Table 2.
Comparison of prediction accuracies between networks
| Domain of predicted performance | Network A | RMSE Mean (SD) | Network B | RMSE Mean (SD) | t | p-value |
|---|---|---|---|---|---|---|
| Significant main effect of network | ||||||
| WM | Power nodes | 1.02 (0.06) | WM | 1.06 (0.06) | −3.30 | 0.025 |
| SOCIAL | 1.08 (0.06) | −4.65 | < 0.001 | |||
| EMO | 1.08 (0.06) | −6.03 | < 0.001 | |||
| SOCIAL | Power nodes | 1.04 (0.07) | WM | 1.11 (0.06) | −6.37 | < 0.001 |
| SOCIAL | 1.11 (0.06) | −5.72 | < 0.001 | |||
| EMO | 1.08 (0.07) | −4.43 | < 0.001 | |||
| EMO | Power nodes | 1.01 (0.06) | WM | 1.07 (0.07) | −5.05 | < 0.001 |
| SOCIAL | 1.09 (0.06) | −7.10 | < 0.001 | |||
| EMO | 1.05 (0.06) | −3.93 | 0.003 | |||
| Significant post-hoc tests for specific network—task—state combinations | ||||||
| WM | Power (List Sorting, REST state) | 1.07 (0.09) | EMO (List Sorting, REST state) | 1.19 (0.09) | −4.41 | 0.002 |
| SOCIAL | Power (Soc. Cog., EMO state) | 1.02 (0.11) | SOCIAL (Soc. Cog., EMO state) | 1.14 (0.11) | −4.80 | < 0.001 |
| Power (Soc. Satisf., EMO state) | 1.01 (0.08) | WM (Soc. Satisf., EMO state) | 1.11 (0.08) | −3.82 | 0.017 | |
| Power (Soc. Satisf., WM state) | 1.04 (0.08) | WM (Soc. Satisf., WM state) | 1.15 (0.08) | −3.68 | 0.027 | |
| EMO | Power (Emo. Recog., REST state) | 1.00 (0.09) | SOCIAL (Emo. Recog., REST state) | 1.13 (0.10) | −4.05 | 0.008 |
| EMO (Emo. Recog., REST state) | 1.11 (0.09) | −4.39 | 0.002 | |||
| Power (Emo. Recog., WM state) | 0.99 (0.09) | SOCIAL (Emo. Recog., WM state) | 1.13 (0.10) | −5.18 | < 0.001 | |
| Power (Emo. Recog., SOCIAL state) | 1.00 (0.09) | SOCIAL (Emo. Recog., SOCIAL state) | 1.11 (0.09) | −4.47 | 0.002 | |
| Power (Emo. Recog., EMO state) | 1.01 (0.09) | SOCIAL (Emo. Recog., EMO state) | 1.12 (0.09) | −4.47 | 0.002 | |
Note. Machine-learning-adjusted t-tests to assess network specificity using the averaged 100 RMSE values obtained from 100-fold cross-validation within the network listed in column “Network A” versus the network listed in column “Network B”. p-values are Bonferroni corrected for multiple comparisons. Post-hoc t-tests between individual predictions of the task in the state (both noted in brackets) and the network listed in column “Network A” versus the state listed in column “Network B”.
Soc. Cog.: Social cognition task (in-scanner task), Soc. Satisf.: Social Satisfaction Questionnaire (out-of-scanner score), Emo. Recog.: Emotional Face Recognition task (out-of-scanner score).