What moves you? The Role of Enhanced Expectancies and Reward Processing in Motor Performance and Learning

Abstract

This dissertation describes a research program focused on dissecting the contributions of motivation and rewards to motor skill acquisition and learning. Two prominent theories served as a framework for this work: the OPTIMAL theory of motor learning (Wulf & Lewthwaite, 2016) and reinforcement learning (RL) theory (Rescorla & Wagner, 1972). The former claims that enhancing learners’ expectancies for future positive outcomes and perception of autonomy leads to higher levels of motivation, which strengthens the coupling of goals to actions, culminating in better motor performance and learning. It is hypothesized that expectancies reflect reward anticipation, which might explain the learning benefits given the association between rewards and the release of dopamine, a neurotransmitter that plays a crucial role in movement, reward processing, and memory consolidation. Rewards, and more specifically, reward-prediction errors (the difference between actual and anticipated reward) are the major driver of RL theory. According to this theory, humans adjust their behavior based on reward-prediction errors to maximize the likelihood of receiving rewards. In short, behaviors that lead to rewards are more likely to re-occur in the future, whereas behaviors that are not rewarded are less likely to re-occur in the future. Together, OPTIMAL and RL theory make predictions about how motivation and rewards affect short- and long-term behavior adaptation. Through a series of studies that combine behavioral, psychophysiological, and meta-analytical research, the present research program investigated how these predictions apply to motor skill acquisition and retention. The first paper (chapter 1), published in the Journal of Motor Learning and Development, sought to tease apart the contributions of extrinsic rewards, a common means to enhance learners’ expectancies, to promoting learning of two components of a motor skill, namely the action selection (i.e., what to do) and action execution component (i.e., how to execute the action). Results showed that giving learners extrinsic rewards during practice did not improve their ability to choose the correct action and execute the movement accurately. Interestingly, learners’ self-reported motivation, irrespective of whether they could receive extrinsic rewards, did predict action selection and action execution performance. The second paper (chapter 2), published in the journal International Review of Sport and Exercise Psychology, used a meta-analytic approach to examine the effect of enhanced expectancies on motor learning and whether the effect depended on the type of manipulation adopted. Results revealed a medium-sized, positive effect of enhanced expectancies on motor learning, which varied as a function of the type of manipulation, and is likely overestimated due to the presence of small-study effects and underpowered studies in the sample. The third paper (chapter 3), published in the journal Psychology of Sport and Exercise, investigated the mechanisms underlying the self-controlled feedback learning benefit. As postulated by OPTIMAL theory, increasing learners’ perception of autonomy leads to higher levels of motivation and consequent better performance and learning. One common autonomy support manipulation consists of giving learners control over their feedback schedule, which has been shown to enhance motor learning, though the underlying mechanisms are still unclear. Since motivational and information processing factors have been suggested as potential underpinnings, the second paper aimed to dissociate their contribution to the self-controlled feedback learning benefit. Results showed no effect of self-controlled feedback on learning, although self-reported motivation predicted post-test performance at the individual level, irrespective of whether learners controlled their feedback schedule. Finally, the fourth paper (chapter 4) investigated RL predictions and their underlying mechanisms in a motor learning context. Specifically, mixed-effect regression models were used to analyze the relationship between learners’ feedback-evoked electroencephalogram (EEG) activity (i.e., reward positivity; RewP) and their short- and long-term behavior adaptation. Results showed that RewP scaled with feedback about learners’ accuracy and explained adjustments in their performance, suggesting that it reflects reward-prediction errors. Moreover, although RewP was implicated in short-term performance adjustments, it did not predict long-term behavior adaptation. Taken together, the studies described in this dissertation provide evidence for some aspects of OPTIMAL theory, such as the relationship between learners’ motivation and their motor learning, but cast doubt on others, such as the benefit of increasing learners’ autonomy and their motor learning. Further, the final study provides evidence that the RewP can shed light on how RL mechanisms (reward-prediction errors) explain short-term performance adjustments when learning a complex motor skill, but indicates that other mechanisms may need to be considered to explain long-term behavior adaptation.