Abstract
Motion perceptual learning is possible even when MT is putatively suppressed using a motion-opponent psychophysical stimulus. Here, using fMRI, we investigated whether this learning took place at MT. Observers performed a motion-axis orientation discrimination task on a stimulus designed to suppress MT via motion opponency. The stimulus consisted of counter-phase paired dots oscillating along a common motion axis (Qian and Andersen, 1994). Care was taken to ensure that no static frame of the stimulus movie provided task-relevant information, ensuring only motion information could be used for discrimination. We first verified that the stimulus indeed suppressed hMT+, as compared to a control stimulus within which the dot phase was changed from 180° to 0° to relieve motion opponency. Observers then trained extensively on one of the two stimuli, and were scanned pre- and post-training. Motion perceptual learning occurred for all trainees and transferred partially to the orientation perpendicular to the trained orientation. No reliable BOLD change was found at hMT+ or V1 in the control group trained without MT suppression. Learning in the experimental group, however, led to enhanced suppression of hMT+ by counter-phase paired dots and to a BOLD reduction at V1. By combining all trainees' data we also identified clusters of activity that were correlated with the amount of behavioral transfer. Reliable negative correlations were found bilaterally in the middle/superior occipital regions and unilaterally in a ventral region of the right occipital lobe that included a portion of the collateral sulcus. Our results indicate that motion perceptual learning with counter-phase paired dots enhanced MT opponency, possibly to suppress noisy responses. This learning also led to a reduced BOLD response in V1, indicating that motion opponency at MT possibly originated at V1. In comparison, when MT functioned normally, motion perceptual learning took place downstream from MT.
Supports: NIH grant EY016391 (BST)