UC San Diego

Motor skill learning, or learning a sequence of movements, is a fundamental process that occurs throughout one’s life, from learning how to pick up your coffee cup to learning to play Für Elise. The primary perspective of when motor skill learning occurs is that it occurs during rest periods, offline, rather than concurrently with practice, online. However, those interpretations have conflated learning with performance. In this dissertation, I will propose an account that assumes learning occurs online, concurrently with practice, and that reactive inhibition, the worsening of performance as task performance continues, are sufficient to explain motor skill learning. Chapter 1 provides evidence that this alternate online account is sufficient to explain the ordinal results of performance improvements following a rest period. However, this experiment was unable to rule out the offline learning account. Chapter 2 implements the underlying assumptions of both the online, offline, and hybrid accounts into 10 different computational models. Overall, a hybrid model was able to best fit the data suggesting that both online and offline learning occur. Further, we demonstrated the necessity of assuming reactive inhibition in any interpretation of performance data. These models also indicated that there may be learning differences resulting from training schedules. Chapter 3 examined if learning rates were dependent on training schedules. Contrary to prior work, we found no evidence that training schedule influenced learning rates in a motor skill task where few errors are made. Taken together, my dissertation provides compelling evidence that learning occurs online, concurrently with the task, and challenges the predominant view that motor skill learning only occurs offline.