Onset rivalry: brief presentation isolates an early independent phase of perceptual competition

Abstract

When the left and right eyes are simultaneously presented with different images, observers typically report exclusive awareness of only one image. This phenomenon is termed binocular rivalry, reflecting the fact that the dominant image alternates every few seconds in a cycle of perceptual competition that continues indefinitely. Despite the apparent continuity in perceptual switching, we now demonstrate that the initial "onset" period is fundamentally different to all subsequent rivalry epochs. Using brief intermittent presentations, rivalry dominance shows strong biases such that the same target is perceived with each successive stimulus onset. These biases remain consistent within any given location, but vary across the visual field in a distribution that is stable over multiple weeks but highly idiosyncratic across observers. If the presentation exceeds approximately 1sec at any location, however, the very different and much more balanced alternations of sustained binocular rivalry become apparent. These powerful onset biases are observed with brief intermittent presentations at a single location or with continual smooth motion of the targets. Periods of adaptation to one of the rivaling targets induced local switches in dominance to the non-adapted target. However, these effects were generally limited to the spatial site of adaptation and had less influence over each subsequent cycle of the target. We conclude that onset rivalry is independent of sustained rivalry and cannot be explained by local regions of monocular dominance or memory of past perceptual history, but rather reflects low-level, spatially localized factors that are stable over periods of weeks. These findings suggest that brief presentation paradigms are inappropriate for their current use in studies of the mechanisms underlying sustained rivalry. However, brief presentations are ideal for investigating early stages of perceptual competition.

Figure 1. Illustration of stimulus and data presentation method.

(A) In experiment 1, binocular rivalry was induced with foveal presentation of red and green orthogonal gratings to the left or right eye respectively. The rivaling targets were red and green sinusoidal grating patches that subtended 2.5° of visual angle. The gratings had a spatial frequency of 1.6 cycles/deg oriented obliquely at either 45° or 135°. Targets were presented for either 60sec of continual presentation or for 60×1sec presentations over 600 sec (i.e. 1sec on 9sec off). In experiment 2, the same targets were presented in one of 8 locations spaced equally around a circular path with a radius of 4.5° around a 0.3° central fixation point. During the intermittent stimulus conditions targets were presented for 1sec with an inter-stimulus interval 250ms (all 8 locations visited every 10sec). In both conditions the targets were presented on a gray background within a 13.5° white square frame 0.2° thick. (B) Schematic illustration of results for an idealized case of random and strongly biased perceptual dominance (note that any switch in perception is only temporary, suggesting the dominance reflects an onset bias rather than stabilization). For foveal rivalry the data are graphically represented as a sequence of 60 adjacent colored bars corresponding (from left to right) with the perceptual dominance for each of the 60×1sec presentation (in the intermittent case the gap interval is not depicted). In experiment 2, perceptual dominance at each location is illustrated by a color patch corresponding to each of the 8 peripheral target locations. Each successive target loop is represented in increasingly outward rings. In this way, each color patch represents the subject's perceptual dominance at a single point in time and space for the entire trial. The left column represents idealized case for random allocation of dominance, while the right column shows complete localized biases.

Figure 2. Results from experiment 1 and 2.

(A) The pattern of perceptual dominance for two representative subjects during foveal presentation of rivalry targets. The left column depicts results for 60 intermittent 1sec presentations (9 sec inter-stimulus intervals). During this condition there was no evidence of perceptual stabilization. Subjects reported large perceptual biases that were broken by brief intermittent dominance of the alternative targets. In contrast, no biases in perceptual dominance were observed during 60sec of sustained target presentation. (B) Data from all 4 subjects tested with the peripheral presentation in experiment 2a–d. The left column shows results from the initial test of sequential (clockwise or counterclockwise) presentation. During the 2min trial duration the targets cycled through the same 8 locations 12 times (each wedge represents a single location in space, with time represented in radial distance from the center ring). In this condition, perceptual dominance shows strong localized biases. Similar biases were also observed when the order of presentation was randomized (columns 2^nd and 3^rd from the left respectively), however, the relative locations of the biases were not necessarily conserved. No systematic biases were observed during 60sec of sustained presentation at each of the 8 locations (4^th column). (C) A plot of the proportion of red dominance reported during intermittent (x-axis) and sustained target presentation (y-axis) for the foveal (black circles) and 8 peripheral target locations (color diamonds correspond to the 4 subjects tested in experiment 2). Despite very little bias in dominance during sustained viewing subjects reported large biases during intermittent viewing across all target locations.

Figure 3. Results from experiment 3, 4 and 5.

(A) Data from all 4 subjects tested with 3 different speeds of smooth motion of the target around a peripheral trajectory (the same radial distance used in experiment 2). The left column shows smooth motion at a cycle speed of one loop every 10sec, a rate equivalent to experiment 2a–c. Throughout the 2min trial strong localized biases were reported consistently across all 12 cycle loops. The 2^nd and 3^rd column show a clear decrease in the strength and location specificity of perceptual bias as the target speed slows down to 20 and 100sec per loop. (B) In experiment 4, the 1sec presentation (identical to experiment 2a–c) was used to determine whether the reduction in bias with slower speeds was related to increases in local presentation duration or the elapsed duration between successive presentations at a single location. Data from 2 (of 4) subjects illustrate that localized biases exist independent of the elapsed duration between successive presentations. Local biases are still observed if the inter-stimulus interval is extended to allow cycle rates of 20 and 100sec and at retest 2 weeks later. (C) 2 subjects were tested with 60sec of adaptation to the dominant target (red in both cases) at 4 locations (0° 90° 180° 270°) indicated with the red triangle at the center of the corresponding radial plots. After the adaptation period the adapting stimulus was removed and the rivalry targets were presented 45° preceding the adapting location. The rivalry target then cycled smoothly through the adaptation location for 2 min (12×10sec cycles). Together the plots show that adaptation to one of the rivalry targets will result in dominance of the non-adapted target, however, the spatial and temporal extent of this effect is limited and varies across subject and adaptation location.