A Caltech Library Service

Inferring brain-wide interactions using data-constrained recurrent neural network models

Perich, Matthew G. and Arlt, Charlotte and Soares, Sofia and Young, Megan E. and Mosher, Clayton P. and Minxha, Juri and Carter, Eugene and Rutishauser, Ueli and Rudebeck, Peter H. and Harvey, Christopher D. and Rajan, Kanaka (2020) Inferring brain-wide interactions using data-constrained recurrent neural network models. . (Unpublished)

[img] PDF (March 11, 2021) - Submitted Version
Creative Commons Attribution Non-commercial No Derivatives.


Use this Persistent URL to link to this item:


Behavior arises from the coordinated activity of numerous anatomically and functionally distinct brain regions. Modern experimental tools allow unprecedented access to large neural populations spanning many interacting regions brain-wide. Yet, understanding such large-scale datasets necessitates both scalable computational models to extract meaningful features of inter-region communication and principled theories to interpret those features. Here, we introduce Current-Based Decomposition (CURBD), an approach for inferring brain-wide interactions using data-constrained recurrent neural network models that directly reproduce experimentally-obtained neural data. CURBD leverages the functional interactions inferred by such models to reveal directional currents between multiple brain regions. We first show that CURBD accurately isolates inter-region currents in simulated networks with known dynamics. We then apply CURBD to multi-region neural recordings obtained from mice during running, macaques during Pavlovian conditioning, and humans during memory retrieval to demonstrate the widespread applicability of CURBD to untangle brain-wide interactions underlying behavior from a variety of neural datasets.

Item Type:Report or Paper (Discussion Paper)
Related URLs:
URLURL TypeDescription Paper ItemCode
Perich, Matthew G.0000-0001-9800-2386
Mosher, Clayton P.0000-0002-9213-3059
Minxha, Juri0000-0003-4942-3269
Rutishauser, Ueli0000-0002-9207-7069
Rudebeck, Peter H.0000-0002-1411-7555
Harvey, Christopher D.0000-0001-9850-2268
Rajan, Kanaka0000-0003-2749-2917
Additional Information:The copyright holder for this preprint is the author/funder, who has granted bioRxiv a license to display the preprint in perpetuity. It is made available under a CC-BY-NC-ND 4.0 International license. Version 1 - December 21, 2020; Version 2 - March 11, 2021. C.A. is supported by a Louis Perry Jones, an Alice and Joseph Brooks, and a Mahoney Postdoctoral Fellowship. S.S. is supported by an European Molecular Biology Organization (EMBO) Postdoctoral Fellowship. U.R. is supported by the National Institutes of Health (NIH) (R01 MH110831 and U01 NS117839) and National Science Foundation (NSF) (BCS-1554105). P.R. is supported by a National Institute of Mental Health BRAINS award (R01MH110822, R01MH118638) and a young investigator grant from the Brain and Behavior Foundation (NARSAD). C.D.H. is supported by an NIH Director’s Pioneer Award (DP1 MH125776) and NINDS R01 NS089521. K.R. is supported by NIH BRAIN Initiative (R01 EB028166), James S. McDonnell Foundation’s Understanding Human Cognition Scholar Award, and NSF FOUNDATIONS award (NSF1926800). We would like to thank Dr. Juan A. Gallego, Dr. Raeed H. Chowdhury, Dr. Larry F. Abbott, Dr. Sheila Cherry, and Dr. Christian D. Marton for comments on earlier drafts of this manuscript. We are indebted to and inspired by Dr. C. R. Rajan and Prabha Rajan for sharing their undying love of learning. Author Contributions: M.G.P. and K.R. conceived of the method. M.G.P. analyzed datasets and generated figures. M.G.P. and K.R. wrote the manuscript. M.G.P., C.A., S.S., C.M., J.M., U.R., P.R., C.D.H., and K.R. edited the manuscript. E.C. ran simulations and implemented code. C.A., S.S., and C.D.H. collected the mouse dataset. M.E.Y., C.M., and P.R. provided the monkey dataset. J.M. and U.R. provided the human dataset. The authors declare no competing interests. Code availability: All modeling and analysis in this manuscript was done in Matlab (The Mathworks, Inc.). Matlab and Python code to train multi-region Model RNNs based on multi-region experimental recordings and perform CURBD using the inferred interactions is available at:
Funding AgencyGrant Number
Louis Perry Jones Postdoctoral FellowshipUNSPECIFIED
Alice and Joseph Brooks Postdoctoral FellowshipUNSPECIFIED
Mahoney Postdoctoral FellowshipUNSPECIFIED
European Molecular Biology Organization (EMBO)UNSPECIFIED
NIHR01 MH110831
NIHU01 NS117839
Brain and Behavior FoundationUNSPECIFIED
NIHDP1 MH125776
NIHR01 NS089521
NIHR01 EB028166
James S. McDonnell FoundationUNSPECIFIED
Record Number:CaltechAUTHORS:20201222-101718854
Persistent URL:
Official Citation:Inferring brain-wide interactions using data-constrained recurrent neural network models. Matthew G Perich, Charlotte Arlt, Sofia Soares, Megan E. Young, Clayton P. Mosher, Juri Minxha, Eugene Carter, Ueli Rutishauser, Peter H. Rudebeck, Christopher D. Harvey, Kanaka Rajan. bioRxiv 2020.12.18.423348; doi:
Usage Policy:No commercial reproduction, distribution, display or performance rights in this work are provided.
ID Code:107254
Deposited By: Tony Diaz
Deposited On:22 Dec 2020 18:45
Last Modified:16 Nov 2021 19:00

Repository Staff Only: item control page