Implicit reward-based motor learning

Abstract

Binary feedback, providing information solely about task success or failure, can be sufficient to drive motor learning. While binary feedback can induce explicit adjustments in movement strategy, it remains unclear if this type of feedback also induce implicit learning. We examined this question in a center-out reaching task by gradually moving an invisible reward zone away from a visual target to a final rotation of 7.5° or 25° in a between-group design. Participants received binary feedback, indicating if the movement intersected the reward zone. By the end of the training, both groups modified their reach angle by about 95% of the rotation. We quantified implicit learning by measuring performance in a subsequent no-feedback aftereffect phase, in which participants were told to forgo any adopted movement strategies and reach directly to the visual target. The results showed a small, but robust (2-3°) aftereffect in both groups, highlighting that binary feedback elicits implicit learning. Notably, for both groups, reaches to two flanking generalization targets were biased in the same direction as the aftereffect. This pattern is at odds with the hypothesis that implicit learning is a form of use-dependent learning. Rather, the results suggest that binary feedback can be sufficient to recalibrate a sensorimotor map.

Fig. 1

Schematic outline of key hypotheses on implicit reward-based motor learning. a. Schematic of a participant in the experimental apparatus. b. Training phase. Participants made center-out reaching movements from a white starting circle to a black training target. A pleasant auditory “ding” was provided when the movement passed within the reward zone (green arch); otherwise, an unpleasant “buzz” was played. c. No-feedback phase. Participants were instructed to reach directly to a visual target. The target appeared at the training location or one of two probe locations (+−15°). Participants were instructed to forgo any strategy adopted during the training phase. Left panel shows implicit learning as measured by an aftereffect, defined as a change in hand angle for reaches to the training target from pre-training (translucent hand) to post-training (solid hand). Middle panel shows probe target reaching predictions for the implicit recalibration hypothesis. Reaches will be biased in same direction for both probe targets independent of size of the perturbation. Right panel shows probe target reaching predictions for the use-dependent learning hypothesis. For the Small perturbation condition, the biases will be in opposite directions since the reaches during training fall between the two probe locations. For the Large perturbation condition, the direction of the bias for the probe target nearest the reward zone will depend on the degree of learning (example here is for a participant who shows full learning)

Fig. 2

The effect of binary reward feedback on reaching. a, b. The training phase. Gradually changing the rewarded hand angles (green zone) leads to learning, as indicated by the change in reach angle. We plot the median (solid thick lines) over all participants with the interquartile range (opaque lines) for the Small perturbation group (a) and Large perturbation group (b). Note that the vertical axes are scaled to the perturbation size. For display purposes, the curves are smoothed with a running average with a window size of 10 trials. c, d. Aftereffect as a function of the final learning for both groups. Each grey dot corresponds to a participant and error bars indicate the interquartile range per group. The green lines indicate perturbation size

Fig. 3

Aftereffects and generalization of learning. Bars and error bars indicate medians and interquartile ranges. a. Reaching biases for the training target (black) and two probe targets (see Fig 1c). Thin lines indicate data from individual participants. b. Asymmetry in reaching biases to probe targets. Dots indicate the individual participants in the two groups. c. Generalization quantified as a percentage of the aftereffect. The participants with large positive and negative values are the ones with a small aftereffect