Abstract
Ocular alignment defects, including strabismus and convergence insufficiency, affect around 5% of people and are associated with binocular vision impairments. Current treatments lack quantitative endpoints and have high levels of recidivism. We develop a rehabilitation method for ocular alignment training and examined the rate of learning, transfer to untrained alignments, and retention over time. Ocular alignment was controlled with a real-time dichoptic feedback paradigm in which a fixation target was presented only to the fixating eye and a gaze-contingent ring was presented only to the alignment eye. Observers were required to move their eyes to center the ring on the target, with feedback provided by the size and color of the ring. By offsetting the ring temporally or nasally, this task required convergent or divergent deviation, respectively, of the alignment eye. Learning was quantified as the time taken to attain target deviation of 2° or 4° for convergence (C2, C4) or divergence (D2, D4) over 40 trials. 32 normally-sighted observers completed 2 training sessions separated by 1 week. Each session, subjects completed 3 alignment tasks and were randomly assigned to a test sequence: C2-C4-D2, C4-C2-D2, D2-D4-C2, or D4-D2-C2. The time taken to achieve alignment was higher for greater deviation angles (p<.001) but there was no significant difference between divergent and convergent directions (p=.910). Previous training on one deviation angle transferred to untrained angles within convergence or divergence (p<.001) but not between these directions (p=.365), and times were significantly shorter in the second session (p<.001). The results show that oculomotor alignment can be rapidly trained with a feedback based dichoptic paradigm. Training is transferred and retained over at least one week within the same deviation direction, but not between deviation directions. Feedback-based oculomotor training may therefore provide a non-invasive method for the rehabilitation of ocular alignment defects.