Reconfigurations of cortical manifold structure during reward-based motor learning

Reviewer #1 (Public Review):

The authors investigated how global brain activity varied during reward-based motor learning. During early learning, they found increased covariance between the sensorimotor and dorsal attention networks, coupled with reduced covariance between the sensorimotor and default mode networks; during late learning, they found the opposite pattern. Individual learning performance varied only with changes in the dorsal attention network. The authors certainly used a wide variety of valuable, state-of-the-art techniques to interrogate whole-brain networks and extract the key components of learning behavior. However, the findings are incomplete, tempered by potential confounds in the experimental design. As such, the underlying claim regarding how these networks jointly support reward-based motor learning is unclear.

Reviewer #2 (Public Review):

This useful investigation of learning-driven dynamics of cortical and some subcortical structures combines a novel in-scanner learning paradigm with interesting analysis approaches. The new task for reward-based motor learning is highly compelling and goes beyond the current state-of-the-art, but it is incomplete with respect to examining different signatures of learning, clarifying probed learning processes, and investigating changes in all relevant subcortical structures is incomplete and would benefit from more rigorous approaches. With the rationale and data presentation strengthened this paper would be of interest to neuroscientists working on motor control and reward-based learning.

Reviewer #3 (Public Review):

The manuscript of Nick and colleagues addresses the intriguing question of how brain connectivity evolves during reward-based motor learning. The concept of quantifying connectivity through changes in extraction and contraction across lower-dimensional manifolds is both novel and interesting and the presented results are clear and well-presented. Overall, the manuscript is a valuable addition to the field. The evidence supporting the presented findings is strong, though at times lacking rigorous statistical quantification. Nevertheless, there are several issues that require attention and clarification.

Author Response

We appreciate the thoughtful comments provided by the editor and reviewers. We were pleased to hear that they appreciated our work's contribution to the field of motor learning as well as our use of state-of-the-art analysis techniques.

We are currently preparing a comprehensive revision of our manuscript to address several of the recommendations of the reviewers. It is our belief that this revision will not only strengthen our paper but also help clarify several areas that were highlighted by the reviewers.

To address the concerns regarding potential confounds in our experimental design, we will be providing a more detailed justification and rationale for the experimental design and analysis choices made during our study. It appears that some reviewers’ comments may stem from misunderstandings concerning certain details of our task and we will carefully revise these sections to ensure that the design and purpose of the study are unambiguous. We will also be improving our characterizations of subjects’ learning behavior, which we believe will clarify some of the reviewers comments and enhance the overall rigor of our analyses. Lastly, we will be dealing with all concerns related to the statistical quantification of our results.

We appreciate the opportunity to improve our manuscript for eLife and are eager to provide a revision that satisfies the majority of the reviewers’ recommendations