Skip to content
DraconiticResearch
Search Sources API
201,642 studies31 sourcesUpdated 12 min ago
Peer-reviewedCC BY-NC-NDNeuron

Sparse component analysis: A method that uncovers separable computations within neural population activity

  • Andrew J. Zimnik1
  • Xinyue An2
  • K. Cora Ames3
  • Andrew Ulmer2
  • Antonio H. Lara1
  • Abigail A. Russo1
  • Vladislav Susoy4
  • Laura N. Driscoll5
  • John Patrick Cunningham6
  • Liam Paninski6
  • Mark M. Churchland7
  • Joshua I. Glaser8
8 affiliations
  1. 1Department of Neuroscience, Columbia University Medical Center, New York, NY, USA; Zuckerman Institute, Columbia University, New York, NY, USA
  2. 2Department of Neurology, Northwestern University, Chicago, IL, USA; Interdepartmental Neuroscience Program, Northwestern University, Chicago, IL, USA
  3. 3Department of Neuroscience, Columbia University Medical Center, New York, NY, USA; Zuckerman Institute, Columbia University, New York, NY, USA; Grossman Center for the Statistics of Mind, Columbia University, New York, NY, USA; Center for Theoretical Neuroscience, Columbia University, New York, NY, USA
  4. 4Department of Physics, Harvard University, Cambridge, MA, USA; Center for Brain Science, Harvard University, Cambridge, MA, USA
  5. 5Department of Electrical Engineering, Stanford University, Stanford, CA, USA; Allen Institute for Neural Dynamics, Allen Institute, Seattle, WA, USA; Department of Neurobiology and Biophysics, University of Washington, Seattle, WA, USA
  6. 6Zuckerman Institute, Columbia University, New York, NY, USA; Grossman Center for the Statistics of Mind, Columbia University, New York, NY, USA; Center for Theoretical Neuroscience, Columbia University, New York, NY, USA; Department of Statistics, Columbia University, New York, NY, USA
  7. 7Department of Neuroscience, Columbia University Medical Center, New York, NY, USA; Zuckerman Institute, Columbia University, New York, NY, USA; Grossman Center for the Statistics of Mind, Columbia University, New York, NY, USA; Kavli Institute for Brain Science, Columbia University Medical Center, New York, NY, USA. Electronic address: [email protected]
  8. 8Department of Neurology, Northwestern University, Chicago, IL, USA; Department of Computer Science, Northwestern University, Evanston, IL, USA; National Institute for Theory and Mathematics in Biology, Chicago, IL, USA. Electronic address: [email protected]
Open at sourceDOICite

Abstract

In many neural populations, the computationally relevant signals are posited to be a set of "latent factors"-signals shared across many individual neurons. A given brain area may perform many computations, each associated with distinct factors, which can together compose an overall action. The methods for uncovering such structure typically require supervision, which can limit the discovery of novel aspects of activity. Here, we introduce sparse component analysis (SCA), an unsupervised approach. SCA facilitated surprisingly clear parcellations of neural activity across a range of behaviors, when seeking both linear and nonlinear embeddings. We applied SCA to motor cortex activity from reaching and cycling monkeys, single-trial imaging data from C. elegans, and activity from a multitask artificial network. SCA revealed both simple and unexpected instances where the overall population response was built compositionally from sets of factors with distinct computational roles.

Related

  1. Genotyping by Sequencing (GBS) uncovers genetic diversity and population structure for salicornia breeding programs in Iran
    Fatemeh Loni, Nayer Azam Khoshkholgh Sima, Mohammad Reza Ghaffari +2
    Scientific ReportsPeer-reviewedCC BY-NC-ND 4.0Full textGenetics & genomicsCrossref
  2. Dual-Component Nanoemulsion Containing Brugmansia suaveolens Leaf Extract and Copaiba Oil: Antimicrobial Activity and Preclinical Wound-Healing Potential
    Sandro Pinheiro da Costa, Verônica da Silva Cardoso, Alane Beatriz Vermelho +9
    Plants (Basel, Switzerland)Peer-reviewedCC BYFull textPlant biologyEurope PMC
  3. SOMATODENDRITIC VASOPRESSIN RELEASE COUPLES INTRINSIC EXCITABILITY TO POPULATION-LEVEL ACTIVITY
    M. K. Kirchner, F. C. Krane, I. A. Ahmed +1
    bioRxivPreprintAll rights reservedFull textNeurosciencebioRxiv
  4. Hippocampal theta states are linked to reconfiguration of population activity with respect to behavioural context
    Laura Masaracchia, Pablo Oyarzo, Felipe Fredes +1
    Communications BiologyPeer-reviewedCC BY 4.0Full textNeuroscienceCrossref
  5. DRIADA: an open-source Python toolkit for linking single-neuron selectivity and population activity to behavior
    Nikita Pospelov, Viktor Plusnin, Olga Rogozhnikova +7
    arXivPreprintCC BY-NC-SA 4.0Full textBioinformatics & computational biologyNeurosciencearXiv

Cite

BibTeXRIS
BibTeX
@article{zimnik2026sparse,
  title = {{Sparse component analysis: A method that uncovers separable computations within neural population activity}},
  author = {Andrew J. Zimnik and Xinyue An and K. Cora Ames and Andrew Ulmer and Antonio H. Lara and Abigail A. Russo and Vladislav Susoy and Laura N. Driscoll and John Patrick Cunningham and Liam Paninski and Mark M. Churchland and Joshua I. Glaser},
  journal = {Neuron},
  year = {2026},
  month = jul,
  doi = {10.1016/j.neuron.2026.05.022},
  url = {https://doi.org/10.1016/j.neuron.2026.05.022},
  note = {CC BY-NC-ND},
}
RIS
TY  - JOUR
TI  - Sparse component analysis: A method that uncovers separable computations within neural population activity
AU  - Andrew J. Zimnik
AU  - Xinyue An
AU  - K. Cora Ames
AU  - Andrew Ulmer
AU  - Antonio H. Lara
AU  - Abigail A. Russo
AU  - Vladislav Susoy
AU  - Laura N. Driscoll
AU  - John Patrick Cunningham
AU  - Liam Paninski
AU  - Mark M. Churchland
AU  - Joshua I. Glaser
PY  - 2026
DA  - 2026/07/01
JO  - Neuron
DO  - 10.1016/j.neuron.2026.05.022
UR  - https://doi.org/10.1016/j.neuron.2026.05.022
AB  - In many neural populations, the computationally relevant signals are posited to be a set of "latent factors"-signals shared across many individual neurons. A given brain area may perform many computations, each associated with distinct factors, which can together compose an overall action. The methods for uncovering such structure typically require supervision, which can limit the discovery of novel aspects of activity. Here, we introduce sparse component analysis (SCA), an unsupervised approach. SCA facilitated surprisingly clear parcellations of neural activity across a range of behaviors, when seeking both linear and nonlinear embeddings. We applied SCA to motor cortex activity from reaching and cycling monkeys, single-trial imaging data from C. elegans, and activity from a multitask artificial network. SCA revealed both simple and unexpected instances where the overall population response was built compositionally from sets of factors with distinct computational roles.
SN  - 0896-6273
SN  - 1097-4199
ER  -