Figures and data
Procedures and transfer effect using orientation task.
(A) Experimental design. Participants completed five sessions, including a pretest, training or reactivation, and a posttest. (B) Orientation detection task. Participants reported which of the two intervals contained the Gabor orientation. (C) Normalized learning gain index (NGI) for trained versus untrained orientations in the Reactivation (Anodal vs. Sham) groups. (D) Thresholds (S/N ratio) in Reactivation/Anodal group. A two-way repeated measures ANOVA (session: pretest vs. posttest × orientation: trained vs. untrained) revealed a significant main effect of session (F(1,23) = 124.406, p < 0.001, η2p = 0.844), but no interaction effect (F(1,23) = 0.010, p = 0.923, BF01 = 3.633), demonstrating comparable learning for the trained (paired t-test: t(23) = 7.338, p < 0.001, Cohen’s d = 1.498) and untrained (paired t-test: t(23) = 7.765, p < 0.001, Cohen’s d = 1.585) orientations. (E) Thresholds (S/N ratio) in Reactivation/Sham group. A two-way repeated measures ANOVA (session: pretest vs. posttest × orientation: trained vs. untrained) revealed a significant interaction (F(1,23) = 16.477, p < 0.001, η2p = 0.417), demonstrating stronger learning for the trained (paired t-test: t(23) = 8.386, p < 0.001, Cohen’s d = 1.712) compared to the untrained (paired t-test: t(23) = 2.599, p = 0.016, Cohen’s d = 0.531) orientation. The central lines in the box plot indicate the median values. The upper and lower lines represent the interquartile range (25th – 75th percentiles). Each dot represents data from one participant. *p < 0.05, ***p < 0.001, n.s. = not significant.
Transfer effect using motion detection task.
(A) Motion detection task. Participants reported which of the two intervals contained the coherent motion dot field. (B) Normalized learning gain index (NGI) for the trained versus untrained direction in the Reactivation (Anodal vs. Sham) groups. (C) Thresholds (motion coherence) in Reactivation/Anodal group. A two-way repeated measures ANOVA (session: pretest vs. posttest × direction: trained vs. untrained) revealed a significant main effect of session (F(1,23) = 102.652, p < 0.001, η2p = 0.817), but no interaction effect (F(1,23) = 0.252, p = 0.621, BF01 = 3.239), demonstrating comparable learning for the trained (paired t-test: t(23) = 9.134, p < 0.001, Cohen’s d = 1.864) and untrained (paired t-test: t(23) = 8.973, p < 0.001, Cohen’s d = 1.832) directions. (D) Thresholds (motion coherence) in Reactivation/Sham group. A two-way repeated measures ANOVA (session: pretest vs. posttest × orientation: trained vs. untrained) revealed a significant interaction (F(1,23) = 9.864, p = 0.005, η2p = 0.300), demonstrating larger learning effect for the trained (paired t-test: t(23) = 7.764, p < 0.001, Cohen’s d = 1.585) compared to the untrained (paired t-test: t(23) = 3.244, p = 0.004, Cohen’s d = 0.662) direction. The central lines in the box plot indicate the median values. The upper and lower lines represent the interquartile range (25th – 75th percentiles). Each dot represents data from one participant. **p < 0.01, ***p < 0.001, n.s. = not significant.
Full-Practice leads to learning specificity.
(A) Normalized learning gain index (NGI) for the trained versus untrained orientation in the Full-Practice (Anodal vs. Sham) groups. (B) Thresholds (S/N ratio) in Full-Practice/Anodal group. A two-way repeated measures ANOVA (session: pretest vs. posttest × orientation: trained vs. untrained) revealed a significant interaction (F(1,23) = 17.961, p < 0.001, η2p = 0.438), demonstrating larger learning effect for the trained (paired t-test: t(23) = 10.689, p < 0.001, Cohen’s d = 2.182) compared to the untrained (paired t-test: t(23) = 2.782, p = 0.011, Cohen’s d = 0.568) orientation. (C) Thresholds (S/N ratio) in Full-Practice/Sham group. A two-way repeated measures ANOVA (session: pretest vs. posttest × orientation: trained vs. untrained) revealed a significant interaction (F(1,23) = 5.380, p = 0.030, η2p = 0.190), demonstrating larger learning effect for the trained (paired t-test: t(23) = 6.932, p < 0.001, Cohen’s d = 1.415) compared to the untrained (paired t-test: t(23) = 4.101, p < 0.001, Cohen’s d = 0.837) orientation. The central lines in the box plot indicate the median values. The upper and lower lines represent the interquartile range (25th – 75th percentiles). Each dot represents data from one participant. *p < 0.05, ***p < 0.001, n.s. = not significant.
