Abstract
Visual perceptual learning (VPL) refers to long-term performance improvement as a result of visual experience (Sagi, 2011; Watanabe & Sasaki, 2015). In almost all VPL studies, artificial stimuli (AS) have been used. Whether VPL of natural scenes (NS) and/or everyday visual environment follows the same rules as VPL of AS has never been systematically investigated. A series of studies have found that task-irrelevant VPL of supra-threshold AS does not occur because attentional systems detect and suppress the supra-threshold task-irrelevant feature signals (Tsushima et al, 2006, 2008; Chang et al, 2014). Here, we tested whether task-irrelevant VPL occurs with supra-threshold NS. There were 2 conditions, NS and AS conditions (N=8 each). During a 10-day exposure stage, participants in the NS condition were repeatedly exposed to a set of supra-threshold task-irrelevant NS in which a particular orientation (trained orientation) was dominant. In the AS condition, participants were exposed to supra-threshold task-irrelevant AS that consisted of the identical mean luminance, orientation and spatial frequency distributions to those of the NS. The participants were asked to conduct a difficult rapid serial visual presentation (RSVP) task throughout the exposure stage. Before and after the exposure stage, participants’ performance on a discrimination task was measured on both the trained orientation and untrained orientation (90 deg rotated from the trained) using a Gabor patch. We found a significant performance improvement for the trained orientation, but not for the untrained orientation, after the exposure stage in the NS condition. In contrast, no significant performance improvement was observed for either orientation in the AS condition. These results indicate that the mechanisms of VPL of NS in everyday life may not be the same as VPL of AS. If so, the framework of VPL based on these 30 years of research would have to be greatly modified.