Abstract
Stroke-induced damage to primary visual cortex causes loss of conscious vision in the contralateral hemifield. Global motion discrimination training can partially reverse this cortical blindness (CB) at trained locations, recovering near-normal coarse direction discrimination, but leaving fine discrimination impaired. Here, we examined several possible mechanisms for this partial recovery. We used random dot stimuli (10°/s, 3 dots/deg2) containing a Gaussian-distributed range of dot directions centered on the signal direction (Figure 1B), to measure tuning functions for direction in 7 CB subjects. Subjects judged whether dots moved above or below the left/rightward direction (Figure 1A), with thresholds defined as the direction difference resulting in 82% correct performance. Measures were collected before and after left-right direction discrimination training in the blind field and at corresponding intact field locations. Gaussian functions were fit to generate tuning curves for motion channel sensitivity to increased dot direction range (Figure 2). Where measureable (n=5), individual tuning curves were significantly wider at retrained, blind field locations than in the intact hemifield, suggesting broader population tuning to motion in retrained locations. In addition, a neurally plausible generative model of motion discrimination was then utilized to determine if model neurons with broader motion tuning would produce similar psychophysical performance as retrained CB patients. Using a bank of MT-like motion detectors, the model suggests that the source of deficits in fine direction discrimination may be a sparser neuronal population and lower firing rates. Overall, these data suggest that performance in retrained CB fields results from a combination of altered tuning and weaker, sparser neural responses. Thus, an additional, fundamental limit to complete recovery in CB fields may be the number of motion selective neurons available for signal processing, particularly in fine difference tasks of this type.
Meeting abstract presented at VSS 2016