Abstract:

PURPOSE: Two theories explain visual perceptual learning (VPL): Early cortical neural plasticity (e.g., Karni & Sagi, 1991; Stoups et al., 2001) and neuron-specific response reweighting (Dosher & Lu, 1987; Law & Gold, 2008), as motivated by observed learning specificities. However, our double training studies (Xiao et al., 2008; Zhang et al., 2010; Wang et al., 2016) demonstrate that VPL often shows complete transfer to untrained locations, orientations/directions, and physical properties, indicating that VPL is primarily a rule-based high-level process operating at a conceptual level. Here we test a hypothesis that VPL requires neither early cortical neural plasticity nor neuron-speci?c response reweighting.
METHODS: In a peripheral orientation-discrimination learning task, training took place with the Gabor stimulus either rotated in 12 locations (anti-clockwise) and 4 orientations 90-deg apart (clockwise), or in a random order. Each location/orientation combination received 2 trials per block of trials, 6 blocks per session, for a total of 60 trials over 5 daily sessions. A staircase controlled the orientation di?erence from condition to condition. The pre/post condition was never practiced.
RESULTS: Both rotating and random training conditions produced significant orientation learning, indicating that the observers can learn orientation discrimination with multiple conditions that minimize neural plasticity in early visual areas and neuron-speci?c response reweighting. Learning transferred to the untrained pre/post location/ orientation condition, so that the improvement was comparable to that in a control group that only practiced the pre/post condition for an equal number of trials. Therefore, orientation learning with multiple stimuli is genuine and unspecific to the more difficult multiple stimulus task. Similar results were seen in a motion-direction learning task with 6 rotating locations and 8 rotating directions 45-deg apart, but without direction/location speci?cities.
CONCLUSIONS: These results indicate that early cortical plasticity and neuron-speci?c input reweighting are unnecessary for VPL. Instead sensory inputs can be reweighted at a later stage in a neuron unspeci?c manner.

Key words: visual perceptual learning, orientation discrimination, early cortical neural plasticity, neuron-specific response reweighting