Abstract
Visual perceptual learning (VPL) studies typically train a specific stimulus feature at a specific retinal location, and learning is often attributed to training-induced V1 plasticity because of the location and feature specificity. However, recent double training studies from our lab challenged the specificity observations and the V1-plasticity explanation. Here we present a new study that minimizes V1 plasticity by design in a orientation discrimination task. Two groups of observers practiced a peripheral orientation discrimination task with Gabor stimuli at 5° retinal eccentricity. Both groups had pre- and post-training thresholds measured at the same test location and orientation. The control group practiced the test condition for five sessions. The experimental group followed the same procedure, but the Gabor location rotated along the 5°-eccentricity circle every trial at 12 preset locations (30° apart between neighboring locations). The Gabor orientation also rotated at 4 preset orientations (45° apart), but it would jump to the next orientation if the Gabor was shown at the test location and its orientation was rotated to the test orientation. For each combined orientation-location condition, only 60 trials were practiced, which was 1/48 of the trials the control group practiced at the test condition. Neurons respectively responding to each stimulus condition in the retinotopic and orientation selective V1 were thus minimally trained and those responding to the test condition were not trained. Training reduced orientation thresholds at the test condition by 32.0 ± 9.4% (p = 0.015, two-tailed paired t-test) in the experimental group and by 31.0 ± 10.5% (p = 0.041) in the control group. There was no significant difference of learning between the two groups (p = 0.944). These results indicate that orientation learning can be achieved with minimal V1 plasticity, which is consistent with our claim that VPL is a high-level cognitive process.
Meeting abstract presented at VSS 2016