Fig. 1.

Example task and simulation parameters. The example illustrates a class of motor learning tasks involving binary reward feedback based on one dimension of the movement, with no feedback on error size and sign. Left panel: simulation parameters consist of two task parameters and three learner parameters. Participants aim for a target but might end up somewhere else due to motor noise and exploration in their movement. Upon the next attempt, the learner may adjust her aim point with a learning fraction. The gray screen blocks the sight of the hand, so that task feedback is limited to binary reward feedback. Right panel: three types of the reward criterion task parameter. A fixed reward criterion only rewards movements at the target. An adaptive reward criterion additionally rewards movements that are closer to the target than the mean or median of the past five attempts. A random reward criterion randomly rewards 50% of the movements randomly, independent of performance