Numerosity adaptation resists filtering: Insights from an illusory contour paradigm

The mechanisms underlying numerosity perception remain debated, with theories proposing either a dedicated system for segmented items or reliance on low-level features such as spatial frequency (SF) and texture-density. Numerosity adaptation—where exposure to an array of objects shifts the perceived numerosity of a subsequently presented array—has been considered strong evidence for a dedicated numerosity-processing system. However, a recent hypothesis suggests that this effect results from filtering previously processed information rather than genuine numerosity adaptation. To clarify the visual features driving this phenomenon, we employed a novel adaptation paradigm in which participants were exposed to arrays of illusory-contours (ICs) items before performing a comparison task on arrays of real dots. ICs arrays, generated using the Ehrenstein-illusion, allowed us to disentangle numerosity adaptation from texture-density/SF or filtering effects. Because ICs-numerosity in the Ehrenstein grid is negatively correlated with texture-density/SF (e.g., fewer ICs cause greater power in low-SFs), the texture-density hypothesis predicts a reduction in the point of subjective equality (PSE) after adaptation, whereas the filtering hypothesis predicts no effect. Our results showed a significant increase in PSE following adaptation, suggesting that adaptation was driven by the number of ICs in the adapting arrays rather than by SF. Moreover, the strong difference between control and adaptation conditions indicates that adaptation was not merely due to filtering. In sum, our findings demonstrate that numerosity adaptation is driven by item quantity, irrespective of stimulus type, providing further support for a numerosity-selective mechanism and challenging theories based solely on low-level visual features or filtering.