
What does fMRI tell us about neuronal activity?
Nature Reviews Neuroscience
,
3
,
142
151
.
Hubel
,
D. H.
, &
Wiesel
,
T. N.
(
1962
).
Receptive fields, binocular interaction and functional architecture in the cat's visual cortex.
Journal of Physiology
,
160
,
106
154
.
Hung
,
C. P.
,
Ramsden
,
B. M.
,
Chen
,
L. M.
, &
Roe
,
A. W.
(
2001
).
Building surfaces from borders in Areas 17 and 18 of the cat.
Vision Research
,
41
,
1389
1407
.
Hung
,
C. P.
,
Ramsden
,
B. M.
, &
Roe
,
A. W.
(
2007
).
A functional circuitry for edge-induced brightness perception.
Nature Neuroscience
,
10
,
1185
1190
.
Jans
,
B.
,
Been
,
M.
,
van de Ven
,
V.
,
Goebel
,
R.
, &
De Weerd
,
P.
(
submitted
).
Pattern-based alignment: A new tool for functional alignment of topographic maps of fMRI activity.
Kim
,
S. G.
,
Richter
,
W.
, &
Ugurbil
,
K.
(
1997
).
Limitations of temporal resolution in functional MRI.
Magnetic Resonance in Medicine
,
37
,
631
636
.
Kinoshita
,
M.
, &
Komatsu
,
H.
(
2001
).
Neural representation of the luminance and brightness of a uniform surface in the macaque primary visual cortex.
Journal of Neurophysiology
,
86
,
2559
2570
.
Komatsu
,
H.
(
2006
).
The neural mechanisms of perceptual filling-in.
Nature Reviews Neuroscience
,
7
,
220
231
.
Kriegeskorte
,
N.
, &
Goebel
,
R.
(
2001
).
An efficient algorithm for topologically correct segmentation of the cortical sheet in anatomical MR volumes.
Neuroimage
,
14
,
329
346
.
Levitt
,
J. B.
, &
Lund
,
J. S.
(
2002
).
The spatial extent over which neurons in macaque striate cortex pool visual signals.
Visual Neuroscience
,
19
,
439
452
.
Levitt
,
J. B.
,
Yoshioka
,
T.
, &
Lund
,
J. S.
(
1994
).
Intrinsic cortical connections in macaque visual area V2: Evidence for interaction between different functional streams.
Journal of Comparative Neurology
,
342
,
551
570
.
Linden
,
D. E.
,
Kallenbach
,
U.
,
Heinecke
,
A.
,
Singer
,
W.
, &
Goebel
,
R.
(
1999
).
The myth of upright vision. A psychophysical and functional imaging study of adaptation to inverting spectacles.
Perception
,
28
,
469
481
.
Listgarten
,
J.
,
Neal
,
R. M.
,
Roweis
,
S. T.
, &
Emili
,
A.
(
2005
).
Multiple alignment of continuous time series. In
Advances in neural information processing systems
(
Vol. 17
, pp.
817
824
).
Cambridge, MA
:
MIT Press
.
Logothetis
,
N. K.
,
Pauls
,
J.
,
Augath
,
M.
,
Trinath
,
T.
, &
Oeltermann
,
A.
(
2001
).
Neurophysiological investigation of the basis of the fMRI signal.
Nature
,
412
,
150
157
.
MacEvoy
,
S. P.
,
Kim
,
W.
, &
Paradiso
,
M. A.
(
1998
).
Integration of surface information in primary visual cortex.
Nature Neuroscience
,
1
,
616
620
.
Malonek
,
D.
, &
Grinvald
,
A.
(
1996
).
Interactions between electrical activity and cortical microcirculation revealed by imaging spectroscopy: Implications for functional brain mapping.
Science
,
272
,
551
554
.
Martinez-Conde
,
S.
,
Macknik
,
S. L.
,
Troncoso
,
X. G.
, &
Dyar
,
T. A.
(
2006
).
Microsaccades counteract visual fading during fixation.
Neuron
,
49
,
297
305
.
Meng
,
M.
,
Remus
,
D. A.
, &
Tong
,
F.
(
2005
).
Filling-in of visual phantoms in the human brain.
Nature Neuroscience
,
8
,
1248
1254
.
Menon
,
R. S.
, &
Kim
,
S. G.
(
1999
).
Spatial and temporal limits in cognitive neuroimaging with fMRI.
Trends in Cognitive Sciences
,
3
,
207
216
.
Neumann
,
H.
,
Pessoa
,
L.
, &
Hansen
,
T.
(
2001
).
Visual filling-in for computing perceptual surface properties.
Biological Cybernetics
,
85
,
355
369
.
Pereverzeva
,
M.
, &
Murray
,
S. O.
(
2008
).
Neural activity in human V1 correlates with dynamic lightness induction.
Journal of Vision
,
8
,
1
10
.
Perna
,
A.
,
Tosetti
,
M.
,
Montanaro
,
D.
, &
Morrone
,
M. C.
(
2005
).
Neuronal mechanisms for illusory brightness perception in humans.
Neuron
,
47
,
645
651
.
Pessoa
,
L.
,
Thompson
,
E.
, &
Noe
,
A.
(
1998
).
Finding out about filling-in: A guide to perceptual completion for visual science and the philosophy of perception.
Behavioral and Brain Sciences
,
21
,
723
748; discussion 748–802
.
Peterhans
,
E.
, &
von der Heydt
,
R.
(
1989
).
Mechanisms of contour perception in monkey visual cortex: II. Contours bridging gaps.
Journal of Neuroscience
,
9
,
1749
1763
.
Peters
,
J. C.
,
Jans
,
B.
,
van de Ven
,
V.
,
De Weerd
,
P.
, &
Goebel
,
R.
(
2010
).
Dynamic brightness induction in V1: Analyzing simulated and empirically acquired fMRI data in a “common brain space” framework.
Neuroimage
,
52
,
973
984
.
Ramsden
,
B. M.
,
Hung
,
C. P.
, &
Roe
,
A. W.
(
2001
).
Real and illusory contour processing in area V1 of the primate: A cortical balancing act.
Cerebral Cortex
,
11
,
648
665
.
Roe
,
A. W.
,
Lu
,
H. D.
, &
Hung
,
C. P.
(
2005
).
Cortical processing of a brightness illusion.
Proceedings of the National Academy of Sciences, U.S.A.
,
102
,
3869
3874
.
Rooney
,
B. J.
, &
Cooper
,
R. M.
(
1988
).
2-Deoxyglucose uptake in the rat visual system following pupil dilation with atropine.
Vision Research
,
28
,
259
262
.
Rossi
,
A. F.
, &
Paradiso
,
M. A.
(
1999
).
Neural correlates of perceived brightness in the retina, lateral geniculate nucleus, and striate cortex.
Journal of Neuroscience
,
19
,
6145
6156
.
Rossi
,
A. F.
,
Rittenhouse
,
C. D.
, &
Paradiso
,
M. A.
(
1996
).
The representation of brightness in primary visual cortex.
Science
,
273
,
1104
1107
.
Sasaki
,
Y.
, &
Watanabe
,
T.
(
2004
).
The primary visual cortex fills in color.
Proceedings of the National Academy of Sciences, U.S.A.
,
101
,
18251
18256
.
Sereno
,
M. I.
,
Dale
,
A. M.
,
Reppas
,
J. B.
,
Kwong
,
K. K.
,
Belliveau
,
J. W.
,
Brady
,
T. J.
,
et al
(
1995
).
Borders of multiple visual areas in humans revealed by functional magnetic resonance imaging.
Science
,
268
,
889
893
.
Shmuel
,
A.
,
Augath
,
M.
,
Oeltermann
,
A.
, &
Logothetis
,
N. K.
(
2006
).
Negative functional MRI response correlates with decreases in neuronal activity in monkey visual area V1.
Nature Neuroscience
,
9
,
569
577
.
Smith
,
A. T.
,
Singh
,
K. D.
,
Williams
,
A. L.
, &
Greenlee
,
M. W.
(
2001
).
Estimating receptive field size from fMRI data in human striate and extrastriate visual cortex.
Cerebral Cortex
,
11
,
1182
1190
.
Stettler
,
D. D.
,
Das
,
A.
,
Bennett
,
J.
, &
Gilbert
,
C. D.
(
2002
).
Lateral connectivity and contextual interactions in macaque primary visual cortex.
Neuron
,
36
,
739
750
.
Talairach
,
J.
, &
Tournoux
,
P.
(
1988
).
Co-planar stereotaxic atlas of the human brain.
New York
:
Thieme
.
Todorovic
,
D.
(
1987
).
The Craik–O'Brien–Cornsweet effect: New varieties and their theoretical implications.
Perception & Psychophysics
,
42
,
545
560
.
von der Heydt
,
R.
, &
Peterhans
,
E.
(
1989
).
Mechanisms of contour perception in monkey visual cortex. I. Lines of pattern discontinuity.
Journal of Neuroscience
,
9
,
1731
1748
.
von der Heydt
,
R.
,
Peterhans
,
E.
, &
Baumgartner
,
G.
(
1984
).
Illusory contours and cortical neuron responses.
Science
,
224
,
1260
1262
.
Weil
,
R. S.
,
Watkins
,
S.
, &
Rees
,
G.
(
2008
).
Neural correlates of perceptual completion of an artificial scotoma in human visual cortex measured using functional MRI.
Neuroimage
,
42
,
1519
1528
.
Worsley
,
K. J.
, &
Friston
,
K. J.
(
1995
).
Analysis of fMRI time-series revisited—Again.
Neuroimage
,
2
,
173
181
.
Zhou
,
H.
,
Friedman
,
H. S.
, &
von der Heydt
,
R.
(
2000
).
Coding of border ownership in monkey visual cortex.
Journal of Neuroscience
,
20
,
6594
6611
.

Author notes

*

These authors contributed equally to the study.