Effects of Perceptual Learning Through Binocular Virtual Reality Exercises on Low Stereopsis

Abstract

Objectives: To evaluate the feasibility and efficacy of a brief, scaffolded virtual reality (VR)-based perceptual-learning protocol to improve stereovision in adults without measurable baseline global stereopsis. Methods: Prospective interventional pilot study with a normal-vision reference group. Fourteen adults were enrolled; nine with nil global stereoacuity formed the intervention group and five with normal binocular vision served as controls. Intervention participants completed 8–10 VR sessions (~30 min each over two weeks) using a disparity-driven depth-matching task with individualized thresholds, virtual prism compensation, and interocular contrast balancing. Primary outcomes were changes in clinical stereoacuity (local and global) and Binocular Function (BF). Secondary outcomes included positive fusional vergence (PFV), binocular imbalance (inter-ocular contrast ratio), and in-game depth discrimination error (arcseconds). Results: Global stereoacuity improved with four of nine participants who initially showed no measurable global stereopsis achieving measurable thresholds after training (post: 861.11 ± 537.08 arcsec, IQR = 800.00; p = 0.06). Local stereoacuity improved near (pre: 616.67 ± 532.68 arcsec, IQR = 1100.00; post: 361.67 ± 537.11, IQR = 200.00; p = 0.03) and at a distance (pre: 715.11 ± 573.54 arcsec, IQR = 1114.00; post: 437.11 ± 510.41, IQR = 405.00; p = 0.03). BF score improved from 4.33 ± 0.50 (IQR = 1.00) to 3.51 ± 1.14 (IQR = 1.30) (p = 0.03). PFV break at distance increased (pre: 17.83 ± 17.53Δ, IQR = 23.50; post: 24.50 ± 17.12Δ, IQR = 28.50; p = 0.04), and the binocular imbalance showed a nonsignificant trend towards improvement (p = 0.09). In-game depth error decreased from 752.11 ± 384.50 arcsec (IQR = 553.00) to 221.78 ± 79.12 (IQR = 105.00) (p < 0.01). Control participants achieved a mean depth error of 43.00 ± 10.30 arcsec after three sessions. Conclusions: A short-dose, individualized VR protocol yielded gains in stereoacuity and binocular function in adults with severe stereodeficiency, with transfer from in-game learning to standard clinical measures. While the sample size was limited, the approach proved feasible and was well tolerated, showing encouraging efficiency versus prior high-dose regimens.