Abstract
Perceptual learning (PL), experience induced gains in discriminating sensory features, has been classically thought to be highly specific to trained retinal location of the stimuli. However, recent research from multiple labs has demonstrated that double training (Xiao et al., 2008) promotes dramatic generalization to untrained locations in paradigms previously considered to produce location-specific learning. While double training studies demonstrate that location specificity may have been misconstrued in some experimental contexts, there exists significant controversy regarding the extent to which double training will promote spatial generalization across different perceptual learning paradigms. To explore this issue, we examined spatial specificity in two different hyperacuity tasks. The first task was previously used by Zhang et al.(2011) to show that double training at the same location with a Vernier-acuity task and an orientation task, could transfer the learning of Vernier task to untrained locations. We replicated Zhang et al. findings under a gaze-contingent display, and found that double training at the same location enabled Vernier learning (either with Gabor or line stimuli) to transfer over untrained locations. We next examined whether this double training would also lead to spatial transfer of learning in a 3-dot hyperacuity task, which has been shown as a location-specific task in our previous study (Hung and Seitz, 2011). We found that learning in the 3-dot hyperacuity task showed no transfer to untrained locations under the same training paradigm. Our results show that the location specificity in PL is not ubiquitously decoupled after double training. We suggest that specificity of learning can arise from decision rules, attention learning, or representational changes and that small differences in the training approach can emphasize some of these mechanisms over the others.
Meeting abstract presented at VSS 2013