Welcome to our Neural Dynamics Lab

————————————— Our work is on the beautiful, nonlinear dynamics of neurons, neural networks, and neural populations. These dynamics are richly varied from setting to setting — at times governed by mechanisms we can distill and explain — and at times still eluding our best analytical tools. 

 Best of all these dynamics reflect learning – over minutes, lifetimes, and evolutionary time – and so invite us to decode their uniquely biological version of intelligence. —————————————

Making progress here means coming together across fields and communities.  Every day we work hard to bring together dynamical systems, neural networks, probability and AI.  And we savor the good fortune of collaborating with fellow theorists of many different backgrounds, and with experimental groups at UW, the Allen Institute, and beyond.

We are grateful to the taxpayers of the US and the state of Washington who make public science possible.  Thank you!   And we thank the institutions that support our work, including the NIH, the Burroughs-Wellcome Fund, the Allen Institute, the NSF, the Swartz Foundation, the Simons Foundation, the Shanahan Foundation, the UW Center for Neurotechnology, and UWIN.

We recognize the high importance of equity with respect to race and ethnicity, gender and gender identity, sexual orientation, economic and social status, religion, ability, other backgrounds, and commit to nondiscrimination and inclusion in our lab work and culture.

How We Work

Interconnected

At UW, our lab bridges a vibrant department and a research center which we’re delighted to have helped to start.


With wonderful partners and leaders at the Allen Institute, with whom we’ve been proudly affiliated for more than a decade, this makes for daily life immersed in a city-wide collaboratory.


Community

Getting to do this work together is a tremendous gift.  We rejoice in this experience and do our best to support and celebrate one another every day.


The Lab Today

Eric Shea-Brown  (PI)

(PI, co-director of UW CNC, and collaborator with all members of our lab)

Ziyu Lu

(dynamics of large-scale neural states)

Yiliu Wang

(Inferring state-dependent dynamics using statistical machine learning)

Zihan Zhang

(local and coarse-grained gradient based methods for training neural networks)

Shirui Chen

(recurrent neural network dynamics and distributional encoding)

James Halzelden

(control and operator perspectives on learning in neural networks)

Tim Kim

(Linking connectomics and nonlinear dynamics in neural networks)

Aditya Deole

(control algorithms for mesoscopic neural dynamics)

Tailai Li

(dynamical systems models for brain-computer interfaces)

Hanson Mo

(biophysical timescales and learning regimes in neural networks)

Research Highlights

Biological learning and credit assignment

Learning in neural networks means assigning the right values to thousands to trillions or more of individual connections, so that the network as a whole produces the desired behavior.  This is the famous problem of “credit assignment”: how can one figure out how to change individual connections so that the network’s behavior improves?   With Uygar Sumbul (Allen), Steven Smith (Stanford), and colleagues, we’re working on how the brain’s multilayered connectomes – which continuously signal via dozens of mysterious genetically encoded neuromodulators -- may solve this problem in a uniquely biological way. Liu et al 1, Liu et al 2, …

Neural representations and biological dynamics

Neural networks do more than solve the tasks they are trained on – they learn to map data so that it can be rapidly reused to solve new problems.  This means using network dynamics to build abstract models of data, and how this data corresponds to what matters in the outside world.  We’re working on a mathematical theory of how and why this mapping occurs – and are fascinated by the ways that this might link to neurobiology of plasticity.  Farrell et al 1 (or arXiv version); Recanatesi et al 1, Recanatesi et al 2, …

Publications

Our papers on Google Scholar

Talks, Podcasts and Interviews

A conversation on neuroscience and AI with Eric on the Brain Inspired Podcast.

Eric's brief chat about trying bridge math and biology with Grey Matters.

A conversation about what motivates us at MIT’s CBBM.

Eric on learning and the other connectome at VVTNS.

Eric on dimensionality and coding at COSYNE.

News Pieces

A Simons Foundation profile of our work on dimensionality and dynamics.

An Allen Institute News piece on our work on the neuromodualtory connectome and learning.

Our take on connectivity and dynamics in SIAM news.

Coming Together

We’re proud and delighted to be part of the UW Computational Neuroscience Center, which we worked with Adrienne Fairhall and colleagues to help found.