pith. sign in

arxiv: 2606.07336 · v1 · pith:3Q6SXFACnew · submitted 2026-06-05 · 🧬 q-bio.NC

Fixed point compositionality via low-rank gluing rules in inhibition-dominated threshold-linear networks

Juliana Londono Alvarez This is my paper

Pith reviewed 2026-06-27 20:05 UTC · model grok-4.3

classification 🧬 q-bio.NC
keywords threshold-linear networksfixed pointscompositionalitymodularitylow-rank gluingsinhibition-dominated networksattractorsgeneralized CTLNs
0
0 comments X

The pith

Low-rank gluings constrain global fixed points in inhibition-dominated threshold-linear networks to combinations of local module fixed points.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proves that when subnetworks are assembled into larger networks using low-rank inter-module couplings, the fixed points of the whole system must arise as combinations of the fixed points that each module would have on its own. This holds for arbitrary internal connectivity inside the modules, provided the overall network remains inhibition-dominated. For the special case of rank-1 gluings the authors give an exact rule that identifies precisely which combinations survive as global fixed points. The same rules extend earlier decomposition results from combinatorial threshold-linear networks to a broader family called generalized CTLNs. The construction supplies an explicit method for building networks whose attractors can be read off from the component pieces rather than solved globally.

Core claim

Global fixed points of low-rank gluing networks are constrained to be combinations of the local fixed points of their constituent modules; for rank-1 gluings a complete characterization determines exactly which combinations yield global fixed points.

What carries the argument

Low-rank gluings: a modular assembly rule in which arbitrary subnetworks are joined by low-rank coupling matrices that enforce fixed-point compositionality.

If this is right

  • Fixed-point sets of the composite network can be enumerated directly from the fixed-point sets of the modules without solving the full system.
  • Decomposition rules previously derived for combinatorial threshold-linear networks remain valid for the larger class of generalized CTLNs.
  • Networks with combinatorially many predictable attractors can be assembled by reusing a small collection of component motifs.
  • The same gluing construction yields both compositional fixed points and compositional limit cycles.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The result supplies a concrete design principle for scaling modular networks while preserving analytic control over their attractors.
  • If cortical circuits implement analogous low-rank inter-area couplings, their global activity patterns may be predictable from local circuit motifs.
  • The rank condition could be tested by measuring effective connectivity matrices between recorded sub-populations and checking whether their singular values decay rapidly.

Load-bearing premise

The couplings between modules must be exactly low-rank (or rank-1) as constructed and the network must remain inhibition-dominated; any deviation removes the compositionality guarantee.

What would settle it

Construct a low-rank gluing network whose inter-module matrices satisfy the rank condition yet contains a global fixed point whose support pattern is not a valid combination of any local module fixed points.

Figures

Figures reproduced from arXiv: 2606.07336 by Juliana Londono Alvarez.

Figure 1
Figure 1. Figure 1: Fixed point compositionality and attractors. [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Low-rank gluings versus rank-1 gluings. Elements with the same shading are constrained to be equal. (A) A connectivity matrix defining a low-rank gluing of component subnetworks W1, . . . , WN . Here, the scalar γ (i) j gives the strength of the synaptic connection from node j to all nodes in τi, corresponding to the subnetwork Wi. (B) A connectivity matrix defining a rank-1 gluing of component subnetworks… view at source ↗
Figure 3
Figure 3. Figure 3: Illustration of Theorems 1 and 2. (A) Three component networks, and their fixed point supports FP(Wi, θ). (B) Structure of a modular network assembled with the components of panel A, and the full set of possible supports when networks are low-rank gluings of those. (C-D) Two example low-rank and rank-1 gluings of the components in panel A, and their fixed point supports FP(W, θ). In panel C, Theorem 1 cons… view at source ↗
Figure 4
Figure 4. Figure 4: Illustration of CTLN theorems for three special low-rank gluings. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Attractor correspondence in CTLNs and gCTLNs. [PITH_FULL_IMAGE:figures/full_fig_p009_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Fixed points and dynamics of CTLN vs. gCTLN for three specific graph constructions. [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Theorem 5 (sign conditions, [44]). Same color indicates the same sgn s σ i for i in the given component. (A) A subset σ is permitted if and only if sgn s σ i = sgn s σ j = idx(σ) for all i, j ∈ σ. sgn s σ k for k ∈ [n] \ σ can be anything. (B) A subset σ is forbidden if and only if there is a mix of signs in it. More precisely, if it exists i, j ∈ σ such that sgn s σ i = − sgn s σ j . (C) A subset σ suppor… view at source ↗
Figure 8
Figure 8. Figure 8: Cartoon showing sgn s σ i = Q i∈I sgn s σi j for three components. Note that sgn s σi j is constant for any j ∈ [n] \ τi, for all i ∈ [N]. On the left hand side, we have the index equalities because σ ∈ FP(W, θ). On the right hand side, we have the index equalities because σi ∈ FP(Wτi , θ). 19 [PITH_FULL_IMAGE:figures/full_fig_p019_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Cartoon showing the assumption that σ1 = ∅ (m = 1). This implies sgn s σ1 i = 1. On the right hand side, we have the index equalities because, as in [PITH_FULL_IMAGE:figures/full_fig_p020_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Checking sign conditions for σ def = S i∈I σi, assuming j ∈ τ2. On the right hand side, we have the index equalities because σi ∈ FP(Wτi , θ) ∪ {∅} for each i ∈ [N] and sgn s σℓ k = − idx σℓ for all ℓ ∈ I. 20 [PITH_FULL_IMAGE:figures/full_fig_p020_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Low-rank gluings versus rank-1 gluings in graph-based networks. [PITH_FULL_IMAGE:figures/full_fig_p021_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: A gCTLN illustration of Theorem 1. Four graphs defined in the same components, with different low-rank gluings. (A) Unions of local fixed points form the set of possible global fixed points. (B - E) Examples of low-rank gluings, along with the resulting global fixed point supports. Adapted from [47]. 3.3.1 Cyclic unions and proof of Theorem 3 Recall the cyclic union construction from Figure 4B. Formally, … view at source ↗
Figure 13
Figure 13. Figure 13: Bound gait as a CTLN and a gCTLN. (A) The bound gait. Drawing adapted from [62]. (B) A cyclic union designed to mimic the limb activations of the bound gait. τ1 (front limbs) and τ3 (back limbs) contain the readout nodes. (C) Fixed point supports for the network in B, identical for both the CTLN (panel D) and gCTLN (panel E). (D-E) Connectivity matrix for the CTLN and gCTLN, respectively, and the attracto… view at source ↗
Figure 14
Figure 14. Figure 14: Composite linear chain. A graph that is a composite linear chain and the connectivity matrix defined by it as a gCTLN. The fixed point supports reduce to those of the last component τ4. When the network is initialized in the first component, the activity trickles down the chain to finally settle in the last component. Adapted from [47]. Proof. By definition, {k} ∈ FP(G, ε, δ) if and only if there exists a… view at source ↗
Figure 15
Figure 15. Figure 15: Comparison of a graph that has only some targets in the other components and a clique union, [PITH_FULL_IMAGE:figures/full_fig_p028_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: gCTLNs graph rules. (A) Cycles and cliques yield unique full-support fixed points. (B) The fixed point supports for independent sets and directed acyclic graphs are sinks and unions of sinks. Adapted from [49]. With these definitions, we obtain the following result: Corollary 2. Let G be a graph on n nodes, and σ ⊆ [n]. (a) if G is a cycle, then FP(G, ε, δ) = {[n]}. (b) If G is a clique, then FP(G, ε, δ) … view at source ↗
Figure 17
Figure 17. Figure 17: A CTLN and a gCTLN defined on a graph satisfying the assumptions of Theorem 9. [PITH_FULL_IMAGE:figures/full_fig_p030_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: Clione’s swimming control. (A) Top: vectors for the swimming directions up/down, right/left, front/back. Bottom: all possible combinations of the swimming directions in top panel. (B) Left: a cyclic union of three independent sets. Center: an equivalent representation of the graph in left panel, with all edges explicitly drawn. Right: a gCTLN-defined connectivity matrix. (C) Principal component projection… view at source ↗
read the original abstract

Brains routinely generate highly flexible and complex behaviors on a relatively stable structure and limited resources. A key mechanism underlying this ability is compositionality, which allows the brain to efficiently decompose complex tasks into simpler, reusable primitives. While network modularity has often been linked to compositionality in biological and artificial networks, a rigorous mathematical characterization of this relationship in nonlinear networks is still lacking. In this work, we formally investigate how structural modularity supports functional compositionality in inhibition-dominated threshold-linear networks (TLNs). We introduce a novel class of modular network assembly called low-rank gluings, where component subnetworks with arbitrary internal connectivity are connected via specific low-rank couplings. We prove that the global fixed points of these networks are constrained to be combinations of the local fixed points of their constituent modules. For a more structured subclass, called rank-1 gluings, we provide a complete characterization that determines which combinations of local fixed points yield global ones. We apply these results to graph-based networks, extending fixed point decomposition rules from combinatorial threshold-linear networks (CTLNs) to the more flexible family of generalized CTLNs (gCTLNs), thereby proving that these structural rules are more robust than initially posited. Finally, we demonstrate that these gluing rules provide a mathematically tractable recipe for engineering compositional dynamics, enabling the construction of networks with a combinatorially large repertoire of predictable attractors that can be understood from simpler component motifs, ranging from compositions of fixed points to compositional limit cycles.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

0 major / 3 minor

Summary. The manuscript introduces low-rank gluings for assembling inhibition-dominated threshold-linear networks (TLNs) from modular subnetworks with arbitrary internal connectivity. It proves that global fixed points are constrained to combinations of the local fixed points of the modules. For the subclass of rank-1 gluings, a complete characterization is provided determining which combinations yield global fixed points. The results are applied to graph-based generalized combinatorial TLNs (gCTLNs), extending fixed-point decomposition rules and proving greater robustness than initially posited. The gluing rules are also presented as a recipe for engineering networks with combinatorially large repertoires of predictable attractors, including compositional limit cycles.

Significance. If the proofs hold, the work supplies a rigorous, conditional mathematical characterization linking structural modularity to functional compositionality in nonlinear dynamical systems. This is significant for theoretical neuroscience. Credit is due for the explicit low-rank gluing construction, the machine-checkable-style proofs for the stated class of networks, the complete rank-1 characterization, and the extension to gCTLNs that demonstrates robustness of the structural rules without data-fitting or circularity.

minor comments (3)
  1. [Abstract] Abstract: the phrase 'low-rank gluings' and 'rank-1 gluings' are introduced without a one-sentence definition or forward reference to the precise matrix condition (e.g., the form of the inter-module coupling matrix); adding this would improve immediate readability.
  2. [§2] The manuscript would benefit from an early, self-contained statement (perhaps in §2) of the exact low-rank condition on the coupling matrices before the main theorems are stated.
  3. [Engineering examples] Figure captions and axis labels in the engineering-examples section could be expanded to indicate which local fixed points are being combined in each panel.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive summary, significance assessment, and recommendation of minor revision. No specific major comments were raised in the report.

Circularity Check

0 steps flagged

No significant circularity identified

full rationale

The paper defines low-rank gluings as a new construction on inhibition-dominated TLNs and proves (via direct mathematical argument) that global fixed points are combinations of local module fixed points precisely when the inter-module couplings satisfy the stated low-rank form. This is a conditional theorem on explicitly constructed objects; the low-rank condition is the hypothesis, not a derived or fitted quantity. No self-citations are invoked as load-bearing uniqueness theorems, no parameters are fitted to data and then relabeled as predictions, and no known empirical patterns are merely renamed. The derivation chain is therefore self-contained within the paper's own definitions and proofs.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 1 invented entities

Review performed on abstract only; therefore the ledger is necessarily incomplete. The work rests on standard mathematical properties of threshold-linear networks and on the newly introduced low-rank coupling definition.

axioms (1)
  • domain assumption Inhibition-dominated threshold-linear dynamics
    Abstract states the networks are inhibition-dominated TLNs; this background assumption is required for the fixed-point claims to hold.
invented entities (1)
  • low-rank gluing no independent evidence
    purpose: Modular assembly rule that enforces fixed-point compositionality
    Newly defined construction; no independent evidence supplied in abstract.

pith-pipeline@v0.9.1-grok · 5791 in / 1397 out tokens · 22427 ms · 2026-06-27T20:05:02.965008+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

76 extracted references · 3 canonical work pages

  1. [1]

    Strogatz

    Steven H. Strogatz. Exploring complex networks.Nature, 410(6825):268–276, March 2001

    2001

  2. [2]

    Complex networks: Structure and dynamics.Physics Reports, 424(4):175–308, February 2006

    Stefano Boccaletti, Vito Latora, Yamir Moreno, Mario Chavez, and Dong-uk Hwang. Complex networks: Structure and dynamics.Physics Reports, 424(4):175–308, February 2006

    2006

  3. [3]

    Structure and function of complex brain networks.Dialogues in Clinical Neuroscience, 15(3):247–262, September 2013

    Olaf Sporns. Structure and function of complex brain networks.Dialogues in Clinical Neuroscience, 15(3):247–262, September 2013

    2013

  4. [4]

    Structural and Functional Brain Networks: From Connections to Cognition.Science, 342(6158):1238411, November 2013

    Hae-Jeong Park and Karl Friston. Structural and Functional Brain Networks: From Connections to Cognition.Science, 342(6158):1238411, November 2013

    2013

  5. [5]

    Bassett and Olaf Sporns

    Danielle S. Bassett and Olaf Sporns. Network neuroscience.Nature Neuroscience, 20(3):353–364, March 2017

    2017

  6. [6]

    Olaf Sporns and Richard F. Betzel. Modular Brain Networks.Annual Review of Psychology, 67:613– 640, January 2016

    2016

  7. [7]

    David Papo and Javier M. Buldú. Biological detail and graph structure in network neuroscience. Frontiers in Network Physiology, 5, October 2025

    2025

  8. [8]

    The balance between integration and segregation drives network dynamics maximizing multistability and metastability.Scientific Reports, 15(1):18811, May 2025

    Javier Palma-Espinosa, Sebastián Orellana-Villota, Carlos Coronel-Oliveros, Jean Paul Maidana, and Patricio Orio. The balance between integration and segregation drives network dynamics maximizing multistability and metastability.Scientific Reports, 15(1):18811, May 2025

    2025

  9. [9]

    Jacob J. W. Bakermans, Joseph Warren, James C. R. Whittington, and Timothy E. J. Behrens. Constructing future behavior in the hippocampal formation through composition and replay.Nature Neuroscience, 28(5):1061–1072, May 2025

    2025

  10. [10]

    Compositionality of Rule Representations in Human Prefrontal Cortex.Cerebral Cortex, 22(6):1237–1246, June 2012

    Carlo Reverberi, Kai Görgen, and John-Dylan Haynes. Compositionality of Rule Representations in Human Prefrontal Cortex.Cerebral Cortex, 22(6):1237–1246, June 2012

    2012

  11. [11]

    Joglekar, H

    Guangyu Robert Yang, Madhura R. Joglekar, H. Francis Song, William T. Newsome, and Xiao-Jing Wang. Task representations in neural networks trained to perform many cognitive tasks.Nature Neuroscience, 22(2):297–306, February 2019

    2019

  12. [12]

    Driscoll, Krishna Shenoy, and David Sussillo

    Laura N. Driscoll, Krishna Shenoy, and David Sussillo. Flexible multitask computation in recurrent networks utilizes shared dynamical motifs.Nature Neuroscience, 27(7):1349–1363, July 2024

    2024

  13. [13]

    Neural Module Networks

    Jacob Andreas, Marcus Rohrbach, Trevor Darrell, and Dan Klein. Neural Module Networks. In Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pages 39–48, 2016

    2016

  14. [14]

    Break It Down: Evidence for Structural Composi- tionality in Neural Networks.Advances in Neural Information Processing Systems, 36:42623–42660, December 2023

    Michael Lepori, Thomas Serre, and Ellie Pavlick. Break It Down: Evidence for Structural Composi- tionality in Neural Networks.Advances in Neural Information Processing Systems, 36:42623–42660, December 2023. 35

    2023

  15. [15]

    Willett, Darrel R

    Francis R. Willett, Darrel R. Deo, Donald T. Avansino, Paymon Rezaii, Leigh R. Hochberg, Jaimie M. Henderson, and Krishna V. Shenoy. Hand Knob Area of Premotor Cortex Represents the Whole Body in a Compositional Way.Cell, 181(2):396–409.e26, April 2020

    2020

  16. [16]

    Bouchacourt, Adel Ardalan, Nikola T

    Sina Tafazoli, Flora M. Bouchacourt, Adel Ardalan, Nikola T. Markov, Motoaki Uchimura, Marcelo G. Mattar, Nathaniel D. Daw, and Timothy J. Buschman. Building compositional tasks with shared neural subspaces.Nature, 650(8100):164–172, February 2026

    2026

  17. [17]

    Kymn, Sonia Mazelet, Anthony Thomas, Denis Kleyko, E

    Christopher J. Kymn, Sonia Mazelet, Anthony Thomas, Denis Kleyko, E. Paxon Frady, Friedrich T. Sommer, and Bruno A. Olshausen. Binding in hippocampal-entorhinal circuits enables composition- ality in cognitive maps. InProceedings of the 38th International Conference on Neural Information Processing Systems, volume 37 ofNIPS ’24, pages 39128–39157, Red Hoo...

  18. [18]

    Curran Associates Inc

  19. [19]

    Amematsro, Eric M

    Elom A. Amematsro, Eric M. Trautmann, Najja J. Marshall, L. F. Abbott, Michael N. Shadlen, Daniel M. Wolpert, and Mark M. Churchland. Motor cortex flexibly deploys a high-dimensional repertoire of subskills. bioRxiv: 2025.09.07.674717, September 2025

    2025

  20. [20]

    Human hippocampal ripples coordinate planning sequences and compositional representations in neocortex.Nature Neuroscience, pages 1–11, May 2026

    Li He, Xiongfei Wang, Jinbo Zhang, Zhibing Xiao, Xiangyu Hu, Philipp Schwartenbeck, Jacob Bak- ermans, Tim Behrens, and Yunzhe Liu. Human hippocampal ripples coordinate planning sequences and compositional representations in neocortex.Nature Neuroscience, pages 1–11, May 2026

    2026

  21. [21]

    Schaub.Modularity and Dynamics on Complex Networks

    Renaud Lambiotte and Michael T. Schaub.Modularity and Dynamics on Complex Networks. Ele- ments in the Structure and Dynamics of Complex Networks. Cambridge University Press, Cambridge, 2022

    2022

  22. [22]

    Schaub, Jean-Charles Delvenne, Martin Rosvall, and Renaud Lambiotte

    Michael T. Schaub, Jean-Charles Delvenne, Martin Rosvall, and Renaud Lambiotte. The many facets of community detection in complex networks.Applied Network Science, 2(1):4, February 2017

    2017

  23. [23]

    O. Sporns. Theoretical Neuroanatomy: Relating Anatomical and Functional Connectivity in Graphs and Cortical Connection Matrices.Cerebral Cortex, 10(2):127–141, February 2000

    2000

  24. [24]

    Shaping Dynamics With Multiple Populations in Low-Rank Recurrent Networks.Neural Computa- tion, 33(6):1572–1615, May 2021

    Manuel Beiran, Alexis Dubreuil, Adrian Valente, Francesca Mastrogiuseppe, and Srdjan Ostojic. Shaping Dynamics With Multiple Populations in Low-Rank Recurrent Networks.Neural Computa- tion, 33(6):1572–1615, May 2021

    2021

  25. [25]

    Clark and Manuel Beiran

    David G. Clark and Manuel Beiran. Structure of activity in multiregion recurrent neural networks. Proceedings of the National Academy of Sciences, 122(10):e2404039122, March 2025

    2025

  26. [26]

    Attractor neural network models of spatial maps in hippocampus.Hippocampus, 9(4):481–489, 1999

    Misha Tsodyks. Attractor neural network models of spatial maps in hippocampus.Hippocampus, 9(4):481–489, 1999

    1999

  27. [27]

    Richard H. R. Hahnloser. Emergence of neural integration in the head-direction system by visual supervision.Neuroscience, 120(3):877–891, September 2003

    2003

  28. [28]

    Hulse, Vivek Jayaraman, Sandro Romani, and Ann M

    Marcella Noorman, Brad K. Hulse, Vivek Jayaraman, Sandro Romani, and Ann M. Hermundstad. Maintaining and updating accurate internal representations of continuous variables with a handful of neurons.Nature Neuroscience, 27(11):2207–2217, November 2024

    2024

  29. [29]

    Cell Assembly Sequences Arising from Spike Threshold Adaptation Keep Track of Time in the Hippocampus.Journal of Neuroscience, 31(8):2828–2834, February 2011

    Vladimir Itskov, Carina Curto, Eva Pastalkova, and György Buzsáki. Cell Assembly Sequences Arising from Spike Threshold Adaptation Keep Track of Time in the Hippocampus.Journal of Neuroscience, 31(8):2828–2834, February 2011

    2011

  30. [30]

    Xiaohui Xie, Richard H. R. Hahnloser, and H. Sebastian Seung. Selectively Grouping Neurons in Recurrent Networks of Lateral Inhibition.Neural Computation, 14(11):2627–2646, November 2002. 36

    2002

  31. [31]

    Richard H. R. Hahnloser, H. Sebastian Seung, and Jean-Jacques Slotine. Permitted and Forbidden Sets in Symmetric Threshold-Linear Networks.Neural Computation, 15(3):621–638, March 2003

    2003

  32. [32]

    Encoding Binary Neural Codes in Networks of Threshold-Linear Neurons.Neural Computation, 25(11):2858–2903, November 2013

    Carina Curto, Anda Degeratu, and Vladimir Itskov. Encoding Binary Neural Codes in Networks of Threshold-Linear Neurons.Neural Computation, 25(11):2858–2903, November 2013

    2013

  33. [33]

    PatternCompletioninSymmetricThreshold-LinearNetworks

    CarinaCurtoandKatherineMorrison. PatternCompletioninSymmetricThreshold-LinearNetworks. Neural Computation, 28(12):2825–2852, December 2016

    2016

  34. [34]

    Diversity of Emergent Dynamics in Competitive Threshold-Linear Networks.SIAM Journal on Applied Dynamical Systems, 23(1):855–884, March 2024

    Katherine Morrison, Anda Degeratu, Vladimir Itskov, and Carina Curto. Diversity of Emergent Dynamics in Competitive Threshold-Linear Networks.SIAM Journal on Applied Dynamical Systems, 23(1):855–884, March 2024

    2024

  35. [35]

    Linking Connectivity, Dynamics, and Computations in Low-Rank Recurrent Neural Networks.Neuron, 99(3):609–623.e29, August 2018

    Francesca Mastrogiuseppe and Srdjan Ostojic. Linking Connectivity, Dynamics, and Computations in Low-Rank Recurrent Neural Networks.Neuron, 99(3):609–623.e29, August 2018

    2018

  36. [36]

    Dynamics of random recurrent networks with correlated low-rank structure.Physical Review Re- search, 2(1):013111, February 2020

    Friedrich Schuessler, Alexis Dubreuil, Francesca Mastrogiuseppe, Srdjan Ostojic, and Omri Barak. Dynamics of random recurrent networks with correlated low-rank structure.Physical Review Re- search, 2(1):013111, February 2020

    2020

  37. [37]

    Hopfield

    John J. Hopfield. Neural networks and physical systems with emergent collective computational abilities.Proceedings of the National Academy of Sciences, 79(8):2554–2558, April 1982

    1982

  38. [38]

    Sebastian Seung

    H. Sebastian Seung. How the brain keeps the eyesstill.Proceedings of the National Academy of Sciences, 93(23):13339–13344, November 1996

    1996

  39. [39]

    McNaughton

    Alexei Samsonovich and Bruce L. McNaughton. Path Integration and Cognitive Mapping in a Con- tinuous Attractor Neural Network Model.Journal of Neuroscience, 17(15):5900–5920, August 1997

    1997

  40. [40]

    Battaglia and Alessandro Treves

    Francesco P. Battaglia and Alessandro Treves. Attractor neural networks storing multiple space representations: Amodelforhippocampalplacefields.Physical Review E,58(6):7738–7753, December 1998

    1998

  41. [41]

    Yoram Burak and Ila R. Fiete. Accurate Path Integration in Continuous Attractor Network Models of Grid Cells.PLOS Computational Biology, 5(2):e1000291, February 2009

    2009

  42. [42]

    Gardner, Erik Hermansen, Marius Pachitariu, Yoram Burak, Nils A

    Richard J. Gardner, Erik Hermansen, Marius Pachitariu, Yoram Burak, Nils A. Baas, Benjamin A. Dunn, May-Britt Moser, and Edvard I. Moser. Toroidal topology of population activity in grid cells. Nature, 602(7895):123–128, February 2022

    2022

  43. [43]

    McNaughton, Francesco P

    Bruce L. McNaughton, Francesco P. Battaglia, Ole Jensen, Edvard I. Moser, and May-Britt Moser. Path integration and the neural basis of the ’cognitive map’.Nature Reviews Neuroscience, 7(8):663– 678, August 2006

    2006

  44. [44]

    Cohen and Stephen Grossberg

    Michael A. Cohen and Stephen Grossberg. Absolute stability of global pattern formation and par- allel memory storage by competitive neural networks.IEEE Transactions on Systems, Man, and Cybernetics, SMC-13(5):815–826, September 1983

    1983

  45. [45]

    Fixed Points of Competitive Threshold-Linear Networks.Neural Computation, 31(1):94–155, January 2019

    Carina Curto, Jesse Geneson, and Katherine Morrison. Fixed Points of Competitive Threshold-Linear Networks.Neural Computation, 31(1):94–155, January 2019

    2019

  46. [46]

    Core motifs predict dynamic attractors in combinatorial threshold-linear networks.PLOS ONE, 17(3):e0264456, March 2022

    Caitlyn Parmelee, Samantha Moore, Katherine Morrison, and Carina Curto. Core motifs predict dynamic attractors in combinatorial threshold-linear networks.PLOS ONE, 17(3):e0264456, March 2022. 37

    2022

  47. [47]

    Rotstein

    Andrea Bel, Romina Cobiaga, Walter Reartes, and Horacio G. Rotstein. Periodic Solutions in Threshold-Linear Networks and Their Entrainment.SIAM Journal on Applied Dynamical Systems, 20(3):1177–1208, January 2021

    2021

  48. [48]

    Sequential At- tractors in Combinatorial Threshold-Linear Networks.SIAM Journal on Applied Dynamical Systems, 21(2):1597–1630, June 2022

    Caitlyn Parmelee, Juliana Londono Alvarez, Carina Curto, and Katherine Morrison. Sequential At- tractors in Combinatorial Threshold-Linear Networks.SIAM Journal on Applied Dynamical Systems, 21(2):1597–1630, June 2022

    2022

  49. [49]

    On graphical domination for threshold-linear networks with recurrent excitation and global inhibition

    Carina Curto. On graphical domination for threshold-linear networks with recurrent excitation and global inhibition. arXiv: 2510.05098, October 2025

    work page arXiv 2025

  50. [50]

    Graph Rules for Recurrent Neural Network Dynamics.Notices of the American Mathematical Society, 70(04):1, April 2023

    Carina Curto and Katherine Morrison. Graph Rules for Recurrent Neural Network Dynamics.Notices of the American Mathematical Society, 70(04):1, April 2023

    2023

  51. [51]

    The binding problem.Current Opinion in Neurobiology, 6(2):171–178, April 1996

    Anne Treisman. The binding problem.Current Opinion in Neurobiology, 6(2):171–178, April 1996

    1996

  52. [52]

    Adina L. Roskies. The Binding Problem.Neuron, 24(1):7–9, September 1999

    1999

  53. [53]

    Long and Michale S

    Michael A. Long and Michale S. Fee. Using temperature to analyse temporal dynamics in the songbird motor pathway.Nature, 456(7219):189–194, November 2008

    2008

  54. [54]

    Tang, Marie L

    Lamont S. Tang, Marie L. Goeritz, Jonathan S. Caplan, Adam L. Taylor, Mehmet Fisek, and Eve Marder. Precise Temperature Compensation of Phase in a Rhythmic Motor Pattern.PLOS Biology, 8(8):e1000469, August 2010

    2010

  55. [55]

    Traub, Nancy J

    Miles Adrian Whittington, Roger D. Traub, Nancy J. Kopell, G. Bard Ermentrout, and Eberhard H. Buhl. Inhibition-based rhythms: Experimental and mathematical observations on network dynamics. International Journal of Psychophysiology, 38(3):315–336, December 2000

    2000

  56. [56]

    Karnani, Masakazu Agetsuma, and Rafael Yuste

    Mahesh M. Karnani, Masakazu Agetsuma, and Rafael Yuste. A blanket of inhibition: Functional inferences from dense inhibitory connectivity.Current Opinion in Neurobiology, 26:96–102, June 2014

    2014

  57. [57]

    Moises Arriaga and Edward B. Han. Dedicated Hippocampal Inhibitory Networks for Locomotion and Immobility.Journal of Neuroscience, 37(38):9222–9238, September 2017

    2017

  58. [58]

    Dense Inhibitory Connectivity in Neocortex.Neuron, 69(6):1188–1203, March 2011

    Elodie Fino and Rafael Yuste. Dense Inhibitory Connectivity in Neocortex.Neuron, 69(6):1188–1203, March 2011

    2011

  59. [59]

    Dirk M. Bucher. Central Pattern Generators. In Larry R. Squire, editor,Encyclopedia of Neuro- science, pages 691–700. Academic Press, Oxford, January 2009

    2009

  60. [60]

    Attractor-based models for sequences and pattern generation in neural circuits

    Juliana Londono Alvarez. Attractor-based models for sequences and pattern generation in neural circuits. arXiv: 2410.11012, October 2024

    work page arXiv 2024

  61. [61]

    Chapter 8 - Predicting Neural Network Dynamics via Graph- ical Analysis

    Katherine Morrison and Carina Curto. Chapter 8 - Predicting Neural Network Dynamics via Graph- ical Analysis. In Raina Robeva and Matthew Macauley, editors,Algebraic and Combinatorial Com- putational Biology, MSE/Mathematics in Science and Engineering, pages 241–277. Academic Press, January 2019

    2019

  62. [62]

    Attractor-Based Models for Se- quences and Pattern Generation in Neural Circuits.Neural Computation, 38(3):257–291, February 2026

    Juliana Londono Alvarez, Katherine Morrison, and Carina Curto. Attractor-Based Models for Se- quences and Pattern Generation in Neural Circuits.Neural Computation, 38(3):257–291, February 2026. 38

    2026

  63. [63]

    van der Weele and Erik J

    Jacobus P. van der Weele and Erik J. Banning. Mode interaction in horses, tea, and other nonlinear oscillators: The universal role of symmetry.American Journal of Physics, 69(9):953–965, September 2001

    2001

  64. [64]

    Rabinovich, Allen I

    Pablo Varona, Mikhail I. Rabinovich, Allen I. Selverston, and Yuri I. Arshavsky. Winnerless competi- tion between sensory neurons generates chaos: A possible mechanism for molluscan hunting behavior. Chaos: An Interdisciplinary Journal of Nonlinear Science, 12(3):672–677, September 2002

    2002

  65. [65]

    Yu. V. Panchin, Yuri I. Arshavsky, Tatiana G. Deliagina, Lyudmila Borisovna Popova, and Grigori N. Orlovsky. Control of locomotion in marine mollusk Clione Limacina. IX. Neuronal mechanisms of spatial orientation.Journal of Neurophysiology, 73(5):1924–1937, May 1995

    1924

  66. [66]

    Gallego, Matthew G

    Juan A. Gallego, Matthew G. Perich, Lee E. Miller, and Sara A. Solla. Neural Manifolds for the Control of Movement.Neuron, 94(5):978–984, June 2017

    2017

  67. [67]

    Neuralcircuitsascomputationaldynamicalsystems.Current Opinion in Neurobiology, 25:156–163, April 2014

    DavidSussillo. Neuralcircuitsascomputationaldynamicalsystems.Current Opinion in Neurobiology, 25:156–163, April 2014

    2014

  68. [68]

    Dynamics on the manifold: Identifying computational dynamical activity from neural population recordings.Current Opinion in Neurobiology, 70:163–170, October 2021

    Lea Duncker and Maneesh Sahani. Dynamics on the manifold: Identifying computational dynamical activity from neural population recordings.Current Opinion in Neurobiology, 70:163–170, October 2021

    2021

  69. [69]

    Perich, Devika Narain, and Juan A

    Matthew G. Perich, Devika Narain, and Juan A. Gallego. A neural manifold view of the brain.Nature Neuroscience, 28(8):1582–1597, August 2025

    2025

  70. [70]

    Golub, David Sussillo, and Krishna V

    Saurabh Vyas, Matthew D. Golub, David Sussillo, and Krishna V. Shenoy. Computation Through Neural Population Dynamics.Annual Review of Neuroscience, 43(Volume 43, 2020):249–275, July 2020

    2020

  71. [71]

    Valentin Schmutz, Ali Haydaroğlu, Shuqi Wang, Yixiao Feng, Matteo Carandini, and Kenneth D. Harris. High-dimensional neuronal activity from low-dimensional latent dynamics: A solvable model. bioRxiv: 2025.06.03.657632, January 2026

    2025

  72. [72]

    Rabinovich, Valentin S

    Mikhail I. Rabinovich, Valentin S. Afraimovich, and Pablo Varona. Heteroclinic binding.Dynamical Systems, 25(3):433–442, September 2010

    2010

  73. [73]

    Rabinovich and Pablo Varona

    Mikhail I. Rabinovich and Pablo Varona. Discrete Sequential Information Coding: Heteroclinic Cognitive Dynamics.Frontiers in Computational Neuroscience, 12, September 2018

    2018

  74. [74]

    Modularityofbiologicalsystems: Alinkbetweenstructureandfunction.Journal of The Royal Society Interface, 20(207):20230505, October 2023

    Claus Kadelka, Matthew Wheeler, Alan Veliz-Cuba, David Murrugarra, and Reinhard Laubenbacher. Modularityofbiologicalsystems: Alinkbetweenstructureandfunction.Journal of The Royal Society Interface, 20(207):20230505, October 2023

    2023

  75. [75]

    ical foundations of discrete dynamical systems

    Daniel Carranza, Chris Kapulkin, Nathan Kershaw, Reinhard Laubenbacher, and Matthew Wheeler. ical foundations of discrete dynamical systems. arXiv: 2506.05190, June 2025

    work page arXiv 2025

  76. [76]

    Diverse Mean-Field Dynamics of Clustered, Inhibition-Stabilized Hawkes Net- works via Combinatorial Threshold-Linear Networks.PRX Life, 3(4), 2025

    Caitlin Lienkaemper. Diverse Mean-Field Dynamics of Clustered, Inhibition-Stabilized Hawkes Net- works via Combinatorial Threshold-Linear Networks.PRX Life, 3(4), 2025. 39

    2025