Figure 7.

Cost function for monkeys and representative human subject for different reward contingencies. The reward internal to the subjects is not accessible and thus a free parameter. For each subject, we compute the cost function for various settings of this free parameter (reward always 1 for correct decisions; punishment for incorrect decisions either 0, 1, or 2). The results do not qualitatively depend on the choice of these parameters. The estimated cost function c(t) itself is per second, such that the total cost for making a decision at time t is the area underneath the cost function up to time t. For the two monkeys, the gray dashed line represents the minimum reward time distribution. The minimum reward time relates to the reaction time of the decision maker. It is here mapped into the decision time by subtracting the nondecision time as estimated for each monkey separately. This estimate is a random variable, such that the minimum reward time also becomes a random variable in the decision time domain.