Abstract
Visual perceptual learning (VPL) is defined as long-term improvement in visual performance as a result of visual experience. In many cases, VPL is largely confined to a feature that was presented during training, which refers to feature specificity of VPL. However, recent research has indicated that feature specificity can be abolished by a training-plus-exposure procedure. This finding led some researchers to suggest that VPL is governed by rule-based learning. However, the role of rule-based learning in VPL has not been empirically clearly examined. Since category learning is known to be a type of rule-based learning, the present study investigated whether and how category learning governs VPL by examining the effect of category learning on VPL. We trained 6 human subjects to classify gabor orientations within a 180° range into two categories. After one session of training of category learning (500 trials), participants underwent 5 days' training on a detection of one gabor orientation presented at the fovea. Before and after training on the orientation, participants were tested for the detectability of an orientation from the same category (SC) as the trained orientation, another orientation from a different category (DC), and also the trained orientation. The distance between the trained and SC orientations was the same as the distance between the trained and DC orientations. The results showed that detection performances on the SC orientation and the trained orientation were significantly greater after than before training, but not on the DC orientation. These results indicate that category learning makes VPL of a feature transferable to features within the same category as the trained feature. Based on the results, we concluded that rule-based learning can drive VPL. The intraparietal sulcus (IPS) has been implicated in rule-based learning and category learning. Thus, the IPS may play a role in generalization VPL.
Meeting abstract presented at VSS 2016