Stimuli and procedure of experiment 1a

Binocular rivalry stimuli were composed of 2 orthogonal sine-wave grating disks, oriented ±45° (diameter: 1°, spatial frequency: 3 cpd, Michaelson contrast: 80%). They were presented dichoptically and foveally with a central red fixation point (0.07° in diameter) and a high contrast checkerboard “frame” (size: 2.5° × 2.5°; 0.25° thick) to facilitate stable binocular fusion.

Each binocular rivalry test was composed of sixteen 60-s rivalry trials. Each trial started with a 5-s blank interval. Then, the rival stimuli were presented for 55 s. Subjects were required to hold down 1 of the 3 keys (left, right, or down arrow) to report their perceptions (clockwise, counterclockwise, or mixed). The orientation related to each eye was kept constant within a trial but randomly varied across the trials.

During 1 h of adaptation, subjects passively viewed dichoptically presented movie images (Fig. 1a) surrounded by a high contrast checkerboard “frame” (size: 12.28° × 19.12°; 0.29° thick). The frame rate of movies was 25 fps. The original movie images were presented to 1 of the 2 eyes, while the corresponding backward movie images were presented to the opposite eye. The backward movie images were offline processed with 3 steps: dividing the regular movie into a series of 20-min segments, generating a backward copy for each 20-min segment in the MediaEditor software (http://www.aijianji.com/medownload.htm), concatenating the 3 segments to produce a backward movie file. That is, for each 20-min segment, the backward version of the movie was formally identical to the original one except for the absence of a logical movie plot.

For the first 6 subjects, the audio track always synchronized with the regular movie images (i.e. the “Synchrony” condition). To examine the potential contribution of audiovisual integration, we added another 2 different adaptation conditions for the 16 later recruited subjects. In the “Asynchrony” condition, the audio track was 5 s ahead of the movie image. Since the time window of audiovisual integration is only several hundred milliseconds wide (Conrey and Pisoni 2006), 5 s is sufficiently long to avoid audiovisual integration. In the “No-sound” condition, the movie was played silently.

Before the formal experiment, all subjects practiced 3 binocular rivalry tests per day (with a 10-min break in between) for 3–7 days, to ensure a stable performance of binocular rivalry (Bao et al. 2018). Because perceptual eye dominance fluctuated widely in the first several trials of a day (Suzuki and Grabowecky 2007), before the practice of each day, subjects completed 5 warm-up binocular rivalry trials, the data of which were not analyzed.

Perceptual eye dominance was determined by the last 3 sessions of the “binocular rivalry practice” stage, with the dominant eye being the one that showed the longer summed phase durations.

The formal experiment included 4 phases: (1) 5 warm-up binocular rivalry trials (data not analyzed), (2) a preadaptation binocular rivalry test, (3) 1 h of dichoptic-backward-movie adaptation, and (4) a post-adaptation binocular rivalry test.

Subjects completed 4 sessions for each adaptation condition, with each eye attended in 2 sessions and the sequence counterbalanced. This resulted in totally 12 adaptation sessions. The order of the 3 adaptation conditions was balanced across subjects.