pith. sign in

arxiv: 2603.12416 · v2 · submitted 2026-03-12 · 🧬 q-bio.NC

Formation of Artificial Neural Assemblies by Biologically Plausible Inhibition Mechanisms

Lucas Hoff , Gustavo Soroka , Matheus Guimar\~aes , Aline Villavicencio , Marco Idiart This is my paper

Pith reviewed 2026-05-15 11:26 UTC · model grok-4.3

classification 🧬 q-bio.NC
keywords neural assembliesHebbian learninggamma oscillationsinhibition mechanismsassembly calculusE%-winners-take-allcortical networksneural dynamics
0
0 comments X

The pith

A gamma-cycle-inspired inhibition rule lets network dynamics set the size of formed neural assemblies and raises their recovery rate.

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

The Assembly Calculus model forms Hebbian assemblies through fixed k-winners-take-all selection and synaptic strengthening. The new E%-WTA model replaces the fixed count with a percentage-based selection drawn from gamma-oscillation statistics and scales inhibition to measured excitatory-inhibitory neuron ratios in cortex. These changes remove the requirement for a preset assembly size, so the number of neurons that end up co-active emerges from the network's own activity patterns. Simulations show that the resulting assemblies are recovered more reliably when the original stimuli are presented again.

Core claim

By implementing an E%-winners-take-all selection process inspired by gamma oscillation cycles together with an inhibition mechanism based on the excitatory to inhibitory neuron ratio, the model permits the network's own dynamics to determine the size of the formed assemblies, yielding higher recovery rates for these assemblies upon evocation of the generating stimuli than the original Assembly Calculus model.

What carries the argument

E%-winners-take-all selection combined with E/I-ratio inhibition, which selects active neurons according to statistical distributions consistent with power-law scaling rather than a fixed number.

If this is right

  • Assembly size is no longer fixed by a preset parameter k but emerges from the network's activity statistics.
  • Recovery of an assembly upon re-presentation of its generating stimulus exceeds the rate obtained with fixed k-winners-take-all selection.
  • The model incorporates observed features of gamma cycles and cortical E/I ratios, bringing its dynamics closer to biological networks.
  • Formed assemblies can represent external stimuli with greater flexibility because their membership is not dictated in advance.

Where Pith is reading between the lines

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

  • Dynamic sizing may allow assemblies to adjust automatically to inputs of varying complexity without manual retuning of k.
  • The approach could be tested by measuring whether assembly sizes in larger or more layered networks continue to follow the same emergent statistics.
  • Incorporating more detailed oscillatory timing or synaptic plasticity rules might further increase the biological realism of the recovery advantage.

Load-bearing premise

That the E%-winners-take-all rule and E/I-ratio inhibition accurately capture the statistical structure of gamma cycles and cortical neuron ratios, and that the simulation outcomes will generalize beyond the tested network sizes and parameters.

What would settle it

Direct comparison of emergent assembly sizes and stimulus-evoked recovery rates in the simulated model against empirical measurements from cortical recordings that show different size distributions or lower recovery success.

Figures

Figures reproduced from arXiv: 2603.12416 by Aline Villavicencio, Gustavo Soroka, Lucas Hoff, Marco Idiart, Matheus Guimar\~aes.

Figure 1
Figure 1. Figure 1: Structure and dynamics of the E%-WTA model. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Neural assembly formation in AC and E%-WTA models. [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Characteristics of the assemblies formed in the AC and E%-WTA models as a function of synaptic plasticity. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: E%-WTA model behavior as a function of a) the probability of synaptic connection (ps) and b) the stimulus size (ks) (200 simulations for each value of ps and ks with β = 0.01). Lower values lead to lower values of success rate, thereby limiting the applicability of operations defined in the original model. The success rate (normalized by the number of simulations) is defined here as the ratio between the n… view at source ↗
read the original abstract

As proposed by Hebb's theory, neural assemblies are groups of excitatory neurons that fire synchronously and exhibit high synaptic density, representing external stimuli and supporting cognitive functions such as language and decision-making. Recently, a model called Assembly Calculus (AC) was proposed, enabling the formation of artificial neural assemblies through the $k$-winners-take-all selection process and Hebbian learning. Although the model is capable of forming assemblies according to Hebb's theory, the adopted selection process does not incorporate essential aspects of biological neural computation, as neural activity, which is often governed by statistical distributions consistent with power-law scaling. Given this limitation, the present work aimed to bring the model's dynamics closer to that observed in real cortical networks. To achieve this, a new selection mechanism inspired by the dynamics of gamma oscillation cycles, called E%-winners-take-all, was implemented, combined with an inhibition process based on the ratio between excitatory and inhibitory neurons observed in various regions of the cerebral cortex. The results obtained from our model (called E%-WTA model) were compared with those of the original model, and the analyses demonstrated that the introduced modifications allowed the network's own dynamics to determine the size of the formed assemblies. Furthermore, the recovery rate of these groups, through the evocation of the stimuli that generated them, became superior to that obtained in the original model.

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

2 major / 2 minor

Summary. The paper proposes the E%-WTA model as an extension of Assembly Calculus, replacing k-winners-take-all with an E%-winners-take-all rule inspired by gamma-oscillation statistics and adding global inhibition scaled by the observed cortical E/I neuron ratio. It claims that these biologically motivated changes allow assembly sizes to emerge from recurrent network dynamics rather than from a fixed parameter, and that the resulting assemblies exhibit higher stimulus-evoked recovery rates than those formed by the original AC model.

Significance. If the central claims are substantiated with quantitative evidence, the work would strengthen the biological plausibility of computational models of Hebbian assembly formation by aligning selection and inhibition mechanisms more closely with cortical gamma cycles and E/I balance. This could improve the fidelity of large-scale simulations of memory and decision-making circuits, provided the size-emergence result generalizes beyond the tested regimes.

major comments (2)
  1. [Abstract and Results] Abstract and Results sections: the claims of 'superior' recovery rates and 'network's own dynamics determine the size' are asserted without any reported quantitative metrics, error bars, statistical tests, network size N, stimulus encoding details, or exact E% value. No tables or figures compare recovery percentages or assembly-size distributions between E%-WTA and the original k-WTA model, rendering the improvement unverifiable.
  2. [Methods] Methods and Model Definition: E% is introduced as a fixed hyper-parameter (analogous to k). The claim that assembly size emerges independently of modeler choice therefore requires explicit evidence that size is insensitive to modest variations in E% once E/I balance is held constant, or that size scales with N according to gamma-cycle statistics rather than E. No such sensitivity analysis or scaling plot is provided.
minor comments (2)
  1. [Methods] Clarify in the Methods section how the E%-WTA threshold is computed on each cycle and how it interacts with the E/I-ratio inhibition term; the current description leaves the precise update rule ambiguous.
  2. [Methods] Add a brief description of network size, connectivity density, and stimulus representation (e.g., number of input neurons per stimulus) so that the reported dynamics can be reproduced or compared with other AC implementations.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which highlight areas where our presentation can be strengthened. We address each major comment below and will revise the manuscript to include the requested quantitative details and analyses.

read point-by-point responses

  1. Referee: [Abstract and Results] Abstract and Results sections: the claims of 'superior' recovery rates and 'network's own dynamics determine the size' are asserted without any reported quantitative metrics, error bars, statistical tests, network size N, stimulus encoding details, or exact E% value. No tables or figures compare recovery percentages or assembly-size distributions between E%-WTA and the original k-WTA model, rendering the improvement unverifiable.

    Authors: We agree that the abstract and high-level results summary present the claims qualitatively. The full results section contains direct comparisons of recovery rates and assembly sizes, but to make these verifiable we will add explicit quantitative metrics (including mean recovery percentages with error bars), statistical tests, network size N, stimulus encoding details, and the exact E% value. A new table and figure will directly compare recovery percentages and assembly-size distributions between E%-WTA and k-WTA models. These additions will be included in the revised manuscript. revision: yes

  2. Referee: [Methods] Methods and Model Definition: E% is introduced as a fixed hyper-parameter (analogous to k). The claim that assembly size emerges independently of modeler choice therefore requires explicit evidence that size is insensitive to modest variations in E% once E/I balance is held constant, or that size scales with N according to gamma-cycle statistics rather than E. No such sensitivity analysis or scaling plot is provided.

    Authors: E% is indeed a fixed parameter chosen to match gamma-oscillation statistics, but the model is designed so that assembly size is then governed by recurrent dynamics and the fixed E/I ratio rather than being prescribed directly. To support this, we will add a sensitivity analysis demonstrating that assembly size remains largely insensitive to modest variations in E% (with E/I balance held constant) and include scaling plots versus network size N that align with gamma-cycle statistics. These will appear in the revised Methods and Results sections. revision: yes

Circularity Check

0 steps flagged

No significant circularity in derivation chain

full rationale

The paper defines E%-WTA and E/I-ratio inhibition from external biological observations (gamma-cycle statistics and cortical neuron ratios) and compares simulation outcomes directly to the independent Assembly Calculus baseline. No equation reduces claimed assembly-size emergence or recovery-rate improvement to a fitted parameter renamed as prediction, a self-citation loop, or a definitional tautology; the central results rest on explicit simulation runs whose inputs (E%, network size, learning rules) remain distinct from the reported metrics.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The model inherits Hebbian learning and the basic assembly-formation goal from prior literature while introducing two new tunable elements (E% threshold and E/I scaling factor) whose values are not reported in the abstract; no new physical entities are postulated.

free parameters (2)
  • E% threshold
    Percentage of neurons allowed to win per cycle; must be chosen to produce plausible assembly sizes and is not derived from first principles in the abstract.
  • E/I inhibition scaling
    Factor controlling inhibitory strength based on observed cortical ratios; appears as a free parameter fitted to biological statistics.
axioms (2)
  • standard math Hebbian learning: co-active neurons strengthen synapses
    Invoked as the plasticity rule for assembly formation, taken from the cited Assembly Calculus and Hebb's original postulate.
  • domain assumption Neural activity follows power-law distributions consistent with gamma oscillations
    Used to justify replacing k-WTA with E%-WTA; stated as observed in real cortex but not re-derived here.

pith-pipeline@v0.9.0 · 5557 in / 1569 out tokens · 50906 ms · 2026-05-15T11:26:43.859478+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

42 extracted references · 42 canonical work pages

  1. [1]

    D. O. Hebb,The organization of behavior: A neuropsycho- logical theory. New York: Wiley, Jun. 1949

    work page 1949

  2. [2]

    Abeles,Corticonics

    M. Abeles,Corticonics. Cambridge University Press, May 1991

    work page 1991

  3. [3]

    Endogenous sequential cortical activity evoked by visual stimuli,

    L. Carrillo-Reid, J.-E. K. Miller, J. P. Hamm, J. Jackson, and R. Yuste, “Endogenous sequential cortical activity evoked by visual stimuli,”J. Neurosci., vol. 35, no. 23, pp. 8813– 8828, Jun. 2015

    work page 2015

  4. [4]

    Visual stimuli recruit intrinsically generated cortical ensem- bles,

    J.-E. K. Miller, I. Ayzenshtat, L. Carrillo-Reid, and R. Yuste, “Visual stimuli recruit intrinsically generated cortical ensem- bles,”Proc. Natl. Acad. Sci. U. S. A., vol. 111, no. 38, pp. E4053–61, Sep. 2014

    work page 2014

  5. [5]

    Imprinting and recalling cortical ensembles,

    L. Carrillo-Reid, W. Yang, Y . Bando, D. S. Peterka, and R. Yuste, “Imprinting and recalling cortical ensembles,”Sci- ence, vol. 353, no. 6300, pp. 691–694, Aug. 2016

    work page 2016

  6. [6]

    Controlling visually guided behavior by holographic re- calling of cortical ensembles,

    L. Carrillo-Reid, S. Han, W. Yang, A. Akrouh, and R. Yuste, “Controlling visually guided behavior by holographic re- calling of cortical ensembles,”Cell, vol. 178, no. 2, pp. 447–457.e5, Jul. 2019

    work page 2019

  7. [7]

    A theory for cerebral neocortex,

    D. Marr, “A theory for cerebral neocortex,”Proc. R. Soc. Lond., vol. 176, no. 1043, pp. 161–234, Nov. 1970

    work page 1970

  8. [8]

    Simple memory: a theory for archicortex,

    D. Marr, “Simple memory: a theory for archicortex,”Phi- los. Trans. R. Soc. Lond. B Biol. Sci., vol. 262, no. 841, pp. 23–81, Jul. 1971

    work page 1971

  9. [9]

    Neuronal ensembles: Building blocks of neural circuits,

    R. Yuste, R. Cossart, and E. Yaksi, “Neuronal ensembles: Building blocks of neural circuits,”Neuron, vol. 112, no. 6, pp. 875–892, Mar. 2024

    work page 2024

  10. [10]

    Neural networks and physical systems with emergent collective computational abilities,

    J. J. Hopfield, “Neural networks and physical systems with emergent collective computational abilities,”Proc. Natl. Acad. Sci. U. S. A., vol. 79, no. 8, pp. 2554–2558, Apr. 1982

    work page 1982

  11. [11]

    Memory re- trieval time and memory capacity of the CA3 network: role of gamma frequency oscillations,

    L. de Almeida, M. Idiart, and J. E. Lisman, “Memory re- trieval time and memory capacity of the CA3 network: role of gamma frequency oscillations,”Learn. Mem., vol. 14, no. 11, pp. 795–806, Nov. 2007

    work page 2007

  12. [12]

    From the neuron doctrine to neural networks,

    R. Yuste, “From the neuron doctrine to neural networks,” Nat. Rev. Neurosci., vol. 16, no. 8, pp. 487–497, Aug. 2015

    work page 2015

  13. [13]

    Large-scale recording of neuronal ensembles,

    G. Buzsáki, “Large-scale recording of neuronal ensembles,” Nat. Neurosci., vol. 7, no. 5, pp. 446–451, May 2004

    work page 2004

  14. [14]

    Two-photon laser scanning fluorescence microscopy,

    W. Denk, J. H. Strickler, and W. W. Webb, “Two-photon laser scanning fluorescence microscopy,”Science, vol. 248, no. 4951, pp. 73–76, Apr. 1990

    work page 1990

  15. [15]

    Richard axel,

    N. Q&A, “Richard axel,”Neuron, vol. 99, no. 6, pp. 1110– 1112, Sep. 2018

    work page 2018

  16. [16]

    Brain com- putation by assemblies of neurons,

    C. H. Papadimitriou, S. S. Vempala, D. Mitropol- sky, M. Collins, and W. Maass, “Brain com- putation by assemblies of neurons,”Proceedings of the National Academy of Sciences, vol. 117, no. 25, pp. 14 464–14 472, 2020. [Online]. Available: https://www.pnas.org/doi/abs/10.1073/pnas.2001893117

    work page doi:10.1073/pnas.2001893117 2020

  17. [17]

    Random Projection in the Brain and Computation with Assemblies of Neurons,

    C. H. Papadimitriou and S. S. Vempala, “Random Projection in the Brain and Computation with Assemblies of Neurons,” in10th Innovations in Theoretical Computer Science Conference (ITCS 2019), ser. Leibniz International Proceedings in Informatics (LIPIcs), A. Blum, Ed., vol. 124. Dagstuhl, Germany: Schloss Dagstuhl–Leibniz-Zentrum fuer Informatik, 2018, pp....

    work page 2019

  18. [18]

    Long Term Memory and the Densest K-Subgraph Problem,

    R. Legenstein, W. Maass, C. H. Papadimitriou, and S. S. Vempala, “Long Term Memory and the Densest K-Subgraph Problem,” in9th Innovations in The- oretical Computer Science Conference (ITCS 2018), ser. Leibniz International Proceedings in Informatics (LIPIcs), A. R. Karlin, Ed., vol. 94. Dagstuhl, Germany: Schloss Dagstuhl–Leibniz-Zentrum fuer In- formatik...

    work page 2018

  19. [19]

    A biologically plausible parser,

    D. Mitropolsky, M. J. Collins, and C. H. Papadim- itriou, “A biologically plausible parser,”Transactions of the Association for Computational Linguistics, vol. 9, pp. 1374–1388, 2021. [Online]. Available: https://aclanthology.org/2021.tacl-1.81/

    work page 2021

  20. [20]

    Com- putation with sequences of assemblies in a model of the brain,

    M. Dabagia, C. H. Papadimitriou, and S. S. Vempala, “Com- putation with sequences of assemblies in a model of the brain,”Neural Comput., vol. 37, no. 1, pp. 193–233, Dec. 2024

    work page 2024

  21. [21]

    Effects of plasticity func- tions on neural assemblies,

    C. Constantinides and K. Nassar, “Effects of plasticity func- tions on neural assemblies,”arXiv, 2021

    work page 2021

  22. [22]

    Understanding the role of recur- rent connections in assembly calculus,

    A. Rangamani and Y . Xie, “Understanding the role of recur- rent connections in assembly calculus,” 07/2022 2022

    work page 2022

  23. [23]

    A second func- tion of gamma frequency oscillations: An e%-max winner- take-all mechanism selects which cells fire,

    L. de Almeida, M. Idiart, and J. E. Lisman, “A second func- tion of gamma frequency oscillations: An e%-max winner- take-all mechanism selects which cells fire,”Journal of Neu- roscience, vol. 29, no. 23, pp. 7497–7503, 2009. [Online]. Available: https://www.jneurosci.org/content/29/23/7497

    work page 2009

  24. [24]

    Thermodynamics and signatures of criticality in a network of neurons,

    G. Tka ˇcik, T. Mora, O. Marre, D. Amodei, S. E. Palmer, M. J. B. II, and W. Bialek, “Thermodynamics and signatures of criticality in a network of neurons,” Proceedings of the National Academy of Sciences, vol. 112, no. 37, pp. 11 508–11 513, 2015. [Online]. Available: https://www.pnas.org/doi/10.1073/pnas.1514188112

    work page doi:10.1073/pnas.1514188112 2015

  25. [25]

    Statistical mechanics for networks of real neurons,

    Anonymous, “Statistical mechanics for networks of real neurons,”Rev. Mod. Phys., Apr. 2025

    work page 2025

  26. [26]

    The gamma cycle,

    P. Fries, D. Nikoli ´c, and W. Singer, “The gamma cycle,” Trends Neurosci., vol. 30, no. 7, pp. 309–316, Jul. 2007

    work page 2007

  27. [27]

    Long term memory and the densest k-subgraph problem,

    R. Legenstein, W. Maass, C. Papapdimitriou, and S. Vem- pala, “Long term memory and the densest k-subgraph problem,” in9th Innovations in Theoretical Computer Sci- ence Conference (ITCS 2018), ser. LIPIcs-Leibniz Inter- national Proceedings in Informatics, vol. 94. Germany: Schloss Dagstuhl - Leibniz-Zentrum für Informatik, 2018, p. 57:1–57:15

    work page 2018

  28. [28]

    Mechanisms of gamma oscillations,

    G. Buzsáki and X.-J. Wang, “Mechanisms of gamma oscillations,”Annual Review of Neuroscience, vol. 35, no. 1, pp. 203–225, 2012, pMID: 22443509. [Online]. Available: https://doi.org/10.1146/annurev-neuro-062111-150444

    work page doi:10.1146/annurev-neuro-062111-150444 2012

  29. [29]

    The gamma rhythm as a guardian of brain health,

    A. M. Ichim, H. Barzan, V . V . Moca, A. Nagy-Dabacan, A. Ciuparu, A. Hapca, K. Vervaeke, and R. C. Muresan, “The gamma rhythm as a guardian of brain health,”eLife, vol. 13, p. e100238, nov 2024. [Online]. Available: https://doi.org/10.7554/eLife.100238

    work page doi:10.7554/elife.100238 2024

  30. [30]

    Stimulus-specific neuronal os- cillations in orientation columns of cat visual cortex,

    C. M. Gray and W. Singer, “Stimulus-specific neuronal os- cillations in orientation columns of cat visual cortex,”Proc. Natl. Acad. Sci. U. S. A., vol. 86, no. 5, pp. 1698–1702, Mar. 1989

    work page 1989

  31. [31]

    Modulation of oscillatory neuronal synchronization by se- lective visual attention,

    P. Fries, J. H. Reynolds, A. E. Rorie, and R. Desimone, “Modulation of oscillatory neuronal synchronization by se- lective visual attention,”Science, vol. 291, no. 5508, pp. 1560–1563, Feb. 2001

    work page 2001

  32. [32]

    Human gamma- frequency oscillations associated with attention and mem- ory,

    O. Jensen, J. Kaiser, and J.-P. Lachaux, “Human gamma- frequency oscillations associated with attention and mem- ory,”Trends Neurosci., vol. 30, no. 7, pp. 317–324, Jul. 2007

    work page 2007

  33. [33]

    Hippocampal gamma oscillations in- crease with memory load,

    M. K. van Vugt, A. Schulze-Bonhage, B. Litt, A. Brandt, and M. J. Kahana, “Hippocampal gamma oscillations in- crease with memory load,”J. Neurosci., vol. 30, no. 7, pp. 2694–2699, Feb. 2010

    work page 2010

  34. [34]

    Gamma oscillations and episodic memory,

    B. J. Griffiths and O. Jensen, “Gamma oscillations and episodic memory,”Trends Neurosci., vol. 46, no. 10, pp. 832–846, Oct. 2023

    work page 2023

  35. [35]

    Excitatory-inhibitory balance modulates the formation and dynamics of neuronal assem- blies in cortical networks,

    S. Sadeh and C. Clopath, “Excitatory-inhibitory balance modulates the formation and dynamics of neuronal assem- blies in cortical networks,”Sci. Adv., vol. 7, no. 45, p. eabg8411, Nov. 2021

    work page 2021

  36. [36]

    Synaptic E-I balance underlies efficient neural coding,

    S. Zhou and Y . Yu, “Synaptic E-I balance underlies efficient neural coding,”Front. Neurosci., vol. 12, Feb. 2018

    work page 2018

  37. [37]

    Neuronal circuits overcome imbalance in excitation and inhibition by adjusting connection num- bers,

    N. Sukenik, O. Vinogradov, E. Weinreb, M. Segal, A. Lev- ina, and E. Moses, “Neuronal circuits overcome imbalance in excitation and inhibition by adjusting connection num- bers,”Proc. Natl. Acad. Sci. U. S. A., vol. 118, no. 12, p. e2018459118, Mar. 2021

    work page 2021

  38. [38]

    Constrained brain volume in an efficient coding model explains the frac- tion of excitatory and inhibitory neurons in sensory cortices,

    A. Alreja, I. Nemenman, and C. J. Rozell, “Constrained brain volume in an efficient coding model explains the frac- tion of excitatory and inhibitory neurons in sensory cortices,” PLoS Comput. Biol., vol. 18, no. 1, p. e1009642, Jan. 2022

    work page 2022

  39. [39]

    The fraction of cortical GABAergic neurons is constant from near the start of cortical neurogenesis to adult- hood,

    S. Sahara, Y . Yanagawa, D. D. M. O’Leary, and C. F. Stevens, “The fraction of cortical GABAergic neurons is constant from near the start of cortical neurogenesis to adult- hood,”J. Neurosci., vol. 32, no. 14, pp. 4755–4761, Apr. 2012

    work page 2012

  40. [40]

    Could electrical coupling contribute to the for- mation of cell assemblies?

    R. D. Traub, M. A. Whittington, N. Maier, D. Schmitz, and J. I. Nagy, “Could electrical coupling contribute to the for- mation of cell assemblies?”Rev. Neurosci., vol. 31, no. 2, pp. 121–141, Jan. 2020

    work page 2020

  41. [41]

    Inhibitory connectivity defines the realm of excitatory plasticity,

    G. Mongillo, S. Rumpel, and Y . Loewenstein, “Inhibitory connectivity defines the realm of excitatory plasticity,”Nat. Neurosci., vol. 21, no. 10, pp. 1463–1470, Oct. 2018

    work page 2018

  42. [42]

    Overcoming catastrophic forget- ting in neural networks,

    J. Kirkpatrick, R. Pascanu, N. Rabinowitz, J. Veness, G. Desjardins, A. A. Rusu, K. Milan, J. Quan, T. Ramalho, A. Grabska-Barwinska, D. Hassabis, C. Clopath, D. Ku- maran, and R. Hadsell, “Overcoming catastrophic forget- ting in neural networks,”Proc. Natl. Acad. Sci. U. S. A., vol. 114, no. 13, pp. 3521–3526, Mar. 2017. 9

    work page 2017