Ryo Kanechi^1,2 (), Neil W. Roach³, Masamichi J. Hayashi^1,2; ¹Center for Information and Neural Networks (CiNet), Advanced ICT Research Institute, National Institute of Information and Communications Technology, ²Graduate School of Frontier Biosciences, The University of Osaka, ³School of Psychology, University of Nottingham

Two stimuli separated by a short temporal interval are perceived as integrated, whereas those separated by a long interval are perceived as segregated. Temporal integration and segregation are thought to play important roles in perceptual stability and change detection, respectively. However, it remains unclear whether these processes are regulated by a single or distinct neural mechanism(s). Here, we characterize and compare the spatial tuning of sensitivity for temporal segregation and integration. We presented a pair of 4 x 4 multi-element panels sequentially, with a variable temporal interval ranging from 30 to 150 ms. Participants performed either an odd-element task (OET) or a missing-element task (MET) to measure temporal segregation and integration performance, respectively. Participants were asked to maintain gaze at a fixation point and identify the location where either an incomplete element appeared (OET) or where no element appeared across the panels (MET). To dissociate the retinal co-ordinates of the critical element from its position in the stimulus array, the fixation point either remained fixed at the center (Experiment 1) or varied between four quadrant locations on a trial-by-trial basis (Experiment 2). The results consistently showed that sensitivity (d’) for both tasks declined as the retinal eccentricity of the critical element increased, but the drop-off was significantly more pronounced for the integration task. These results may reflect a greater dependence of temporal integration on the parvocellular pathway, within which cell density decreases steeply with eccentricity. In contrast, temporal segregation may be more dependent on the magnocellular pathway, which is characterized by more stable cell density across eccentricity. Overall, these findings indicate that temporal integration and segregation may be governed, at least in part, by distinct neural mechanisms.