Sensory eye dominance plasticity in the human adult visual cortex

Abstract

Sensory eye dominance occurs when the visual cortex weighs one eye's data more heavily than those of the other. Encouragingly, mechanisms underlying sensory eye dominance in human adults retain a certain degree of plasticity. Notably, perceptual training using dichoptically presented motion signal-noise stimuli has been shown to elicit changes in sensory eye dominance both in visually impaired and normal observers. However, the neural mechanisms underlying these learning-driven improvements are not well understood. Here, we measured changes in fMRI responses before and after a five-day visual training protocol to determine the neuroplastic changes along the visual cascade. Fifty visually normal observers received training on a dichoptic or binocular variant of a signal-in-noise (left-right) motion discrimination task over five consecutive days. We show significant shifts in sensory eye dominance following training, but only for those who received dichoptic training. Pattern analysis of fMRI responses revealed that responses of V1 and hMT+ predicted sensory eye dominance for both groups, but only before training. After dichoptic (but not binocular) visual training, responses of V1 changed significantly, and were no longer able to predict sensory eye dominance. Our data suggest that perceptual training-driven changes in eye dominance are driven by a reweighting of the two eyes' data in the primary visual cortex. These findings may provide insight into developing region-targeted rehabilitative paradigms for the visually impaired, particularly those with severe binocular imbalance.

Keywords: dichoptic perceptual training; fMRI; perceptual learning; plasticity; sensory eye dominance.

Figure 1

Schematics of the (A) dichoptic and (B) binocular variant of a signal-in-noise (SNR) motion task and (C) the general experimental procedure. For the dichoptic variant, signal and noise dots were presented to different eyes on each trial. Two configurations were used such that we presented signal dots to either the left (configuration 1) or the right eye (configuration 2) on each trial. For the binocular variant, signal and noise dots were presented to both eyes on each trial.

Figure 2

Behavioral results showing the degree of learning and changes in sensory eye dominance. Early and late training thresholds for the (A) dichoptic (N = 24) and (B) binocular (N = 25) training groups were derived from averaging the first and the last three training blocks, respectively. (C) Sensory eye dominance in the pre- and post-test for the two training groups as indexed by the binocular balance index derived from the dichoptic signal-in-noise motion test task. An index of zero represents no dominance. Error bars represent ±1 SEM, *p < 0.05.

Figure 3

Group-averaged training data presented independently for each training group. Each point represents a three-block moving average. Error bars represent ±1 SEM.

Figure 4

Differences in GLM beta weights [signals presented to the dominant eye (DE signal) – signals presented to the non-dominant eye (NDE signal)] before and after training, presented independently for the (A) dichoptic and (B) binocular training group. A positive bar represents higher univariate responses when signal dots were presented to the dominant eye. Error bars represent ±1 SEM.

Figure 5

SVM classification accuracies for discriminating fMRI patterned responses between the two stimulus configurations, i.e., [(signal dots presented to the dominant eye) vs. (signal dots presented to the non-dominant eye)] before and after training, presented independently for the (A) LGN, (B) V1 and (C) hMT+. Asterisks denote significant above-baseline (0.5) accuracies. Error bars represent ±1 SEM.

Figure 6

Correlations between SVM accuracies and binocular balance index in V1 and hMT+ before and after training, presented independently for the two training groups. A positive correlation in this context indicates that individuals with stronger eye dominance were associated with a higher pattern-discriminability for the two stimulus configurations in a given region of interest.