Shaping Dynamics With Multiple Populations in Low-Rank Recurrent Networks
Open accessThe DOI matches its Crossref record; 43 of the 43 references it cites were found in a registry or an index.
Abstract
An emerging paradigm proposes that neural computations can be understood at the level of dynamic systems that govern low-dimensional trajectories of collective neural activity. How the connectivity structure of a network determines the emergent dynamical system, however, remains to be clarified. Here we consider a novel class of models, gaussian-mixture, low-rank recurrent networks in which the rank of the connectivity matrix and the number of statistically defined populations are independent hyperparameters. We show that the resulting collective dynamics form a dynamical system, where the rank sets the dimensionality and the population structure shapes the dynamics. In particular, the collective dynamics can be described in terms of a simplified effective circuit of interacting latent variables. While having a single global population strongly restricts the possible dynamics, we demonstrate that if the number of populations is large enough, a rank R network can approximate any R-dimensional dynamical system.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
The source backs part of the claim, not all of it — or backs it for a different population, condition or comparison.
In the citing paper
“When RNN architecture is explicitly designed to include dedicated neural subpopulations, these subpopulations can improve model performance on particular types of tasks (Beiran et al., 2021; Dubreuil et al., 2022).”
From Remapping in a recurrent neural network model of navigation and context inference (2023) · cites this paper as Beiran et al., 2021
Experimental — accuracy not yet measured
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
The publisher hasn’t deposited this paper’s reference list.
Research works printed but not linked: 4
- Dubreuil, A., Valente, A., Beiran, M., Mastrogiuseppe, F., and Ostojic, S. (2020). Complementary roles of dimensionality and population structure in neural computations. bioRxiv , page 2020.07.03.185942.
- Eliasmith, C. and Anderson, C. H. C. H. (2003). Neural engineering : computation, representation, and dynamics in neurobiological systems . MIT Press.
- Nakatsukasa, Y. (2019). The low-rank eigenvalue problem. arXiv , page 1905.11490.
- Schuessler, F., Mastrogiuseppe, F., Dubreuil, A., Ostojic, S., and Barak, O. (2020b). The interplay between randomness and structure during learning in RNNs. arXiv , page 2006.11036.
Cited by
The role of population structure in computations through neural dynamics
Dubreuil, A., Valente, A., Beiran, M., Mastrogiuseppe, F., & Ostojic, S. (2022) · Nature Neuroscience · doi:10.1038/s41593-022-01088-4 · Cited by 237
“(36)); the dashed line shows the total input, which changes signs between the two contexts, leading to opposite responses.”
BackgroundSelf-citationFlexible multitask computation in recurrent networks utilizes shared dynamical motifs
Driscoll, L., Shenoy, K., & Sussillo, D. (2022) · doi:10.1101/2022.08.15.503870 · Cited by 217
“Recent work on subpopulation structure for the implementation of multiple complex tasks provides insight for these findings 21,29 .”
BackgroundFlexible multitask computation in recurrent networks utilizes shared dynamical motifs
Driscoll, L. N., Shenoy, K., & Sussillo, D. (2024) · Nature Neuroscience · doi:10.1038/s41593-024-01668-6 · Cited by 214
“…the all-to-all connectivity possible in our networks as well as the fact that no regularizations or constraints to induce modularity were employed in the training of the RNNs. Recent work on subpopulation structure for the implementation of multiple tasks provides insight for these findings 22,33 .”
Jazayeri, M., & Ostojic, S. (2021) · Current Opinion in Neurobiology · doi:10.1016/j.conb.2021.08.002 · Cited by 196
“Remarkably, these latent dynamics can be represented in terms of effective circuits that describe how incoming sensory latent variables interact with internally generated latent variables [86].”
BackgroundSelf-citationThe low-rank hypothesis of complex systems
Thibeault, V., Allard, A., & Desrosiers, P. (2024) · Nature Physics · doi:10.1038/s41567-023-02303-0 · Cited by 148
“Despite its frequent use—often implicit, but sometimes very explicit 35,36 —the low-rank hypothesis has yet to be verified experimentally for real networks in all their diversity.”
Parametric control of flexible timing through low-dimensional neural manifolds
Beirán, M., Meirhaeghe, N., Sohn, H., Jazayeri, M., & Ostojic, S. (2023) · Neuron · doi:10.1016/j.neuron.2022.12.016 · Cited by 103
Self-citationDiscovering cognitive strategies with tiny recurrent neural networks
Ji-An, L., Benna, M. K., & Mattar, M. G. (2025) · Nature · doi:10.1038/s41586-025-09142-4 · Cited by 64
Learning to represent continuous variables in heterogeneous neural networks
Darshan, R., & Rivkind, A. (2022) · Cell Reports · doi:10.1016/j.celrep.2022.110612 · Cited by 56
Early selection of task-relevant features through population gating
Barbosa, J., Proville, R. D., Rodgers, C. C., DeWeese, M. R., Ostojic, S., & Boubenec, Y. (2023) · Nature Communications · doi:10.1038/s41467-023-42519-5 · Cited by 69
“To obtain easilyinterpretable RNNs,we constrained the recurrent connectivity matrix to be of low rank, allowing us to reverse-engineer the mechanisms employed by the trained networks 18,19,30 .”
InfluentialSelf-citationMental programming of spatial sequences in working memory in the macaque frontal cortex
Tian, Z., Chen, J., Zhang, C., Min, B., Xu, B., & Wang, L. (2024) · Science · doi:10.1126/science.adp6091 · Cited by 62
“( ii) Multiple subspaces interact with each other through a population structure – enabled gain modulation mechanism ( 56 ).”
BackgroundComputational role of structure in neural activity and connectivity
Ostojic, S., & Fusi, S. (2024) · Trends in Cognitive Sciences · doi:10.1016/j.tics.2024.03.003 · Cited by 62
Self-citationNetwork dynamics underlying OFF responses in the auditory cortex
Bondanelli, G., Deneux, T., Bathellier, B., & Ostojic, S. (2021) · eLife · doi:10.7554/elife.53151 · Cited by 39
“(44) shows that the dynamics of a rank-R system are determined by the matrix of scalar products between left and right connectivity vectors, which we refer to as the overlap matrix (Schuessler et al., 2020; Beiran et al., 2020) J = VU.”
BackgroundSelf-citationForgetting Leads to Chaos in Attractor Networks
Pereira, U., Aljadeff, J., & Brunel, N. (2023) · Physical Review X · doi:10.1103/physrevx.13.011009 · Cited by 43
“fore, the mean input current due to the low-rank component in low-rank networks is equivalent to the mean input current of the retrieved memory in attractor networks, while the random component is equivalent to the quenched noise that arises from the stored memories (see [72]).”
BackgroundDe novo motor learning creates structure in neural activity that shapes adaptation
Chang, J. C., Perich, M. G., Miller, L. E., Gallego, J. Á., & Clopath, C. (2024) · Nature Communications · doi:10.1038/s41467-024-48008-7 · Cited by 28
“How these different regions interact to affect skill learning and adaptation is an ongoing area of study, and future work could use modular and area-speci fi c networks that are constrained by neural data to tease apart their contributions 61 – 63 .”
Valente, A., Ostojic, S., & Pillow, J. W. (2022) · Neural Computation · doi:10.1162/neco_a_01522 · Cited by 35
“Nonlinear low-rank RNNs without noise can be directly reduced to nonlinear latent dynamics with linear observations following the same mapping as in section 4 (Mastrogiuseppe & Ostojic, 2018; Schuessler et al., 2020; Beiran et al., 2021; Dubreuil et al., 2022) and therefore define a natural class of nonlinear LDS mode…”
BackgroundMethodInfluentialSelf-citationGeometry of population activity in spiking networks with low-rank structure
Cimeša, L., Ciric, L., & Ostojic, S. (2023) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1011315 · Cited by 25
Self-citationLinking neural manifolds to circuit structure in recurrent networks
Pezon, L., Schmutz, V., & Gerstner, W. (2026) · Neuron · doi:10.1016/j.neuron.2025.12.047 · Cited by 29
The impact of sparsity in low-rank recurrent neural networks
Herbert, E., & Ostojic, S. (2022) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1010426 · Cited by 29
“At the level of the population, the collective trajectory x(t) is embedded in a low-dimensional linear subspace [12, 13].”
BackgroundMethodInfluentialSelf-citationPrediction of neural activity in connectome-constrained recurrent networks
Beirán, M., & Litwin-Kumar, A. (2025) · Nature Neuroscience · doi:10.1038/s41593-025-02080-4 · Cited by 39
“To facilitate analysis, we studied a class of low-rank RNNs whose activity can be understood analytically 36,46,47 .”
InfluentialRelating local connectivity and global dynamics in recurrent excitatory-inhibitory networks
Shao, Y., & Ostojic, S. (2023) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1010855 · Cited by 28
“76 The global representation of connectivity instead refers to the situation where J is 77 defined as a low-rank matrix [30,31,33]: 78”
InfluentialSelf-citationAntagonistic inhibitory subnetworks control cooperation and competition across cortical spacePreprint
Mossing, D. P., Veit, J., Palmigiano, A., Miller, K. D., & Adesnik, H. (2021) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2021.03.31.437953 · Cited by 16
“Future work could seek to model this variability (Beiran et al., 2021), which might generate further hypotheses to be tested with more sophisticated optogenetic perturbations.”
BackgroundPrediction of neural activity in connectome-constrained recurrent networks
Beiran, M., & Litwin-Kumar, A. (2024) · doi:10.1101/2024.02.22.581667 · Cited by 16
Trained recurrent neural networks develop phase-locked limit cycles in a working memory task
Pals, M., Macke, J. H., & Barak, O. (2024) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1011852 · Cited by 35
Foucault, C., & Meyniel, F. (2021) · eLife · doi:10.7554/elife.71801 · Cited by 17
Structure of activity in multiregion recurrent neural networks
Clark, D. G., & Beirán, M. (2025) · Proceedings of the National Academy of Sciences · doi:10.1073/pnas.2404039122 · Cited by 30
Control Principles of Neural Dynamics Revealed by the Neurobiology of Timing
Stine, G. M., & Jazayeri, M. (2025) · Annual Review of Neuroscience · doi:10.1146/annurev-neuro-091724-015512 · Cited by 17
“A promising alternative to generic chaotic networks is low-rank neural networks (Beiran et al. 2021, 2023; Dubreuil et al. 2022; Mastrogiuseppe & Ostojic 2018).”
A ‘programming’ framework for recurrent neural networks
Beirán, M., Spencer-Salmon, C. A., & Rajan, K. (2023) · Nature Machine Intelligence · doi:10.1038/s42256-023-00674-w · Cited by 10
Symmetries and Continuous Attractors in Disordered Neural CircuitsPreprint
Clark, D. G., Abbott, L. F., & Sompolinsky, H. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.01.26.634933 · Cited by 10
STDP-based associative memory formation and retrieval
Yoon, H.-G., & Kim, P. (2023) · Journal of Mathematical Biology · doi:10.1007/s00285-023-01883-y · Cited by 7
“…to perform a task (Sussillo and Barak 2013; Schaeffer et al. 2020), there have been an alternative approach to develop algorithm to construct networks that produce dynamicswith a predefinedmanifold (Beiran et al. 2021;Darshan andRivkind 2022; Pollock and Jazayeri 2020; Claudi and Branco 2022).”
BackgroundFlexible selection of task-relevant features through population gatingPreprint
Barbosa, J., Proville, R. D., Rodgers, C. C., DeWeese, M. R., Ostojic, S., & Boubenec, Y. (2022) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2022.07.21.500962 · Cited by 8
Self-citationLow-dimensional neural manifolds for the control of constrained and unconstrained movementsPreprint
Altan, E., Ma, X., Miller, L. E., Perreault, E. J., & Solla, S. A. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.05.25.542264 · Cited by 10
Remapping in a recurrent neural network model of navigation and context inferencePreprint
Low, I. I. C., Giocomo, L. M., & Williams, A. H. (2023) · eLife · doi:10.7554/elife.86943.2 · Cited by 8
Aligned and oblique dynamics in recurrent neural networksPreprint
Schuessler, F., Mastrogiuseppe, F., Ostojic, S., & Barak, O. (2024) · eLife · doi:10.7554/elife.93060 · Cited by 17
“The assumption of Gaussian connectivity greatly simplifies the math, while its restrictions are irrelevant to the task considered here [4, 12, 61].”
Self-citationConnectivity Structure and Dynamics of Nonlinear Recurrent Neural Networks
Clark, D. G., Marschall, O., van Meegen, A., & Litwin-Kumar, A. (2025) · Physical Review X · doi:10.1103/2jt7-c8cq · Cited by 18
“This formulation introduces an extensive set of latent variables z a ð t Þ , in contrast to the finite set of latent variables, conventionally denoted by κ a ð t Þ , in low-rank neural networks [42,44].”
Linking task structure and neural network dynamics
Márton, C. D., Zhou, S., & Rajan, K. (2022) · Nature Neuroscience · doi:10.1038/s41593-022-01090-w · Cited by 7
Impact of Local Connectivity Patterns on Excitatory-Inhibitory Network Dynamics
Shao, Y., Dahmen, D., Recanatesi, S., Shea‐Brown, E., & Ostojic, S. (2025) · PRX Life · doi:10.1103/prxlife.3.023008 · Cited by 19
Self-citationDifferential Geometry Methods for Constructing Manifold-Targeted Recurrent Neural Networks
Claudi, F., & Branco, T. (2022) · Neural Computation · doi:10.1162/neco_a_01511 · Cited by 7
Bayesian inference is facilitated by modular neural networks with different time scales
Ichikawa, K., & Kaneko, K. (2024) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1011897 · Cited by 6
Structured flexibility in recurrent neural networks via neuromodulation
Costacurta, J. C., Bhandarkar, S., Zoltowski, D. M., & Linderman, S. W. (2024) · neural information processing systems · doi:10.52202/079017-0062 · Cited by 15
Emergent Rate-Based Dynamics in Duplicate-Free Populations of Spiking Neurons
Schmutz, V., Brea, J., & Gerstner, W. (2025) · Physical Review Letters · doi:10.1103/physrevlett.134.018401 · Cited by 9
“If the number of patterns p is kept constant as N → ∞ , the limit RNN is a low-rank mean-field model (without disorder) [7,45].”
Zhang, Y., Feng, J., & Min, B. (2024) · eLife · doi:10.7554/elife.103636.3 · Cited by 8
Stochastic activity in low-rank recurrent neural networks
Mastrogiuseppe, F., Carmona, J., & Machens, C. K. (2025) · PLoS Computational Biology · doi:10.1371/journal.pcbi.1013371 · Cited by 7
Self-citationRemapping in a recurrent neural network model of navigation and context inference
Low, I. I. C., Giocomo, L. M., & Williams, A. H. (2023) · doi:10.1101/2023.01.25.525596 · Cited by 3
“When RNN architecture is explicitly designed to include dedicated neural subpopulations, these subpopulations can improve model performance on particular types of tasks (Beiran et al., 2021; Dubreuil et al., 2022).”
Ning, C., Fu, G., Zhang, Y.-Y., Meyniel, F., & Wang, L. (2026) · Neuron · doi:10.1016/j.neuron.2025.11.014 · Cited by 5
Aligned and oblique dynamics in recurrent neural networks
Schuessler, F., Mastrogiuseppe, F., Ostojic, S., & Barak, O. (2024) · eLife · doi:10.7554/elife.93060.3 · Cited by 22
Self-citationNetwork dynamics underlying OFF responses in the auditory cortexPreprint
Bondanelli, G., Deneux, T., Bathellier, B., & Ostojic, S. (2019) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/810655 · Cited by 4
Self-citationDe novo motor learning creates structure in neural activity space that shapes adaptationPreprint
Chang, J. C., Perich, M. G., Miller, L. E., Gallego, J. Á., & Claudia Clopath, A. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.05.23.541925 · Cited by 3
“How these different regions interact to affect skill learning and adaptation is an ongoing area of study, and future work could use modular and area-specific networks that are constrained by neural data to tease apart their contributions 46–48 .”
Aligned and oblique dynamics in recurrent neural networksPreprint
Schuessler, F., Mastrogiuseppe, F., Ostojic, S., & Barak, O. (2024) · eLife · doi:10.7554/elife.93060.2 · Cited by 7
Self-citationA theory of multi-task computation and task selectionPreprint
Marschall, O., Clark, D. G., & Litwin-Kumar, A. (2025) · eLife · doi:10.64898/2025.12.12.693832 · Cited by 5
Flexible Use of Limited Resources for Sequence Working Memory in Macaque Prefrontal Cortex
Li, S., Chen, J., Zhang, C., Tang, S., Xie, Y., & Wang, L. (2025) · Nature Communications · doi:10.1038/s41467-025-65380-0 · Cited by 4
“For example, the multiple rank subspaces can interact with each other through a population structure-enabled gain modulation mechanism 46 during the gating-in/out processes during the sample and retrieval periods.”
Emergence of Universal Computations Through Neural Manifold Dynamics
Gort, J. (2024) · Neural Computation · doi:10.1162/neco_a_01631 · Cited by 3
“For instance, regarding the relation between low-rank connectivity, neural manifold dynamics and universal approximation properties, akin results were recently announced in [28], [29] as well as in [16], annex 1.”
BackgroundMethodResultInfluentialHigh-dimensional neuronal activity from low-dimensional latent dynamics: a solvable model
Schmutz, V., Haydaroğlu, A., Wang, S., Feng, Y., Carandini, M., & Harris, K. D. (2025) · neural information processing systems · doi:10.52202/085713-3949 · Cited by 13
Bifurcations of a Neural Network Model with Symmetry
Parker, R., & Barreiro, A. K. (2022) · SIAM Journal on Applied Dynamical Systems · doi:10.1137/22m1470451 · Cited by 3
“Several authors have examined how low-rank, asymmetric perturbations create an effectively feed-forward structure that allows coherent dynamics to co-exist with random fluctuations in an orthogonal subspace [14, 13, 33, 32]; the dimensionality of the dynamical subspace can be related to the dimension of the low-rank p…”
BackgroundClaudi, F., Chandra, S., & Fiete, I. (2025) · eLife · doi:10.7554/elife.107224.1 · Cited by 1
Multitasking Recurrent Networks Utilize Compositional Strategies for Control of MovementPreprint
Lazzari, J., & Saxena, S. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.09.10.675375 · Cited by 7
“This work has inspired the study of compositional representations in recurrent neural networks (RNNs) through multitasking and continual learning [8–12].”
Remapping in a recurrent neural network model of navigation and context inference
Low, I. I., Giocomo, L. M., & Williams, A. H. (2023) · eLife · doi:10.7554/elife.86943.3 · Cited by 1
“50,51 Thus, there is an emerging conclusion that RNNs use simple dynamical motifs as building blocks for more general and complex computations, which our results support.”
BackgroundAttractor-Based Models for Sequences and Pattern Generation in Neural Circuits
Alvarez, J. L., Morrison, K., & Curto, C. (2026) · Neural Computation · doi:10.1162/neco.a.1492 · Cited by 11
Shaping Slow Manifolds in Equivariant Recurrent Neural Networks
Bernardo, A. D., Valente, A., Mastrogiuseppe, F., & Ostojic, S. (2026) · Neural Computation · doi:10.1162/neco.a.1577 · Cited by 4
Self-citationSymmetries and continuous attractors in disordered neural circuits
Clark, D. G., Abbott, L. F., & Sompolinsky, H. (2026) · Nature Neuroscience · doi:10.1038/s41593-026-02452-4 · Cited by 17
Sequence Generator Network for Neuromechanical Control of Rat Hindlimbs
Jackson, C. B., Nourse, W. R. P., & Quinn, R. D. (2024) · Lecture notes in computer science · doi:10.1007/978-3-031-72597-5_25 · Cited by 1
Parametric Control of Flexible Timing Through Low-Dimensional Neural ManifoldsPreprint
Beirán, M., Meirhaeghe, N., Sohn, H., Jazayeri, M., & Ostojic, S. (2021) · SSRN Electronic Journal · doi:10.2139/ssrn.3967676 · Cited by 1
Self-citationRemapping in a recurrent neural network model of navigation and context inferencePreprint
Low, I. I. C., Giocomo, L. M., & Williams, A. H. (2023) · eLife · doi:10.7554/elife.86943.1 · Cited by 3
Interpretable compositional computation with recurrent neural networksPreprint
Pezon, L., & van Meegen, A. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.06.23.733979 · Cited by 1
“In the context of low-rank networks, this distribution is typically assumed to be Gaussian [27] or a mixture of Gaussians [31, 37, 45]; recent studies went beyond Gaussianity to link connectivity and computation in the case of more general connectivity structures [43, 75].”
Qiu, S. (2021) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2021.01.15.426814
“Work on low rank structure [29-32] exists but does not emphasize how it is related to the thalamus.”
BackgroundEmergence of universal computations through neural manifold dynamicsPreprint
Vicente, J. G. (2023) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2023.02.21.529079
Aligned and oblique dynamics in recurrent neural networksPreprint
Schuessler, F., Mastrogiuseppe, F., Ostojic, S., & Barak, O. (2024) · eLife · doi:10.7554/elife.93060.1 · Cited by 2
Self-citationZhang, Y., Feng, J., & Min, B. (2024) · eLife · doi:10.7554/elife.103636.1
Ghost Mechanism: An Analytical Model of Abrupt Learning in Recurrent Networks
Dinç, F., Cirakman, E., Kurtkaya, B., Yüksekgönül, M., Jiang, Y., Schnitzer, M. J., & Tanaka, H. (2026) · Physical Review X · doi:10.1103/mjcl-lb4x · Cited by 8
“Rank-one RNNs are theoretically capable of solving the DA task by virtue of their universal approximation property [47, 48], i.e. , they can approximate the canonical model in Eq.”
InfluentialClaudi, F., Chandra, S., & Fiete, I. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.05.07.652608 · Cited by 3
“The network engineering approach [80, 81, 82, 83, 84, 82, 85, 86, 87] constructs circuits starting from the detailed desired dynamics of a system (precise states, fixed points, or specific tuning curves), then directly searching or solving for some network connectivity with those dynamics.”
Zhang, Y., Feng, J., & Min, B. (2025) · eLife · doi:10.7554/elife.103636.2
Claudi, F., Chandra, S., & Fiete, I. (2025) · eLife · doi:10.7554/elife.107224
Sensorimotor remapping drives task specialization in prefrontal cortex
Tissot, H., Boucher, J., Reinert, S., Goltstein, P. M., & Boubenec, Y. (2026) · Nature Communications · doi:10.1038/s41467-026-76104-3 · Cited by 2
“While such an extension holds great promise as a tool to identify functionally distinct neural ensembles, it currently lacks the theoretical foundations that underpins connectivity-based population decomposition [44].”
Improved interpretability in LFADS models using a learned, context-dependent per-trial biasPreprint
Shah, N. P., Meschede-Krasa, B., Kunz, E. M., Hahn, N. V., Kamdar, F. B., Avansino, D. T., Hochberg, L. R., Henderson, J. M., & Sussillo, D. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.10.03.680303 · Cited by 1
Multiple Routes to Metacognitive Judgments of Working Memory in the Macaque Prefrontal CortexPreprint
Ning, C., Fu, G., Zhang, Y., Meyniel, F., & Wang, L. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.07.15.664842
Control of Synchrony by Inhibitory Motifs in Spiking Neural NetworksPreprint
Biswas, A., & Kumar, A. (2025) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.1101/2025.11.07.687261 · Cited by 2
“For instance, local connectivity [5, 18, 20, 26, 31], global matrix features e.g. low-rankness [3, 4, 14] and degree distributions [15, 23] can predict the activity state of neuronal networks.”
Temporal computations in recurrent neural networks
Beirán, M. (2020) · theses.fr (ABES) · doi:10.70675/808e8f9fz7127z4bacza4b9z1ccea046cec1
Trognon, A. (2022) · theses.fr (ABES) · doi:10.70675/a61baf2dz0493z4668z80b0za8538881eb16 · Cited by 4
Low-rank network models of neural computations
Valente, A. (2022) · theses.fr (ABES) · doi:10.70675/51cf5234z10f7z42b2z8d46z13894170c780
Self-citationZhang, Y., Li, X., Shen, X., Li, F., Okazawa, G., Wang, L., Feng, J., & Min, B. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.04.16.718933 · Cited by 1
Predictive pursuit emerges in high-dimensional recurrent neural networksPreprint
Redman, W. T., Dinç, F., Lin, X., Chan, M. G., & Alexander, A. S. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.04.23.720457 · Cited by 2
“Earlier studies have primarily focused on visual, motor, or memory tasks that are often solvable via low-dimensional networks with limited computational resources [13, 30–34 , 36].”
From chaos to synchrony in recurrent excitatory–inhibitory networks with target-specific inhibition
Martorell, C., Calvo, R. M., Annibale, A., & Muñoz, M. A. (2026) · Chaos Solitons & Fractals · doi:10.1016/j.chaos.2026.118979
From Chaos to Synchrony in Recurrent Excitatory–Inhibitory Networks with Target-Specific InhibitionPreprint
Muñoz, M. A., Martorell, C., Annibale, A., & Calvo, R. M. (2026) · SSRN Electronic Journal · doi:10.2139/ssrn.6829428
Rank Dependency of Rescaled Pruning in Recurrent Neural NetworksPreprint
Wang, A. Q., Kim, S. H., & Choi, H. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.05.30.728930
“However, rank one is a restrictive case and cannot support the richer dynamics required for more complex cognitive tasks [15].”
de Vicente, A., Mitelut, C., Mendes, R. A. V., Marianelli, L., Colomer-Rosell, M., Brückner, D. B., Bardella, G., & Donato, F. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.06.04.730137
Low-Rank Recurrent Neural Networks and Their Role in Modeling Cognition
Beirán, M., & Mastrogiuseppe, F. (2026) · Oxford University Press eBooks · doi:10.1093/9780197851494.003.0561
Self-citationChau, H. Y., Oldenburg, I. A., Miller, K. D., & Palmigiano, A. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.08.27.746829
Braun, L. M., Nordal, M. K., Rambø, S.-N. H., Clopath, C., & Cogno, S. G. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.09.02.748960
Moving beyond linear summation to infer interaction order from neural and biological dynamics
Altshuler, G., & Barak, O. (2026) · Cell Reports Methods · doi:10.1016/j.crmeth.2026.101603 · Cited by 1
Condruz, N., Bulygin, I., Chintaluri, C., & Vogels, T. P. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.09.14.751508
“…such as targeted gain modulation, which can silence functional sub-populations so that not every neuron participates in every movement, 17,20,81,82 context-dependent dynamics and plasticity turn our networks into reservoirs with vast capacity, in the spirit of universal function…”
Neural circuit mechanisms for multi-taskingPreprint
Beirán, M., Marschall, O., Zemlianova, K. O., Bernardo, A. D., Ostojic, S., & Litwin-Kumar, A. (2026) · bioRxiv (Cold Spring Harbor Laboratory) · doi:10.64898/2026.10.01.755960
Self-citationXiaojun, W., Lin, Z., Deng, B., Liu, X., & Yu, H. (2026) · Neuroscience · doi:10.1016/j.neuroscience.2026.10.016
Ring attractor networks go full circle
Mastrogiuseppe, F., & Ostojic, S. (2026) · Nature Neuroscience · doi:10.1038/s41593-026-02394-x
Self-citation
Metadata
- Authors
- Beiran, M., Dubreuil, A., Valente, A., Mastrogiuseppe, F., & Ostojic, S.
- Affiliations
- Laboratoire de Neurosciences Cognitives et Computationnelles, INSERM U960, Ecole Normale Superieure. PSL University, 75005 Paris, France manuel.beiran@ens.fr
- Inserm (FR)
- Gatsby Computational Neuroscience Unit (GB)
- Venue
- Neural Computation
- Publisher
- MIT Press - Journals
- Published
- 2021-05-13
- Volume
- 33
- Issue
- 6
- Pages
- 1572-1615
- ISSN
- 0899-7667 (ISSN-L)
- 1530-888X
- OpenAlex
- W3039250185
- Cited by
- 92
- References
- 0
- Citations by year
- 2026: 22 · 2025: 21 · 2024: 15 · 2023: 16 · 2022: 10 · 2021: 6
- FWCI
- 6.17
- Type
- journal article
- Language
- en
- Open access
- green
- arxiv.org
- Locations
- Neural Computation (publishedVersion)
- PubMed (publishedVersion)
- arXiv (Cornell University) (submittedVersion)
- SDL (submittedVersion, other-oa)
- HAL (submittedVersion, other-oa)
- arXiv (Cornell University)
- Funders
- Agence Nationale de la Recherche (ANR-17-EURE-0017, ANR-16-CE37-0016)
- Topics
- Life Sciences › Neuroscience › Cognitive Neuroscience › Neural dynamics and brain function
- Physical Sciences › Computer Science › Artificial Intelligence › Neural Networks and Applications
- Physical Sciences › Physics and Astronomy › Statistical and Nonlinear Physics › Nonlinear Dynamics and Pattern Formation
- Keywords
- Gaussian mixture model, population structure, latent variables, dynamical systems, network connectivity
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.