Perceptual learning to reduce sensory eye dominance beyond the focus of top-down visual attention

Abstract

Perceptual learning is an important means for the brain to maintain its agility in a dynamic environment. Top-down focal attention, which selects task-relevant stimuli against competing ones in the background, is known to control and select what is learned in adults. Still unknown, is whether the adult brain is able to learn highly visible information beyond the focus of top-down attention. If it is, we should be able to reveal a purely stimulus-driven perceptual learning occurring in functions that are largely determined by the early cortical level, where top-down attention modulation is weak. Such an automatic, stimulus-driven learning mechanism is commonly assumed to operate only in the juvenile brain. We performed perceptual training to reduce sensory eye dominance (SED), a function that taps on the eye-of-origin information represented in the early visual cortex. Two retinal locations were simultaneously stimulated with suprathreshold, dichoptic orthogonal gratings. At each location, monocular cueing triggered perception of the grating images of the weak eye and suppression of the strong eye. Observers attended only to one location and performed orientation discrimination of the gratings seen by the weak eye, while ignoring the highly visible gratings at the second, unattended, location. We found SED was not only reduced at the attended location, but also at the unattended location. Furthermore, other untrained visual functions mediated by higher cortical levels improved. An automatic, stimulus-driven learning mechanism causes synaptic alterations in the early cortical level, with a far-reaching impact on the later cortical levels.

Figure 1

Stimuli for measuring the balance contrast and the push-pull training. (a) and (b) Orthogonal gratings used to measure the balance contrast in the LE and RE, respectively. (c) For the training, two retinal locations, one for the attended and the other for the unattended condition, are simultaneously stimulated. At each location, a cue (white rectangular frame) attracts transient attention to the weak eye, causing its (vertical and near-vertical) gratings to be perceived while the (horizontal) gratings in the strong eye are suppressed. The observer discriminates the orientation of the gratings seen by the weak eye at the attended location.

Figure 2

Results of the push-pull training at the attended and unattended locations. (a) The average interocular balance contrast at the attended location obtained, respectively, with grating whose orientation was the same as, or orthogonal to, the grating used in the training. The same interocular balance contrast is the measured contrast in the weak eye minus 1.5 log unit (fixed contrast of grating in the strong eye); whereas the orthogonal interocular balance contrast is the measured contrast in the strong eye minus 1.5 log unit (fixed contrast of grating in the weak eye). The interocular balance contrast reduces significantly with in training, particularly when tested with the same orientation grating. (b) The average interocular balance contrast at the unattended location exhibits a similar trend as that at the attended location. (c) SED, defined as the difference between the same and orthogonal interocular balance contrast reduces significant at both the attended and unattended locations as the training progresses.

Figure 3

Boundary contour-based SED. (a) Stimulus with vertical grating surrounding the vertical and horizontal grating discs. The spatial phase of the vertical grating disc relative to the vertical surround is shifted to obtain the point of neutrality. (b) Similar to (a) except that the gratings are oriented 45° and 135° and the point of neutrality is obtained from the relative phase shift of the 135° grating disc. (c) The BC-based SED is significantly reduced after the training at the attended location but not at the unattended location, with both stimuli (a) and (b).

Figure 4

Dynamics of interocular dominance and suppression before (pre) and after (post) the training, measured with gratings whose orientations were either the same as, or orthogonal to, the training gratings. The data are plotted as a ratio of the performance of the weak eye to the strong eye. Thus, a ratio of greater than unity indicates a superior performance in the weak eye for that stimulus. (a) The predominance ratios increase significantly with the same grating after the training at both the attended and unattended locations, indicating an improvement of the weak eye. (b) The trend of the dominance duration ratios is similar to (a). (c) The dominance frequency ratios do not change significantly with training.

Figure 5

Transfer of perceptual learning to stereopsis. (a) The random-dot stereogram used for measuring binocular disparity threshold for seeing a disc target in depth. (b) Binocular disparity thresholds are significantly reduced at both the attended and unattended locations after the training.

Figure 6

Monocular contrast thresholds are significantly reduced after the training at the attended and unattended locations in both the weak and strong eyes. However, these generalized and small reductions are unlikely to be associated with the reduction in SED.