Generality of learning as driven by dichoptic visual training

Abstract

Training protocols involving dichoptic presentation of motion stimuli have been shown to be effective in promoting eye rebalancing and improving binocular functions; however, the mechanisms underlying these improvements have yet to be clarified. Here, we aimed to characterise mechanisms underlying learning-induced improvements in sensory eye balance and binocular functions by testing the generality of learning effects that can be attained with different training and test cues. Two groups of visually-normal observers were trained on one of two tasks: a dichoptic signal-noise motion task or a dichoptic fine motion task. In the signal-noise task, signal dots carrying a coherent motion direction and randomly-moving noise dots were presented dichoptically. The observers’ task was to indicate the net motion direction of the stimulus. In the fine task, stimuli carried a center-surround configuration such that a reference motion direction in the surround was presented to one eye while a target direction in the center was presented to the alternate eye. The observers’ task was to judge whether the motion direction carried by the central target was offset clockwise or counter-clockwise with respect to the reference motion. Observers received training over three consecutive days (3600 trials) and were tested on a binocular phase combination task to index sensory eye balance, signal-noise and fine depth discrimination tasks to measure stereopsis, and dichoptic signal-noise and fine tasks involving both motion and orientation cues before and after training. Results showed that both training tasks altered sensory eye dominance and improved stereopsis. Interestingly, training on both tasks generalized to improvements in the orientation task, but training using the signal-noise manipulation drove a broader transfer of learning. Our data suggest that in addition to the dichoptic presentation mode itself, the nature of the training stimulus may also exert influence on training effectiveness, potentially implicating signal-noise segregation mechanisms in dorsal cortex.