Skip to main content
. 2015 Aug 26;9:225. doi: 10.3389/fnbeh.2015.00225

Table 2.

Results from leave-one-out validation.

Subject -Bloc 1 -Bloc 2 -Bloc 3 -Bloc 4 All blocs
1 E-Coord W-Mix E-Coord E-Coord W-Mix
2 VPI-select E-Coord E-Coord E-Coord E-Coord
3 E-Coord E-Coord E-Coord E-Coord E-Coord
4 E-Coord E-Coord E-Coord E-Coord E-Coord
5 W-Mix E-Coord E-Coord E-Coord E-Coord
6 E-Coord E-Coord E-Coord W-Mix E-Coord
7 E-Coord VPI-select VPI-select W-Mix W-Mix
8 W-Mix VPI-select VPI-select E-Coord VPI-select
9 VPI-select VPI-select VPI-select VPI-select W-Mix
10 VPI-select VPI-select VPI-select VPI-select VPI-select
11 E-Coord E-Coord E-Coord E-Coord E-Coord
12 E-Coord W-Mix E-Coord W-Mix E-Coord
13 E-Coord E-Coord E-Coord E-Coord E-Coord
14 E-Coord E-Coord E-Coord E-Coord E-Coord

Each bloc is removed systematically for optimization. Discordant models compared with optimization with all blocs are shown in bold. (Q-L, Q-Learning; BWM, Bayesian Working Memory; VPI-select, VPI-based selection; W-Mix, Weight-based mixture; E-Coord, Entropy-based coordination).