The perception of even the most elementary features of the visual environment depends strongly on their spatial context. In the study reported here, we asked at what level of abstraction such effects require conscious processing of the context. We compared two visual illusions that alter subjective judgments of brightness: the simultaneous brightness contrast illusion, in which two circles of identical physical brightness appear different because of different surround luminance, and the Kanizsa triangle illusion, which occurs when the visual system extrapolates a surface without actual physical stimulation. We used a novel interocular masking technique that allowed us to selectively render only the context invisible. Simultaneous brightness contrast persisted even when the surround was masked from awareness. In contrast, participants did not experience illusory contours when the inducing context was masked. Our findings show that invisible context is resolvable by low-level processes involved in surface-brightness perception, but not by high-level processes that assign surface borders through perceptual completion. In the study reported here, we asked at what level of abstraction such effects require conscious processing of the context. We compared two visual illusions that alter subjective judgments of brightness: the simultaneous brightness contrast illusion, in which two circles of Such simultaneous brightness contrast (SBC) is observed, for example, when a gray circle appears brighter against a dark background than against a light background ( Fig. 1a ). Similarly, spatially distinct elements of a visual scene can induce an illusory percept of luminance contours if the organization of those elements induces the visual system to extrapolate the presence of an occluding surface ( Fig. 1b ), as in the Kanizsa triangle illusion ( Kanizsa, 1979 ). Fig. 1. Stimuli and masks used in (a) Experiment 1 and (b) Experiment 3. For each experiment, an example stimulus is shown on the left, and a continuous flash suppression (CFS) mask is shown on the right. Experiment 1 utilized the simultaneous brightness contrast illusion. The suppressed eye was shown two gray circles that were physically identical but appeared different in luminance because of the context. The CFS mask covered most of the illusion stimulus except for two “holes” (highlighted here by arrows) that coincided with the two gray circles. Experiment 3 utilized the Kanizsa triangle illusion. The suppressed eye was shown four inducer elements, three of which created the illusory impression of a triangle. Through a mirror stereoscope, participants viewed an SBC illusion or a Kanizsa triangle ( Fig. 1 , left) with one eye while viewing brightly colored flashing patterns ( Fig. 1 , right) with the other eye. These flashing patterns almost entirely overlapped the retinotopic location of the stimulus in the other eye, sparing only the locations of the illusory percept. This composite stimulus resulted in stable and selective suppression of the background context through continuous flash suppression (CFS; Tsuchiya & Koch, 2005 ), but left the critical portion of the stimuli phenomenally accessible to consciousness for behavioral report. We found that illusory perception decreased significantly, F (4, 20) = 4.435, p = .01, with increasing circle size ( Fig. 2b ), which indicates that mechanisms at the border are indeed important for this illusion. Experiment 3: Kanizsa Triangle Under Dichoptic Masking At what level of abstraction is consciousness necessary for visual processing of spatial context? In Experiment 3, we used our dichoptic masking procedure to test whether a presumably high-level process, the perception of illusory contours through amodal completion (a Kanizsa triangle), could also occur without contextual awareness. Procedure At the start of every trial, participants were presented with a 200-ms blank stimulus, in which only the tile frame, black square, and fixation dot were visible. Then, one eye was presented with the Kanizsa triangle stimulus for 200 ms, followed by 100 ms of the stimulus containing only the Gaussian luminance profiles without the black wedges. We chose a short stimulus duration to ensure that the CFS mask adequately masked the illusion stimulus. However, we observed comparable results in pilot experiments with longer stimulus durations. The three conditions, which are illustrated by the icons below the graph, all involved a Kanizsa triangle stimulus. In the visible condition, the stimulus was not masked. In the selective-mask and invisible conditions, a continuous flash suppression mask was presented in two different ways. Error bars indicate 1 SEM . Discussion In the series of experiments reported here, we combined a novel selective-masking technique with two contextual visual illusions that modulated subjective judgments of brightness. We examined whether conscious processing of spatial context was necessary for these illusions to occur. This may require an abstract concept of depth that may not arise without conscious analysis of the visual scene and is probably learned through experience with the environment ( Walk, 1966 ). In Experiment 3, we tested whether visible Kanizsa inducers were necessary for the perception of illusory contours. When not viewing stimuli through a CFS mask, participants found it easy to discriminate the shape of an illusory Kanizsa triangle; however, when the elements of the stimulus inducing the illusion were masked from awareness, behavioral performance dropped to chance levels—comparable to levels observed when the entire stimulus was masked—confirming that awareness of inducers was necessary for perception of illusory contours. Consistent with this assertion, findings of other studies have shown that patients deprived of visual experience from an early age often do not experience this illusion when their vision is restored in later life ( Fine et al., 2003 ).