To minimize the influences of procedural learning, on the first day the naive subjects were given a few tens of practice trials with suprathreshold differences in stimulus orientation. Orientation discrimination thresholds were measured with a standard one-up, two-down staircase

procedure converging at 70.7% correct responses. The orientation step size of the staircase Y 27632 was 0.05 log units. Each staircase (i.e. a block of trials) consisted of eight reversals. The geometric mean of the last six reversals was calculated as the threshold. A typical staircase comprised 35–50 trials. The subjects compared a difference in orientation between two successively presented stimuli. Thirteen naive subjects were randomly assigned to practice under either the congruent condition (left panels in Fig.1A; Group I subjects, n = 6) or the incongruent condition (right panels in Fig. 1A; Group II subjects, n = 7). After the training,

their thresholds were measured under both the congruent and incongruent conditions, and at the trained 55° orientation as well as at an untrained orientation of 140° (Fig. 1B). Note that the two stimuli in learn more a trial occupied two different retinal locations, but these two retinal locations were the same in the congruent and incongruent conditions. Therefore, the congruent and incongruent spatial

relations of the two stimuli were in terms of a spatiotopic, rather than a retinotopic, reference frame (referred to as the spatiotopic stimulus relation throughout the text). As in many perceptual learning tasks, training decreased the subjects’ thresholds for orientation discrimination by approximately a factor of two in Group I subjects trained under the congruent condition (pre-training threshold 7.62° ± 0.48° vs. post-training Astemizole threshold 4.07° ± 0.3°, t = 6.41, P = 0.001, paired t-test) and in Group II subjects trained under the incongruent condition (pre-training threshold 7.44° ± 1.00° vs. post-training threshold 3.71° ± 0.32°, t = 4.35, P = 0.002). However, when the spatiotopic stimulus relation was switched from trained to untrained without changing the stimulus location on the retina, there was a significant elevation of the mean thresholds at the trained 55° orientation in both Group I subjects (t = 5.06, P = 0.004; left panel in Fig. 1B, compare the two bars corresponding to the 55° condition) and Group II subjects (t = 4.33, P = 0.005; right panel in Fig. 1B, compare the two bars corresponding to the 55° condition), indicating spatiotopic location specificity of the learning. This observation suggests that spatiotopic processing mechanisms can be tuned in favor of the trained spatiotopic stimulus relation.