A long-standing question in psych and neuro is what serves as a dominant "teaching signal" over which to optimize when learning new skills. Instead of approaching this behaviourally, we compared monkey neural recordings to modelling predictions under the same objective function using MotorNet.

We found that models trained with RL aligned much better to monkey data in an matched reaching task. This was true with a crude geometrical metric (CCA) and dynamics metric (DSA) over tasks/datasets, and monkeys.