Figure 1.

Extrinsic and intrinsic reward paradigm. A, Subjects (n = 33) experienced both extrinsic and intrinsic rewards on each trial. A trial starts with subjects selecting from one or two available options each associated with an implicit extrinsic reward. One option on average leads to the larger reward (mean = 50, SD = 10) whereas the other leads to a lower reward (mean = 25, SD = 10) with a reversal every 19–23 trials. Four barriers then appear along the path to the outcome and a cursor appears at the bottom of the screen which automatically advances after a 1-s delay. Subjects press left and right keys to navigate around barriers, constituting a form of skilled performance that can be intrinsically rewarding. Successfully avoiding a barrier turns it green whereas contact with a barrier turns it red. There is no financial penalty for contact with barriers nor financial benefit for avoiding them. Earnings depend only on the outcome delivered at the end of the trial. After every two to three trials, subjects report their current happiness by moving a cursor on a rating line. B, Probability of choice to the initial high-reward option averaged across subjects (n = 33) in black. Shaded areas correspond to SEM. Gray vertical bands represent intervals where probability reversals could occur. C, D, Happiness trajectories and model fits for a computational model with both reward and performance parameters are displayed for two example subjects (C, r² = 0.45; D, r² = 0.42). Also see Figures 2, 3 and Tables 1, 2.