Abstract
In continuous flash suppression (CFS), a rapid, dynamic sequence of Mondrian images presented to one eye can suppress a static target in the other eye for many seconds. Because of its robustness, CFS has become popular for studying unconscious perception, despite limited understanding of its underlying mechanisms. Recent evidence implicates significant contributions from low-level properties (e.g., orientation), but might higher-order influences also impact CFS's potency? For example, the random assortment of shapes in successive Mondrian updates generates pattern information uncertainty that could strengthen suppression exerted on the target. Here we examine the effect of spatial and temporal pattern predictability on CFS potency. Temporally, predictable information entailed updating the Mondrian every 100 ms or sinusoidally modulating pixel luminance at 2 Hz. Irregular update rates and stochastic pixel luminance changes produced temporal uncertainty. Spatially, we reduced predictability by updating the Mondrian with different spatial patterns or we maintained spatial predictability by presenting the same pattern over time, modulating only its luminance contrast. To quantify the effectiveness of these maskers, we had participants track the visibility of a target (contrast modulated concentric grating) over the course of a minute. We also pitted various pairs of discriminable CFS maskers against one another in rivalry. Our results showed that, although the unpredictable spatial masker with fixed 10 Hz change rate dominated the rivalry competition, stronger suppression for spatial uncertainty was only obtained when the target modulated at 4 Hz. We found no effect of spatial and temporal predictability when the target contrast modulated at 0.125 Hz. Since the Mondrian temporal frequency spectrum is characteristically 1/f and hence less compatible with the 4 Hz target, our results suggest that in practice, information unpredictability primarily enhances suppression of a more weakly suppressed target.
Meeting abstract presented at VSS 2018