Figure 1.
Task procedure and results of the behavioral session.
(A) The 2IFC global motion detection task in the behavioral session. One stimulus interval contains 15% coherent motion while the other interval contains random motion (0% coherence). After two motion presentations, the subjects were asked to report the interval (first or second) which contained 15% coherent motion. (B) Mean performance across the subjects during the training stage. A motion direction used in this stage was defined as a trained direction for each subject. After 10-day training, a significant training effect was obtained (day 1 vs day 10, paired t-test, P<10−4). Error bars represent SEM. (C) Mean behavioral performance across the subjects for the trained direction in the pre-test (blue) and post-test (red) stages. Significant performance improvement was found after the training stage (paired t-test, P = 0.01). Error bars represent SEM.
Figure 2.
Comparing the two tuning improvement functions.
(A) A behavioral tuning improvement function. The behavioral tuning improvement function was defined by mean performance change across the subjects between the pre- and post-test stages for each of 9 motion directions. Error bars represent SEM. (B) Comparison between the behavioral tuning improvement function (blue) and decoded tuning improvement function for V3A (red). Each improvement function was scaled from 0 to 1 for visualization purpose. Between the two improvement functions, a significant correlation was found (r = 0.86, P<0.05, multiple correction by the number of ROIs with false discovery rate).
Figure 3.
A decoded tuning improvement function for each ROI.
Mean decoding performance improvement across the subjects was calculated by subtracting decoding accuracy in the pre-fMRI stage from that in the post-fMRI stage for each of 9 motion directions. Significant improvement for the trained direction was obtained only in V3A (paired t-test, P<0.05, false discovery rate, corrected by the number of the ROIs). Error bars represent SEM.