So honored to have been named a 2025 Rising Star of Neuroscience by @thetransmitter.bsky.social! 😁

Big thanks to @avramholmes.bsky.social and @ambrains.bsky.social for the nomination 🤗, as well as to my past and present mentors for helping me get here! ❤️

www.thetransmitter.org/early-career...

Finally, because we can’t resist studying individual differences, we examined whether our approach improved the prediction of various cognitive scores from the HCP. Across multiple cognitive measures, we found that it did, demonstrating the translational value of our approach.

Importantly, these multi-scale effects were replicated in a second human dataset (www.nature.com/articles/s41...) and the mouse brain! (www.nature.com/articles/nat...). This result shows that our model uncovers regional variations in INTs that are conserved across species.

Using Allen gene data, as well as newly developed approaches from the @holmeslab-bhi.bsky.social (www.nature.com/articles/s41...), we found that our model-based INTs tracked cortical variation in SST and PVALB genes, including diverging concentration gradients of chandelier and basket cells!

Next, we examined whether our model-based INTs corresponded with empirical INTs, measured using fMRI, and found that they do! This result validates our approach by showing that slower time scales in our optimized model correspond to slower empirically measured time scales.

To understand these results, we first found that our optimized model was controllable from significantly fewer control nodes than the standard NCT model (A), which was driven by optimized INT maps giving rise to more diverse connectome dynamics (wider range of eigenmodes; B).

We uncover spatially patterned and variable maps of INTs (B), and discover that such maps require less control energy to transition between states (C), indicating that our model significantly improved whole-brain structure-function coupling (lower energy = better coupling).

We applied automatic differentiation to optimize regions’ self-inhibition to facilitate changes in empirical brain states. This algorithm discovers subject- & context-specific whole-brain maps of INTs that mediate the relationship between structural & functional connectivity.

A limitation of NCT is that it trades biophysical realism for simplicity by making a set of simplifying assumptions. One of those assumptions relates to regions’ self-inhibition, which is set uniformly in NCT, rendering regions’ intrinsic neural timescales fixed across the crtx.

NCT is a flexible and quantitative framework that we are continuing to develop (www.nature.com/articles/s41...) that allows us to examine how the topology and structure of the connectome enables and constrains changes to large-scale brain activity.