When The Weaker Conquer: A Contrast-Based Illusion of Visual Numerosity and Its Dependence on Segregation

Humans and many non-human species alike are endowed with a capability to perceive the approximate numerosity of a large set. In the visual domain, various factors have been found to affect perceived numerosity. These include continuous extent variables (i.e., length, area, density, etc.), spatial structure of the stimulus (e.g. clustering) and visual contexts. In the current study, the effect of luminance contrast on approximate numerosity perception was investigated.;A numerosity illusion was found in which low-contrast elements appear to be more numerous than high-contrast ones when they were intermixed in the same display. Using a numerosity discrimination task, the illusion, with an average magnitude of ∼10%, was observed with either positive (white and light gray disks on a dark field) or negative contrast (black and dark gray disks on a light field). The illusion was eliminated when gray and white stimuli were segregated by visual field, motion (whites moved left when grays moved right, and vice-versa), orientation (using white and gray vertical and horizontal bars) or stereoscopic depth. Therefore, the illusion might be due to a failure in segregation between the two sets of elements.;When matching the numerosity of gray or white disks in an intermixed patch to the numerosity of an homogenous patch (gray or white disks unmixed), the numerosity of white disks in the intermixed patch was underestimated relative to the homogenous patch, but the numerosity of gray disks was veridically matched, evidence of an asymmetrical interaction between the two disk sets in an intermixed patch, such that the stronger (white) elements appeared less numerous. The same results were obtained when the irrelevant (not-to-be-matched) set of disks in the intermixed patch was replaced by visual noise of the same contrast, refuting explanations of the illusion in terms of misclassification (whites being seen as grays more often than the reverse) or occlusion (whites hiding potential grays).;Lastly, it was also shown that the numerosity illusion was not specific to the numerosity discrimination task but also occurred for absolute numerosity estimation. The reported numerosity of gray disks in an intermixed patch was consistently higher than that of white ones, regardless of how subjects allocated their attention between the two disk sets, which points to a pre-attentive locus of the effect.;To conclude, when unsegregated, an asymmetrical sensory/perceptual interaction occurs automatically between low-contrast and high-contrast elements, resulting in an illusory reduction in the numerosity of high-contrast ones. To account for the illusion, a thresholding process that operates at an early stage of visual processing to separate the two sets of disks for enumeration is hypothesized.