Table 2. Results of the predictions made for the held-out test data.
| cognitive domain | correlations | NeuroSynth | group | |||
|---|---|---|---|---|---|---|
| MH,MG,DN | SH,FH,LB,AH,SH | RV,AK,JS | DP,MS,JA,SZ | |||
| language | 5 (0) | 2 (-) | 5 (1) fruits |
5 (3) animals, tools, fruits |
5 (3) animals, tools, fruits |
5 (3) animals, tools, fruits |
| sensory-motor skills | 5 (2) tennis, climbing |
5 (-) | 4 (1) climbing |
4 (3) badminton, tennis, climbing |
3 (2) tennis, climbing |
4 (3) badminton, tennis, climbing |
| visuo-spatial memory | 5 (0) | 3 (-) | 5 (0) | 5 (1) market |
5 (0) | 5 (1) market |
| visual processing of faces | 5 (2) athletes, friends |
1 (-) | 5 (1) movie actors |
4 (3) movie actors, athletes, friends |
3 (1) athletes |
5 (2) athletes, friends |
| resting | 4 (-) | 4 (-) | 4 (-) | 5 (-) | 3 (-) | 4 (-) |
| total | 24 (4) | 15 (-) | 23 (3) | 23 (10) | 19 (6) | 23 (9) |
Number of correct predictions for each domain, for the 5 blocks of the test data per condition. Predictions are based on the correlation approach (top 1% of voxel, no smoothing), the NeuroSynth analysis, and based on the submissions made by the four teams of human raters. Numbers in brackets indicate the number of correct predictions regarding content. (-) indicates that content predictions were not preformed. Content predictions were made for all cognitive domains except rest. Abbreviations indicate author names.