Figure 1.

Reaching paradigm. A, Participants reached to an array of targets using a robotic manipulandum. B, Time course of a normal trial. Participants reached at a single target and earned money based on their performance speed. If they were too slow (MTRT < ${\mathit{\tau }}_2$), a message “Too slow!” appeared instead of the reward information. Transition times are indicated below for each screen. A uniform distribution was used for the transition time jitter. C, Time course of a distractor trial. Occasionally, a distractor target appeared, indicated by a color different from the starting position. Participants were told to wait for the second, correct target to appear and reach toward the latter. D, The faster participants completed their reach to the target, the more money they were rewarded. The speed of the the response was quantified as the sum of movement time and reaction time (i.e., MTRT), and the function mapping MTRT to reward varied based on two parameters ${\mathit{\tau }}_1$ and ${\mathit{\tau }}_2$. ${\mathit{\tau }}_1$ and ${\mathit{\tau }}_2$ enabled the reward function to adjust throughout the task as a function of individual performance history, to ensure all participants received a similar amount of reward (see Task design). Top, Bottom, How how the function varied as a function of ${\mathit{\tau }}_1$ (${\mathit{\tau }}_2$ fixed at 800 ms) and ${\mathit{\tau }}_2$ (${\mathit{\tau }}_1$ fixed at 400 ms), respectively, for a 10 p trial.