Tomoki J Takahashi^1,2 (), Masamichi J Hayashi^1,2; ¹Graduate School of Frontier Biosciences, Osaka University, Suita, Japan, ²Center for Information and Neural Networks (CiNet), Advanced ICT Research Institute, National Institute of Information and Communications Technology, Suita, Japan

Humans and animals have the ability to estimate approximate numerosity without explicitly counting individual elements. One psychophysical evidence supporting the existence of numerosity-tuned neural mechanisms is the susceptibility of numerosity perception to adaptation. Although conventional studies often use dots of opposite luminance polarity (e.g., half black and half white, presented on a gray background) to control the overall luminance of the display, it remains unclear whether numerosity perception is bound to such low-level visual features. To address this question, we conducted a series of numerosity adaptation experiments. In our first experiment, observers were adapted to either 20 or 80 dots and responded whether the subsequent test numerosity, ranging from 24 to 72 dots, was larger or smaller than the reference numerosity of 40 dots. The congruence of luminance polarity between the adapting and test dot arrays was manipulated. The results showed significant negative aftereffects in both congruent and incongruent conditions, although weaker in the incongruent condition. To minimize the possible effect of retinal afterimages, our second experiment introduced white noise masking stimuli after the adaptation and test stimuli, yielding results consistent with the first experiment. Subjective reports of the frequency and intensity of afterimages did not correlate with the individual differences in the aftereffect size, confirming the little influence of the afterimage. Finally, we found that randomly switching the luminance polarity of the test stimuli across trials did not eliminate aftereffects, suggesting that the predictability of luminance polarity is not relevant for the presence of aftereffects in the incongruent condition. Taken together, our results show that the negative aftereffects in numerosity adaptation are robust to the changes in luminance polarity. Nevertheless, aftereffect magnitude is still affected by the changes in luminance polarity, suggesting that numerosity perception may not be completely independent of the processing of such low-level visual features.