Samuel McDougle^1,2, Zekun Sun¹; ¹Department of Psychology, Yale University, ²Wu Tsai Institute, Yale University

"Motor equivalence" refers to the fact that people can perform complex movements in multiple ways to achieve the same goal (e.g., signing their name with different effectors). This impressive ability suggests that the mind recruits abstract, high-level representations of actions that transcend lower-level muscle commands. However, evidence that these abstract representations actually exist, or contribute to motor skill, is lacking. For example, an overlearned visual template (e.g., a letter in your native alphabet) could help one perform novel visually-guided movements using online feedback control, without any need for abstraction. We reasoned that if abstractions contribute to real-world motor skills (e.g., handwriting), training the abstraction itself – but not lower-level muscle commands or visual templates – should improve skill. We compiled novel symbols from the Omniglot database, and generated videos that demonstrated to participants (n = 30) the hand movements required to write each symbol. In each task block, participants first imitated writing a target symbol a single time after viewing a demo. Next, they completed 3 training trials, where the symbol they attempted to copy was rotated by one of three extreme angles (90, 180, 270 degrees). Thus, training trials did not recruit the same muscle commands or visual outcomes as the initial learning trial. After training, participants again copied the symbol at its initial orientation. Performance was evaluated by computing both copying speed and error (the Procrustes distance, a measure of shape similarity between the symbol and participant copy). Participants’ performance improved following training (P < 0.0001). In other words, participants' handwriting improvements appeared to be driven by refining high-level, abstract representations of the required movements. Follow-up experiments (n = 200) replicated this result, and ruled out effects of visual familiarity and general motor practice. Our work suggests that the mind represents abstract visuomotor plans to guide skilled behavior.

Acknowledgements: This work is supported by NIH grant R01 NS13292