3.1.Behavioral results
The 3 (Group: IC training/ Active control/ Passive control) × 2 (Test session: Pre-test/ Post-test) × 2 (Trial type: Switch/ Non-switch) repeated measures ANOVA conducted on error rates revealed a main effect of trial type, F (1, 56) = 17.50,p < 0.001, ηp2 = 0.238, such that switch trials led to more errors than non-switch trials (Fig. 3A).
The interaction between trial type and group was significant, F (2, 56) = 5.36,p = 0.007, ηp2 = 0.161. Bonferroni corrected follow-up comparisons showed that individuals in the IC training group were less accurate in switch trials than non-switch trials, F (1,56) = 24.58, p < 0.001, ηp2 = 0.305. The difference between switch and non-switch trials was marginally significant in the passive control group, F (1,56) = 4.02, p = 0.050, ηp2 = 0.067 and non-significant in the active control group, F (1,56) < 1.
The interaction between group and test session was significant, F (2,56) = 4.50, p = 0.015, ηp2 = 0.139. Bonferroni corrected follow-up comparisons showed that individuals in the active control group performed more accurate in the post-test than in the pre-test, F (1,56) = 9.68, p = 0.003, ηp2 = 0.147. There was no significant change in the IC training group (F (1,56) < 1) and passive control group ( F (1,56) < 1).No other significant main effect or interactions were found (Test session: F (1,56) = 1.46, p = 0.232, ηp2 = 0.025; Group : F (2,56) < 1; Test session × Trial type: F (1,56) < 1; Group × Test session × Trial type: F (2,56) = 1.34, p = 0.271, ηp2 = 0.046).
The 3 (Group: IC training/ Active control/ Passive control) × 2 (Test session: Pre-test/ Post-test) × 2 (Trial type: Switch/ Non-switch) repeated measures ANOVA conducted on reaction times (RTs) showed a significant main effect of trial type, F (1, 56) = 102.68,p < 0.001༌ηp2 = 0.647, indicating a significant switch cost (Fig. 3B).
The main effect of test session was also significant, F (1, 56) = 16.41,p < 0.001༌ηp2 = 0.227, indicating that participants were overall faster in the post-test than the pretest session. The main effect of group was significant, F (2, 56) = 4.74༌p = 0.013, ηp2 = 0.145. Further analysis (Bonferroni corrected) revealed that RTs were slower in the active control group than in the IC training (p = 0.028) and passive control (p = 0.030) groups, but the training group and passive control groups did not differ between them (p = 1). The interaction between group and test session was significant, F (2, 56) = 4.58༌p = 0.014༌ηp2 = 0.1. Further analysis showed that both the IC training group and the active control group tended to respond faster after training (IC training group: F (1,56) = 3.86, p = 0.054, ηp2 = 0.065; active control group: F (1,56) = 21.00, p < 0.001, ηp2 = 0.273), but there was no significant improvement in the passive control group (F (1,56) = 0.17, p = 0.683, ηp2 = 0.003). No other significant interactions were found (Group × Trial type: F (2,56) = 0.715, p = 0.494, ηp2 = 0.025; Test session × Trial type: F (1,56) = 0.59, p = 0.446, ηp2 = 0.010; Group × Test session × Trial type: F (2,56) = 1.01, p = 0.370, ηp2 = 0.035).
3.2.Whole brain results
The 3 (Group: IC training/ Active control/ Passive control) × 2 (Trial type: Switch/ Non-switch) repeated measures ANOVA performed on the three groups in the pre-test session showed significant main effect of trial type in the left medial superior gyrus, the left supplementary motor area and critically, the left DLPFC, but also the left inferior parietal lobule, the left precuneus, the right orbital inferior frontal gyrus, the right anterior and middle cingulate cortex, the right cerebellum, and bilateral caudate nucleus (see Table 2 and Fig. 4). However, the main effect of group and the interaction between group and trial type were not significant in any of the brain regions.
Table 2
The brain areas with significant activation for the main effect of trial type revealed by the 3 (Group: IC training/ Active control/ Passive control) × 2 (Trial type: Switch/ Non-switch) repeated measures ANOVA (FDR corrected, p < 0.05, cluster corrected, k > 15).
Brain regions | BA | Cluster size | MNI coordinates (x, y, z) | F-value |
L Medial Superior Frontal Gyrus | 10 | 86 | -6 | 66 | 24 | 32.47 |
L Caudate nucleus | - | 80 | -24 | 6 | 18 | 17.84 |
L Supplementary Motor Area / DLPFC | 6 | 463 | -3 | 15 | 54 | 38.25 |
L Precuneus | 7 | 197 | -9 | -66 | 36 | 31.59 |
L Inferior Parietal Lobule | 40 | 80 | -33 | -48 | 36 | 28.28 |
R Orbital Inferior Frontal Gyrus | 13 | 27 | 48 | 21 | -9 | 21.23 |
R Caudate nucleus | - | 115 | 12 | 3 | 3 | 18.28 |
R Anterior Cingulate Cortex | 32 | 25 | 3 | 36 | 21 | 24.99 |
R Middle Cingulate Cortex | 23 | 18 | 6 | -21 | 27 | 20.35 |
R Cerebellum | | 19 | 33 | -87 | -39 | 28.58 |
Note. BA = Brodmann area; L = left; R = right. |
3.3.ROI results
In the DLPFC, a 2 (Test session: Pre-test/Post-test) × 3 (Group: IC training/ Active control/ Passive control) ANOVA of beta values did not validate a main effect of test session, F(1, 56) = 1.11, p > 0.2, ηp2 = 0.019, or a main effect of group, F(1, 56) = 1.73, p = 0.187, ηp2 = 0.058, (Fig. 5). However, the interaction between test session and group was significant, F(2, 56) = 3.17, p < 0.05, ηp2 = 0.102. Bonferroni corrected follow up tests showed a significant increase in beta values between post- and pre-test session, in the IC training group, F(1, 56) = 6.71, p = 0.012, ηp2 = 0.107, but such differences did not reach significance in either the active control or the passive control groups (Fs(1, 56) < 1).
For the dACC, the ANOVA showed a significant main effect of test session, F(1, 56) = 5.27, p < 0.05, ηp2 = 0.086, suggesting that activation in the ACC was stronger after training. The main effect of group was not significant, F(1, 56) = 2.28, p > 0.1, ηp2 = 0.075 and neither was the interaction between group and test session, F(2, 56) < 1.
For the LCN, the ANOVA showed that the main effect of test session was not significant, F(1, 56) < 1. The main effect of group was significant, F(1, 56) = 3.75, p < 0.05, ηp2 = 0.118. Post hoc analysis (Bonferroni corrected) showed significant difference between IC training group and active control group, p < 0.05, but no other overall difference in beta values between groups (IC training group vs. passive control group: p = 0.179; active control group vs. passive control group: p = 1). The interaction between group and test session was not significant, F(2, 56) < 1.
3.4.Correlation results
A correlation analysis showed that the change in the beta values associated with switch costs (i.e., the contrast between switch and non-switch trials) in the left DLPFC was negatively correlated with the change in the behavioral switch costs, r = -0.440, p = 0.026, in the IC training group (Fig. 5. No such correlation was found in the dACC (r = 0.150, p = 0.264) or LCN (r = -0.283, p = 0.113) in the same group. In the two control groups, we found no significant correlation between beta value change and switch cost change in any of the three ROIs (active control group: DLPFC (r = 0.307, p = 0.100), dACC (r = 0.310, p = 0.098), or LCN (r = 0.253, p = 0.148); passive control group: DLPFC (r = -0.279, p = 0.117), dACC (r = -0.145, p = 0.270), or LCN (r = 0.335, p = 0.075).