Learning effects in experiment 2.

<p>A. The effect training had on orientation discrimination in experiment 2 was quantified as the change in orientation discrimination sensitivity between test sessions 1 and 5 (the training effect). This was calculated by subtracting the slopes for the pre-test (session 1) from the slopes for the post-test (session 5) for each participant and then calculating the average across all participants. The training effect was significantly greater around the trained than the untrained reference orientations. B. Tilt illusion magnitude (with surround) and perceptual biases (without surround) were estimated by extrapolating the PSE from the psychometric curve fitted to the test sessions. We calculated the difference in PSE between session 1 and session 5 by subtracting the PSE for the pre-test (session 1) from the PSE for the post-test (session 5) and averaging these differences across all participants. Critically, illusion magnitude (which was bias corrected by subtracting the PSE without surround for each participant) did not change significantly between session 1 and session 5 at the trained or untrained conditions. In both panels error bars indicate ±1 standard error of the mean across participants. The insets illustrate the stimulus conditions but note that which orientation was trained was counterbalanced across participants.