Fig. 7: Learning effects on constituent features, pre-post comparison.
From: Visual perceptual learning of feature conjunctions leverages non-linear mixed selectivity
a Orientation discrimination performance significantly improved after training in all experiments (Experiment 1, p = 0.0002, d = 1.9544; Experiment 2, p = 0.0021, d = 1.119; Experiment 3, p = 0.002, d = 2.0685; Experiment 4, p = 0.002, d = 2.3461). b Likewise, color discrimination performance improved in all experiments where color was task-relevant (Experiment 1, p = 0.0255, d = 0.7; Experiment 2, p = 0.0014, d = 1.2995; Experiment 3, p = 0.0195, d = 1.2632), but not in Experiment 4 where only orientation was trained (p = 0.4258, d = 0.4684). CLGF Conjunction Learning with Global Feedback, CLFF Conjunction Learning with Feature-specific Feedback, SFL Separate Feature Learning, OL Orientation Learning. Error bars represent the standard error of the mean. *** denotes p < 0.001, ** denotes p < 0.01, * denotes p < 0.05, n.s. denotes non-significant.
