Skip to main content
. 2018 Mar 26;9:1227. doi: 10.1038/s41467-018-03657-3

Fig. 2.

Fig. 2

Group-level results of Experiment 1 and Experiment 2. a Real-time sampling behavior of neuroadaptive Bayesian optimization and b group-level predictions across task space identify the Tower of London and Deductive Reasoning tasks to be optimal for dissociating the dFPN from the vFPN. c Real-time sampling behavior of optimization algorithm and d group-level predictions (observations colored by subject) show difficulty levels 13–15 to be optimal for the Deductive Reasoning task. e Real-time sampling behavior of optimization algorithm for the Tower of London task suggests 6–7 steps to be optimal