Feature selective adaptation of numerosity perception

ABSTRACT Perceptual adaptation has been used to infer the existence of numerosity detectors, which allow humans to quickly estimate the number of objects in a scene. While adaptation was originally thought to affect numerosity perception regardless of the low-level features of the items, a recent study demonstrated that adaptation is more pronounced when the adapting and adapted (test) stimuli share the same color, compared to when they were colored differently. In this study we explored whether such adaptation reduction depends on a novelty effect induced by changes in stimulus features or whether this effect is observed only when implying an identity change of the stimuli. To this aim, we performed six experiments in which numerosity adaptation was investigated in conditions in which adapting and adapted stimuli were either matched or differed for several low-level (color, luminance, shape, and motion) or high-level (letters’ identity, face emotions) features. Numerosity adaptation was consistently observed across all conditions, but it was reduced when adaptor and test differed in color, luminance and shape. However, when stimuli differed in their motion profile, a very salient perceptual change that does not imply a change in items’ identity, adaptation selectivity vanished. Moreover, adaptation selectivity was not observed when items’ identity was changed by spatial rotations of the same stimulus (letters) or when stimuli were matched for the global configuration (face outline) but differed for the arrangement of local features (mouth, nose, eyes). Interestingly, image dissimilarity between test and adaptor, as quantified by Gabor filters simulating a simplified model of the primary visual cortex, nicely predicted the strength of numerosity adaptation across all conditions. Overall, changes in stimulus identity defined by low-level features, rather than novelty in general, determined the strength of the adaptation effects, provided that the changes were readily noticeable. Our findings suggest that numerosity mechanisms may be able to operate on segregated and categorized visual items in addition to the total quantity of the set, with part of the aftereffects induced by numerosity adaptation occurring after feature-binding.