Flow-field inference from neural data using deep recurrent networks

Timothy Doyeon Kim, Thomas Zhihao Luo, Tankut Can, Kamesh Krishnamurthy, Jonathan W. Pillow, Carlos D Brody
Proceedings of the 42nd International Conference on Machine Learning, PMLR 267:30567-30590, 2025.

Abstract

Neural computations underlying processes such as decision-making, working memory, and motor control are thought to emerge from neural population dynamics. But estimating these dynamics remains a significant challenge. Here we introduce Flow-field Inference from Neural Data using deep Recurrent networks (FINDR), an unsupervised deep learning method for inferring low-dimensional, nonlinear, stochastic dynamics underlying neural population activity. Using spike train data from frontal brain regions of rats performing an auditory decision-making task, we demonstrate that FINDR performs competitively with existing methods in capturing the heterogeneous responses of individual neurons. When trained to disentangle task-relevant and irrelevant activity, FINDR uncovers interpretable low-dimensional dynamics. These dynamics can be visualized as flow fields and attractors, enabling direct tests of attractor-based theories of neural computation. We suggest FINDR as a powerful method for revealing the low-dimensional task-relevant dynamics of neural populations and their associated computations.

Cite this Paper

BibTeX
@InProceedings{pmlr-v267-kim25z, title = {Flow-field inference from neural data using deep recurrent networks}, author = {Kim, Timothy Doyeon and Luo, Thomas Zhihao and Can, Tankut and Krishnamurthy, Kamesh and Pillow, Jonathan W. and Brody, Carlos D}, booktitle = {Proceedings of the 42nd International Conference on Machine Learning}, pages = {30567--30590}, year = {2025}, editor = {Singh, Aarti and Fazel, Maryam and Hsu, Daniel and Lacoste-Julien, Simon and Berkenkamp, Felix and Maharaj, Tegan and Wagstaff, Kiri and Zhu, Jerry}, volume = {267}, series = {Proceedings of Machine Learning Research}, month = {13--19 Jul}, publisher = {PMLR}, pdf = {https://raw.githubusercontent.com/mlresearch/v267/main/assets/kim25z/kim25z.pdf}, url = {https://proceedings.mlr.press/v267/kim25z.html}, abstract = {Neural computations underlying processes such as decision-making, working memory, and motor control are thought to emerge from neural population dynamics. But estimating these dynamics remains a significant challenge. Here we introduce Flow-field Inference from Neural Data using deep Recurrent networks (FINDR), an unsupervised deep learning method for inferring low-dimensional, nonlinear, stochastic dynamics underlying neural population activity. Using spike train data from frontal brain regions of rats performing an auditory decision-making task, we demonstrate that FINDR performs competitively with existing methods in capturing the heterogeneous responses of individual neurons. When trained to disentangle task-relevant and irrelevant activity, FINDR uncovers interpretable low-dimensional dynamics. These dynamics can be visualized as flow fields and attractors, enabling direct tests of attractor-based theories of neural computation. We suggest FINDR as a powerful method for revealing the low-dimensional task-relevant dynamics of neural populations and their associated computations.} }
Endnote
%0 Conference Paper %T Flow-field inference from neural data using deep recurrent networks %A Timothy Doyeon Kim %A Thomas Zhihao Luo %A Tankut Can %A Kamesh Krishnamurthy %A Jonathan W. Pillow %A Carlos D Brody %B Proceedings of the 42nd International Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2025 %E Aarti Singh %E Maryam Fazel %E Daniel Hsu %E Simon Lacoste-Julien %E Felix Berkenkamp %E Tegan Maharaj %E Kiri Wagstaff %E Jerry Zhu %F pmlr-v267-kim25z %I PMLR %P 30567--30590 %U https://proceedings.mlr.press/v267/kim25z.html %V 267 %X Neural computations underlying processes such as decision-making, working memory, and motor control are thought to emerge from neural population dynamics. But estimating these dynamics remains a significant challenge. Here we introduce Flow-field Inference from Neural Data using deep Recurrent networks (FINDR), an unsupervised deep learning method for inferring low-dimensional, nonlinear, stochastic dynamics underlying neural population activity. Using spike train data from frontal brain regions of rats performing an auditory decision-making task, we demonstrate that FINDR performs competitively with existing methods in capturing the heterogeneous responses of individual neurons. When trained to disentangle task-relevant and irrelevant activity, FINDR uncovers interpretable low-dimensional dynamics. These dynamics can be visualized as flow fields and attractors, enabling direct tests of attractor-based theories of neural computation. We suggest FINDR as a powerful method for revealing the low-dimensional task-relevant dynamics of neural populations and their associated computations.
APA
Kim, T.D., Luo, T.Z., Can, T., Krishnamurthy, K., Pillow, J.W. & Brody, C.D.. (2025). Flow-field inference from neural data using deep recurrent networks. Proceedings of the 42nd International Conference on Machine Learning, in Proceedings of Machine Learning Research 267:30567-30590 Available from https://proceedings.mlr.press/v267/kim25z.html.

Related Material