A theory of multineuronal dimensionality, dynamics and measurement
Peiran Gao, Eric M. TrautmanniD, Byron M. YuiD, Gopal Santhanam, Stephen I. RyuiD, K.V. ShenoyiD, Surya GanguliiD
Abstract
In many experiments, neuroscientists tightly control behavior, record many trials, and obtain trial-averaged firing rates from hundreds of neurons in circuits containing billions of behaviorally relevant neurons. Di-mensionality reduction methods reveal a striking simplicity underlying such multi-neuronal data: they can be reduced to a low-dimensional space, and the resulting neural trajectories in this space yield a remarkably insightful dynamical portrait of circuit computation. This simplicity raises profound and timely conceptual questions. What are its origins and its implications for the complexity of neural dynamics? How would the situation change if we recorded more neurons? When, if at all, can we trust dynamical portraits obtained from measuring an infinitesimal fraction of task relevant neurons? We present a theory that answers these questions, and test it using physiological recordings from reaching monkeys. This theory reveals conceptual insights into how task complexity governs both neural dimensionality and accurate recovery of dynamic portraits, thereby providing quantitative guidelines for future large-scale experimental design.
02 /Specification & Classification
Click any tag to browse matching papers- Citations
- 332
- Assay
- Research Focus
- Approach
Abstract summary
Claude Opus 5.5 · xhigh effort · 18 Oct 2026 · from the abstract · CC BY-NC-ND 4.0
Task complexity sets how many dimensions multineuronal activity occupies and when dynamical portraits recovered from a tiny fraction of neurons can be trusted; tested on monkey reaching recordings.
Experiments record hundreds of neurons from circuits that contain billions of behaviorally relevant ones, and dimensionality reduction makes the result look strikingly simple, leaving it unclear whether the low-dimensional trajectories describe the circuit or the sampling. The authors offer a theory that ties both the measured dimensionality and the recovery of dynamics to task complexity, and turn it into guidance for designing larger recordings.
Key results
ClaimedA theory is presented that explains why trial-averaged population activity appears low-dimensional and derives how that dimensionality depends on the complexity of the task the animal performs.
Framed as answers to the origins of the observed simplicity and to what would change with more recorded neurons; the abstract states no dimensionality values, scaling exponents or formulas. · Abstract
ClaimedThe same theory gives conditions under which a dynamical portrait estimated from a vanishingly small fraction of task-relevant neurons still reflects the underlying circuit dynamics.
Posed as the question of when such portraits can be trusted and said to be answered, but no recovery error, sample-size threshold or bound is quoted in the abstract. · Abstract
SuggestedThe predictions are checked against physiological recordings from monkeys performing reaching movements, and are said to yield quantitative guidelines for the design of large-scale experiments.
One empirical test setting is named; the abstract reports no neuron counts, sessions, fit statistics or the guideline values themselves. · Abstract
Methods, data and reproducibility
An analytical theory of how the dimensionality measured in neural population recordings, and the fidelity of the dynamical trajectories extracted from them, depend on task complexity and on the fraction of task-relevant neurons sampled. The theory is then tested against physiological recordings made in monkeys during a reaching task.
- physiological population recordings during reaching · monkey
Limitations (2)
- The empirical test uses a single paradigm, reaching in monkeys; no other species, task families or recording modalities are reported. · noted in this summary
- No quantitative result is given in the abstract (no dimensionality estimates, neuron counts or error bounds), so the promised design guidelines cannot be assessed from this text. · noted in this summary
Who should read:Read this if you plan large-scale neural population recordings, or interpret low-dimensional trajectories, and want a principled account of how many neurons a given task demands.
Reproducibility ledger
Nobody has reported trying to reproduce this paper yet. Did you replicate an experiment or run the code? Your report saves the next lab a week.
Sign in to file a reproducibility report.
Related papers
- Simultaneous, cortex-wide and cellular-resolution neuronal population dynamics reveal an unbounded scaling of dimensionality with neuron number
Jason Manley, Sihao Lu, … Alipasha Vaziri·NeuronPeer-reviewed·Mar 6, 2024
- Accurate estimation of neural population dynamics without spike sorting
Eric M. Trautmann, Sergey D. Stavisky, … K.V. Shenoy·NeuronPeer-reviewed·Jun 3, 2019
- Estimating the dimensionality of the manifold underlying multi-electrode neural recordings
Ege Altan, Sara A. Solla, … Eric Jon Perreault·PLoS Computational BiologyPeer-reviewed·Nov 29, 2021
- A predictive theory of experimental design for inferring neural population geometry in large-scale recordings
Itamar Daniel Landau, Gabriel C. Mel, … Surya Ganguli·bioRxiv·Jul 23, 2026
- Context-dependent low-dimensional neural dynamics unfold in distinct subspaces, dimensionality, and dynamical strength for natural walking and reaching
Alissa S. Ling, Michael P. Silvernagel, … Paul Nuyujukian·bioRxiv·Dec 23, 2025
- Stimuli reduce the dimensionality of cortical activity
Luca Mazzucato, Alfredo Fontanini, Giancarlo La Camera·bioRxiv·Feb 4, 2016