Abstract
Several studies have demonstrated that numerosity perception is susceptible to adaptation: after exposure to high numerosities, participants underestimate the numerosity of subsequently patterns, and over-estimate after adaptation to low numerosities. We have recently shown that motor routines such as hand-tapping also affect numerosity estimates, and that the motor adaptation is confined to the tapping region, pointing to shared mechanisms encoding the quantity of both internally generated actions and externally generated events. However, it has been suggested that adaptation studies may reflect response biases rather than alteration of sensory processing. To disentangle the two possibilities, we studied visual and motor adaptation on numerosity perception (determined by the point of subjective equality in a 2AFC discrimination task), while also measuring participant confidence and reaction-times. If adaptation occurs at the sensory level, peaks in reaction-times and troughs in confidence should shift with adaptation to follow the new PSE, where the stimuli are most difficult to discriminate (Gallagher et al., 2019). On the other hand, if they result from response biases when confidence is low, the confidence (and also reaction-time) curves should remain centered at the unadapted PSE after adaptation. We replicate previous studies showing that both sensory and motor adaptation robustly distort numerosity estimates by 23% and 14% respectively. Importantly, in both cases, the shifts in perceived numerosity were almost perfectly mirrored by shifts in confidence and reaction-times, with the maximum uncertainty and longest response-times occurring at the point of subjective equality rather than at the point of physical equality. Taken together our results suggest that both forms of numerosity adaptation do not arise by decisional processes but directly act on the sensory representation of numerosity