# Robust modeling of within- and across-area population dynamics using recurrent neural networks - Admin Suppl

> **NIH NIH RF1** · EMORY UNIVERSITY · 2023 · $312,984

## Abstract

Abstract
Over the past several decades, the ability to record from large populations of neurons (e.g., multi-electrode
arrays, neuropixels, calcium imaging) has increased exponentially, opening new avenues for understanding
the brain. While these data promise a qualitatively different view of brain activity than was previously possible,
interpreting them will require the development of new analytical tools. A framework is provided by the tools of
dynamical systems, which offer the means to uncover coordinated time-varying activation patterns expressed
across an interconnected network of recorded neurons, and to characterize how these patterns relate to
behavior. This framework has provided fundamental new insights into information processing in cortical
circuits, including those underlying motor, sensory, and cognitive processes. However, analytical approaches
to uncovering dynamics have typically been developed and tested in specific brain areas, for limited behaviors,
in restricted behavioral settings.
 We are addressing the challenge of generalized applicability to diverse brain areas through a powerful
new open-source toolkit for automated discovery of neural population dynamics. Our approach, AutoLFADS,
uses recurrent neural networks to uncover dynamics and allows straightforward application to data from
different brain areas and behaviors. We have now validated the generalizability of AutoLFADS using
electrophysiology data from monkeys and rats, and two-photon imaging data from mice.
 Our objective in this supplement is to enable neuroscientist of from a variety of backgrounds to easily
and robustly our powerful tools for modeling the dynamics of large populations of neurons. The specific aims of
this supplement are 1) technical development to increase the flexibility of AutoLFADS and decrease its
computational requirements, to facilitate broad dissemination, and 2) working directly with end users to
increase adoption and lower barriers to entry, first through usability case studies with individual labs, and then
through virtual workshops, office hours, and one-on-one sessions with interested end users. All generated
content (code, documentation, tutorial notebooks, workshop videos) will be made freely available online.
 Through the aims of this supplement, we will produce a powerful, general, and easy-to-use resource for
neuroscientists to probe the dynamics of large neural populations. In line with the goals of the BRAIN initiative,
the resulting tools will be uniquely suited to help advance our understanding of the complex, nonlinear, and
nonintuitive activity of large neural populations that underlies brain function and behavior.

## Key facts

- **NIH application ID:** 10694582
- **Project number:** 3RF1DA055667-01S1
- **Recipient organization:** EMORY UNIVERSITY
- **Principal Investigator:** Lee Miller
- **Activity code:** RF1 (R01, R21, SBIR, etc.)
- **Funding institute:** NIH
- **Fiscal year:** 2023
- **Award amount:** $312,984
- **Award type:** 3
- **Project period:** 2022-10-01 → 2024-08-31

## Primary source

NIH RePORTER: https://reporter.nih.gov/project-details/10694582

## Citation

> US National Institutes of Health, RePORTER application 10694582, Robust modeling of within- and across-area population dynamics using recurrent neural networks - Admin Suppl (3RF1DA055667-01S1). Retrieved via AI Analytics 2026-05-22 from https://api.ai-analytics.org/grant/nih/10694582. Licensed CC0.

---

*[NIH grants dataset](/datasets/nih-grants) · CC0 1.0*