pith. sign in

arxiv: 2605.14680 · v1 · pith:3COJSTHQnew · submitted 2026-05-14 · 🧬 q-bio.NC

Are cortical microcircuits optimized for information flux? -- A simulation-based reverse engineering study

Claus Metzner , Ali Ghebleh , Karin Prebeck , Achim Schilling , Andreas Maier , Thomas Kinfe , Patrick Krauss This is my paper

Pith reviewed 2026-06-30 19:45 UTC · model grok-4.3

classification 🧬 q-bio.NC
keywords cortical microcircuitsinformation fluxrecurrence resonanceneural network modellayer 5mutual informationreservoir computers
0
0 comments X

The pith

A larger embedding network around a dense core population markedly increases the core's information flux through biases and stochastic drive.

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

The paper asks whether biological cortical microcircuits might be organized to maximize information flux between successive states. In a simulation model of layer 5, a densely connected core sits inside a larger supporting network. The embedding network is found to raise flux by shifting core neurons into a higher-entropy regime via effective biases and by injecting fluctuations that avoid trapping in simple attractors through recurrence resonance. The flux can be pushed higher still with individually tuned biases, which themselves can arise via a basic self-organization rule.

Core claim

In the model, the embedding network exerts a pronounced flux-enhancing effect on the core dynamics by generating effective biases that shift core neurons into a higher-entropy operating regime and by supplying stochastic fluctuations that prevent trapping in simple fixed-point or oscillatory attractors through the mechanism of Recurrence Resonance.

What carries the argument

The embedding network, which generates effective biases and stochastic fluctuations to enable recurrence resonance in the core.

If this is right

  • Information flux rises further when core neurons receive individually optimized biases.
  • Such biases can emerge from a simple self-organization principle.
  • The mechanism is relevant for interpreting biological neural circuits and for designing artificial recurrent systems such as reservoir computers.

Where Pith is reading between the lines

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

  • This embedding structure may represent a general design principle for recurrent networks to sustain high-entropy dynamics.
  • Similar embedding could improve performance in engineered reservoir computers by mimicking the biological case.
  • Testing whether real cortical tissue shows higher flux than de-embedded cores would directly check the model's prediction.

Load-bearing premise

The simplified model of cortical layer 5 architecture, with its densely interconnected core embedded in a larger supporting network, sufficiently captures the structural and dynamical features relevant to information flux in real biological microcircuits.

What would settle it

Direct measurement of mutual information between successive states in biological cortical layer 5 circuits, compared against versions where the surrounding network is removed or silenced.

Figures

Figures reproduced from arXiv: 2605.14680 by Achim Schilling, Ali Ghebleh, Andreas Maier, Claus Metzner, Karin Prebeck, Patrick Krauss, Thomas Kinfe.

Figure 1
Figure 1. Figure 1: The model system. • (a) Three neural sub-populations and their connections. • (b) Connection weight matrix of isolated core. • (c) Connection weight matrix of the full network. • (d) Distributions of scaled input sums p(u/t0) in the three sub-populations (using wi = −5), for two different neural temperatures t0 =1 and t0 =5. • (e) Intra- and inter-triplet information flux in the embedded core as a function… view at source ↗
Figure 2
Figure 2. Figure 2 [PITH_FULL_IMAGE:figures/full_fig_p022_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Simulated lesion experiments. Specific directional connections between the three neural populations (compare circuit diagram on top) are cut and the effect on the information flux in the core network is investigated. The left two columns (a,c) show the effect on the intra-triplet flux and right two columns (b,d) on the inter-triplet flux. For the final magenta bars in (c) and (d), all outgoing connections … view at source ↗
Figure 4
Figure 4. Figure 4: Activation statistics in the full embedded core network. • (a) Activations of all 125 Boltzmann neurons over 100 successive time steps, with active states marked by black dots. • (b) Matrix of instantaneous pairwise mutual information for the full 125 × 125 neuron system. • (c) Reduced mutual information matrix obtained from panel (b) by averaging over all individual neuron pairs within each combination of… view at source ↗
Figure 5
Figure 5. Figure 5: Analysis of Signals from the Embedding Networks to the Core. • (a) Probability density distributions of the total synaptic input ui,emb(t) from the embedding networks, that is, from the interneurons and the peripheral population, shown separately for each core neuron i. Continuous distributions were obtained using Gaussian kernel density estimation with a kernel width of 0.1 times the standard deviation of… view at source ↗
Figure 6
Figure 6. Figure 6: Effect of Noise and Biases from the Embedding Networks. • (a) Information flux in the core, measured by a flux indicator, in five different configurations: Core receiving the full signals from the embedding networks; Core receiving only temporal averages (biases); Biases plus normal distributed noise; Uniform (bi =b) optimal biases; Individually optimized biases. • (b) Flux indicator as a function of a uni… view at source ↗
read the original abstract

A sufficiently large information flux in recurrent neural networks, quantified by the mutual information between successive network states, is considered a prerequisite for rich information processing capabilities. This raises the question of whether biological neural networks, such as cortical microcolumns, may be structurally organized to enhance information flux. To investigate this possibility, we study a simplified model of the cortical layer 5 architecture, in which a densely and strongly interconnected core population is embedded within a larger supporting network. Surprisingly, we find that the embedding network exerts a pronounced flux-enhancing effect on the core dynamics. Systematic reverse-engineering analyses reveal that the embedding network provides two key contributions: first, it generates effective biases that shift core neurons into a higher-entropy operating regime; second, it supplies stochastic fluctuations that prevent the network from becoming trapped in simple fixed-point or oscillatory attractors through the mechanism of Recurrence Resonance. We further show that the information flux can be increased even beyond the biologically embedded case by applying individually optimized biases to the core neurons, and that these biases can emerge from a simple self-organization principle. Our findings are relevant both for the functional interpretation of biological neural circuits and for the design of artificial recurrent systems such as reservoir computers.

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

3 major / 2 minor

Summary. The paper claims that in a simplified model of cortical layer 5, a densely interconnected core embedded in a larger supporting network exhibits enhanced information flux (mutual information between successive states) due to effective biases shifting neurons to higher-entropy regimes and stochastic fluctuations enabling 'Recurrence Resonance' to avoid simple attractors; it further claims this flux can be increased via individually optimized biases arising from a self-organization principle, with implications for biological circuit interpretation and reservoir computing design.

Significance. If the model architecture and flux measure prove representative of biological layer 5 and the enhancement is robust, the work would provide a mechanistic account of how embedding connectivity could support rich dynamics, offering testable predictions for circuit optimization and design rules for artificial recurrent systems. The reverse-engineering approach is a methodological strength when accompanied by parameter sweeps and external benchmarks.

major comments (3)
  1. [Model architecture] Model architecture (Methods/Results sections): the claim that the embedding network exerts a 'pronounced flux-enhancing effect' relies on a core-plus-surround topology, but the manuscript provides no comparison of the chosen connection probabilities, weight distributions, or neuron-type ratios to empirical layer-5 statistics from paired recordings or EM reconstructions; without this, the enhancement may be an artifact of the specific dense-core parameters rather than a general feature of cortical architecture.
  2. [Results] Information flux definition and Recurrence Resonance (abstract and §Results): the mutual-information measure is defined internally to the simulations and the mechanism is identified post-hoc; the manuscript does not report external benchmarks, alternative flux measures, or controls that would demonstrate the enhancement is not reducible to quantities fitted within the same model runs.
  3. [Results] Validation and robustness (Results): the abstract and described analyses report no quantitative parameter values, statistical controls, sensitivity analyses across connection strengths, or direct validation against biological firing statistics, leaving the central flux-enhancement claim difficult to assess for robustness or artifact.
minor comments (2)
  1. [Abstract] The abstract would benefit from one or two quantitative effect sizes (e.g., fold-change in mutual information) to allow readers to gauge the magnitude of the reported enhancement.
  2. [Methods] Notation for the information-flux quantity and the precise definition of 'Recurrence Resonance' should be introduced with an equation in the main text rather than left implicit.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We are grateful to the referee for their thorough review and valuable suggestions, which have helped us identify areas for improvement. Below, we address each major comment in detail and outline the changes we plan to implement in the revised version of the manuscript.

read point-by-point responses

  1. Referee: [Model architecture] Model architecture (Methods/Results sections): the claim that the embedding network exerts a 'pronounced flux-enhancing effect' relies on a core-plus-surround topology, but the manuscript provides no comparison of the chosen connection probabilities, weight distributions, or neuron-type ratios to empirical layer-5 statistics from paired recordings or EM reconstructions; without this, the enhancement may be an artifact of the specific dense-core parameters rather than a general feature of cortical architecture.

    Authors: We acknowledge that the manuscript does not include direct comparisons to empirical layer-5 statistics. Our model is a simplified representation designed to investigate the general principle of embedding effects on information flux. In the revision, we will add to the Methods section references to empirical studies on cortical connectivity and discuss how our parameters are consistent with reported biological ranges. This will help mitigate concerns about the results being specific to arbitrary choices. revision: partial

  2. Referee: [Results] Information flux definition and Recurrence Resonance (abstract and §Results): the mutual-information measure is defined internally to the simulations and the mechanism is identified post-hoc; the manuscript does not report external benchmarks, alternative flux measures, or controls that would demonstrate the enhancement is not reducible to quantities fitted within the same model runs.

    Authors: While the mutual information measure is defined based on the simulation states, it is a standard and objective metric for information flux. The identification of Recurrence Resonance was based on systematic variations and ablations. We will revise the Results section to include alternative flux measures (e.g., state entropy and transfer entropy) and additional controls with randomized networks to demonstrate that the enhancement is due to the architecture. revision: yes

  3. Referee: [Results] Validation and robustness (Results): the abstract and described analyses report no quantitative parameter values, statistical controls, sensitivity analyses across connection strengths, or direct validation against biological firing statistics, leaving the central flux-enhancement claim difficult to assess for robustness or artifact.

    Authors: The manuscript includes quantitative results, but we agree that more explicit statistical controls and sensitivity analyses are warranted. We will add sensitivity analyses varying connection strengths and report means and standard deviations from repeated simulations. For direct validation against biological firing statistics, this lies outside the scope of the current simulation study; we will include a new subsection on model limitations and qualitative comparisons to known cortical dynamics. revision: partial

Circularity Check

0 steps flagged

No circularity in simulation-based derivation chain

full rationale

The paper defines information flux as mutual information between successive network states, then runs simulations on a simplified L5 model to observe that embedding enhances flux via biases and stochastic fluctuations (named Recurrence Resonance). These effects are emergent outputs of the model dynamics, not presupposed or fitted to force the result. No load-bearing step reduces a claimed prediction to a self-defined quantity, fitted input, or self-citation chain; the central claims rest on explicit simulation comparisons rather than tautological re-expression of inputs. The derivation is therefore self-contained.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Insufficient detail in the abstract to enumerate free parameters, axioms, or invented entities; the model architecture and information flux definition are treated as given without explicit justification.

pith-pipeline@v0.9.1-grok · 5762 in / 1137 out tokens · 18245 ms · 2026-06-30T19:45:10.418536+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

51 extracted references · 3 canonical work pages · 1 internal anchor

  1. [1]

    Deep learning.nature, 521(7553):436–444, 2015

    Yann LeCun, Yoshua Bengio, and Geoffrey Hinton. Deep learning.nature, 521(7553):436–444, 2015

    2015

  2. [2]

    Re- view of deep learning: Concepts, cnn architectures, challenges, applications, future directions

    Laith Alzubaidi, Jinglan Zhang, Amjad J Humaidi, Ayad Al-Dujaili, Ye Duan, Omran Al- Shamma, Jos´ e Santamar´ ıa, Mohammed A Fadhel, Muthana Al-Amidie, and Laith Farhan. Re- view of deep learning: Concepts, cnn architectures, challenges, applications, future directions. Journal of big Data, 8(1):1–74, 2021

    2021

  3. [3]

    Recent advances in natural language pro- cessing via large pre-trained language models: A survey.ACM Computing Surveys, 56(2):1–40, 2023

    Bonan Min, Hayley Ross, Elior Sulem, Amir Pouran Ben Veyseh, Thien Huu Nguyen, Oscar Sainz, Eneko Agirre, Ilana Heintz, and Dan Roth. Recent advances in natural language pro- cessing via large pre-trained language models: A survey.ACM Computing Surveys, 56(2):1–40, 2023

    2023

  4. [4]

    Universality and individuality in neural dynamics across large populations of recurrent networks

    Niru Maheswaranathan, Alex H Williams, Matthew D Golub, Surya Ganguli, and David Sussillo. Universality and individuality in neural dynamics across large populations of recurrent networks. Advances in neural information processing systems, 2019:15629, 2019

    2019

  5. [5]

    Recurrent neural networks are univer- sal approximators

    Anton Maximilian Sch¨ afer and Hans Georg Zimmermann. Recurrent neural networks are univer- sal approximators. InInternational Conference on Artificial Neural Networks, pages 632–640. Springer, 2006

    2006

  6. [6]

    Universal approximation of flows of control systems by recurrent neural networks

    Miguel Aguiar, Amritam Das, and Karl H Johansson. Universal approximation of flows of control systems by recurrent neural networks. In2023 62nd IEEE Conference on Decision and Control (CDC), pages 2320–2327. IEEE, 2023

    2023

  7. [7]

    echo state

    Herbert Jaeger. The “echo state” approach to analysing and training recurrent neural networks- with an erratum note.Bonn, Germany: German National Research Center for Information Technology GMD Technical Report, 148(34):13, 2001

    2001

  8. [8]

    Optimal sequence memory in driven random networks.Physical Review X, 8(4):041029, 2018

    Jannis Schuecker, Sven Goedeke, and Moritz Helias. Optimal sequence memory in driven random networks.Physical Review X, 8(4):041029, 2018

    2018

  9. [9]

    Connectivity, dynamics, and memory in reservoir computing with binary and analog neurons.Neural computation, 22(5):1272–1311, 2010

    Lars B¨ using, Benjamin Schrauwen, and Robert Legenstein. Connectivity, dynamics, and memory in reservoir computing with binary and analog neurons.Neural computation, 22(5):1272–1311, 2010

    2010

  10. [10]

    Information process- ing capacity of dynamical systems.Scientific reports, 2(1):1–7, 2012

    Joni Dambre, David Verstraeten, Benjamin Schrauwen, and Serge Massar. Information process- ing capacity of dynamical systems.Scientific reports, 2(1):1–7, 2012

    2012

  11. [11]

    Randomly connected networks have short temporal memory.Neural computation, 25(6):1408–1439, 2013

    Edward Wallace, Hamid Reza Maei, and Peter E Latham. Randomly connected networks have short temporal memory.Neural computation, 25(6):1408–1439, 2013

    2013

  12. [12]

    Fading memory echo state networks are universal.Neural Networks, 138:10–13, 2021

    Lukas Gonon and Juan-Pablo Ortega. Fading memory echo state networks are universal.Neural Networks, 138:10–13, 2021

    2021

  13. [13]

    Gradient-based learning drives robust representations in recurrent neural networks by balancing compression and expansion.Nature Machine Intelligence, 4(6):564–573, 2022

    Matthew Farrell, Stefano Recanatesi, Timothy Moore, Guillaume Lajoie, and Eric Shea-Brown. Gradient-based learning drives robust representations in recurrent neural networks by balancing compression and expansion.Nature Machine Intelligence, 4(6):564–573, 2022

    2022

  14. [14]

    Stimulus-dependent suppression of chaos in recurrent neural networks.Physical Review E, 82(1):011903, 2010

    Kanaka Rajan, LF Abbott, and Haim Sompolinsky. Stimulus-dependent suppression of chaos in recurrent neural networks.Physical Review E, 82(1):011903, 2010

    2010

  15. [15]

    arXiv preprint arXiv:1403.3369 , year=

    Herbert Jaeger. Controlling recurrent neural networks by conceptors.arXiv preprint arXiv:1403.3369, 2014. 17

    work page arXiv 2014

  16. [16]

    Understanding and controlling memory in recurrent neural networks

    Doron Haviv, Alexander Rivkind, and Omri Barak. Understanding and controlling memory in recurrent neural networks. InInternational Conference on Machine Learning, pages 2663–2671. PMLR, 2019

    2019

  17. [17]

    Suppressing chaos in neural networks by noise.Physical review letters, 69(26):3717, 1992

    Lutz Molgedey, Johannes Schuchhardt, and Heinz Georg Schuster. Suppressing chaos in neural networks by noise.Physical review letters, 69(26):3717, 1992

    1992

  18. [18]

    Noise-modulated neural networks as an application of stochastic resonance.Neurocomputing, 277:29–37, 2018

    Shuhei Ikemoto, Fabio DallaLibera, and Koh Hosoda. Noise-modulated neural networks as an application of stochastic resonance.Neurocomputing, 277:29–37, 2018

    2018

  19. [19]

    Recurrence resonance” in three-neuron motifs.Frontiers in computational neuroscience, 13, 2019

    Patrick Krauss, Karin Prebeck, Achim Schilling, and Claus Metzner. Recurrence resonance” in three-neuron motifs.Frontiers in computational neuroscience, 13, 2019

    2019

  20. [20]

    Control of noise- induced coherent oscillations in time-delayed neural motifs.arXiv preprint arXiv:2106.11361, 2021

    Florian B¨ onsel, Patrick Krauss, Claus Metzner, and Marius E Yamakou. Control of noise- induced coherent oscillations in time-delayed neural motifs.arXiv preprint arXiv:2106.11361, 2021

    work page arXiv 2021

  21. [21]

    Dynamics and information import in recurrent neural net- works.Frontiers in Computational Neuroscience, 16, 2022

    Claus Metzner and Patrick Krauss. Dynamics and information import in recurrent neural net- works.Frontiers in Computational Neuroscience, 16, 2022

    2022

  22. [22]

    Recurrent neural networks as versatile tools of neuroscience research.Current opinion in neurobiology, 46:1–6, 2017

    Omri Barak. Recurrent neural networks as versatile tools of neuroscience research.Current opinion in neurobiology, 46:1–6, 2017

    2017

  23. [23]

    Highly nonrandom features of synaptic connectivity in local cortical circuits.PLoS biology, 3(3):e68, 2005

    Sen Song, Per Jesper Sj¨ ostr¨ om, Markus Reigl, Sacha Nelson, and Dmitri B Chklovskii. Highly nonrandom features of synaptic connectivity in local cortical circuits.PLoS biology, 3(3):e68, 2005

    2005

  24. [24]

    Is cortical connectivity optimized for storing information?Nature neuroscience, 19(5):749–755, 2016

    Nicolas Brunel. Is cortical connectivity optimized for storing information?Nature neuroscience, 19(5):749–755, 2016

    2016

  25. [25]

    Exploring Sparsity in Recurrent Neural Networks

    Sharan Narang, Erich Elsen, Gregory Diamos, and Shubho Sengupta. Exploring sparsity in recurrent neural networks.arXiv preprint arXiv:1704.05119, 2017

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  26. [26]

    Sparsity through evolutionary pruning prevents neuronal networks from overfitting.Neural Networks, 128:305– 312, 2020

    Richard C Gerum, Andr´ e Erpenbeck, Patrick Krauss, and Achim Schilling. Sparsity through evolutionary pruning prevents neuronal networks from overfitting.Neural Networks, 128:305– 312, 2020

    2020

  27. [27]

    Effect of dilution in asym- metric recurrent neural networks.Neural Networks, 104:50–59, 2018

    Viola Folli, Giorgio Gosti, Marco Leonetti, and Giancarlo Ruocco. Effect of dilution in asym- metric recurrent neural networks.Neural Networks, 104:50–59, 2018

    2018

  28. [28]

    Analysis of structure and dynamics in three-neuron motifs.Frontiers in Computational Neuroscience, 13:5, 2019

    Patrick Krauss, Alexandra Zankl, Achim Schilling, Holger Schulze, and Claus Metzner. Analysis of structure and dynamics in three-neuron motifs.Frontiers in Computational Neuroscience, 13:5, 2019

    2019

  29. [29]

    Weight statistics controls dynamics in recurrent neural networks.PloS one, 14(4):e0214541, 2019

    Patrick Krauss, Marc Schuster, Verena Dietrich, Achim Schilling, Holger Schulze, and Claus Met- zner. Weight statistics controls dynamics in recurrent neural networks.PloS one, 14(4):e0214541, 2019

    2019

  30. [30]

    Nonlinear neural dynamics and classification accuracy in reservoir computing.Neural Computation, 37(8):1469–1504, 2025

    Claus Metzner, Achim Schilling, Andreas Maier, and Patrick Krauss. Nonlinear neural dynamics and classification accuracy in reservoir computing.Neural Computation, 37(8):1469–1504, 2025

    2025

  31. [31]

    Organizational regularities in recurrent neural networks.Frontiers in Complex Systems, 3:1636222, 2025

    Claus Metzner, Achim Schilling, Andreas Maier, and Patrick Krauss. Organizational regularities in recurrent neural networks.Frontiers in Complex Systems, 3:1636222, 2025

    2025

  32. [32]

    Modular brain networks.Annual Review of Psychology, 67:613–640, 2016

    Olaf Sporns and Richard F Betzel. Modular brain networks.Annual Review of Psychology, 67:613–640, 2016

    2016

  33. [33]

    Modular and hierarchically mod- ular organization of brain networks.Frontiers in Neuroscience, 4:200, 2010

    David Meunier, Renaud Lambiotte, and Edward T Bullmore. Modular and hierarchically mod- ular organization of brain networks.Frontiers in Neuroscience, 4:200, 2010. 18

    2010

  34. [34]

    Neural networks and physical systems with emergent collective computational abilities.Proceedings of the national academy of sciences, 79(8):2554–2558, 1982

    John J Hopfield. Neural networks and physical systems with emergent collective computational abilities.Proceedings of the national academy of sciences, 79(8):2554–2558, 1982

    1982

  35. [35]

    Intrinsic noise improves speech recognition in a computational model of the auditory pathway.Frontiers in Neuroscience, 16:908330, 2022

    Achim Schilling, Richard Gerum, Claus Metzner, Andreas Maier, and Patrick Krauss. Intrinsic noise improves speech recognition in a computational model of the auditory pathway.Frontiers in Neuroscience, 16:908330, 2022

    2022

  36. [36]

    Patrick Krauss, Konstantin Tziridis, Claus Metzner, Achim Schilling, Ulrich Hoppe, and Holger Schulze. Stochastic resonance controlled upregulation of internal noise after hearing loss as a putative cause of tinnitus-related neuronal hyperactivity.Frontiers in neuroscience, 10:597, 2016

    2016

  37. [37]

    Achim Schilling, Konstantin Tziridis, Holger Schulze, and Patrick Krauss. The stochastic res- onance model of auditory perception: A unified explanation of tinnitus development, zwicker tone illusion, and residual inhibition.Progress in Brain Research, 262:139–157, 2021

    2021

  38. [38]

    Predictive coding and stochastic resonance as fundamental principles of auditory phantom perception.Brain, 146(12):4809–4825, 2023

    Achim Schilling, William Sedley, Richard Gerum, Claus Metzner, Konstantin Tziridis, Andreas Maier, Holger Schulze, Fan-Gang Zeng, Karl J Friston, and Patrick Krauss. Predictive coding and stochastic resonance as fundamental principles of auditory phantom perception.Brain, 146(12):4809–4825, 2023

    2023

  39. [39]

    Recurrence resonance- noise-enhanced dynamics in recurrent neural networks.Frontiers in Complex Systems, 2:1479417, 2024

    Claus Metzner, Achim Schilling, Andreas Maier, and Patrick Krauss. Recurrence resonance- noise-enhanced dynamics in recurrent neural networks.Frontiers in Complex Systems, 2:1479417, 2024

    2024

  40. [40]

    Computation at the edge of chaos: Phase transitions and emergent compu- tation.Physica D: Nonlinear Phenomena, 42(1-3):12–37, 1990

    Chris G Langton. Computation at the edge of chaos: Phase transitions and emergent compu- tation.Physica D: Nonlinear Phenomena, 42(1-3):12–37, 1990

    1990

  41. [41]

    Real-time computation at the edge of chaos in recurrent neural networks.Neural computation, 16(7):1413–1436, 2004

    Nils Bertschinger and Thomas Natschl¨ ager. Real-time computation at the edge of chaos in recurrent neural networks.Neural computation, 16(7):1413–1436, 2004

    2004

  42. [42]

    The mechanism of stochastic resonance

    Roberto Benzi, Alfonso Sutera, and Angelo Vulpiani. The mechanism of stochastic resonance. Journal of Physics A: mathematical and general, 14(11):L453–L457, 1981

    1981

  43. [43]

    Stochastic resonance and the benefits of noise: from ice ages to crayfish and squids.Nature, 373(6509):33–36, 1995

    Kurt Wiesenfeld and Frank Moss. Stochastic resonance and the benefits of noise: from ice ages to crayfish and squids.Nature, 373(6509):33–36, 1995

    1995

  44. [44]

    drivers” from “modulators

    S Murray Sherman and RW Guillery. On the actions that one nerve cell can have on another: distinguishing “drivers” from “modulators”.Proceedings of the National Academy of Sciences, 95(12):7121–7126, 1998

    1998

  45. [45]

    Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects.Nature neuroscience, 2(1):79–87, 1999

    Rajesh PN Rao and Dana H Ballard. Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects.Nature neuroscience, 2(1):79–87, 1999

    1999

  46. [46]

    A theory of cortical responses.Philosophical transactions of the Royal Society B: Biological sciences, 360(1456):815–836, 2005

    Karl Friston. A theory of cortical responses.Philosophical transactions of the Royal Society B: Biological sciences, 360(1456):815–836, 2005

    2005

  47. [47]

    Homeostatic plasticity in the developing nervous system

    Gina G Turrigiano and Sacha B Nelson. Homeostatic plasticity in the developing nervous system. Nature reviews neuroscience, 5(2):97–107, 2004

    2004

  48. [48]

    Theory for the development of neu- ron selectivity: orientation specificity and binocular interaction in visual cortex.Journal of Neuroscience, 2(1):32–48, 1982

    Elie L Bienenstock, Leon N Cooper, and Paul W Munro. Theory for the development of neu- ron selectivity: orientation specificity and binocular interaction in visual cortex.Journal of Neuroscience, 2(1):32–48, 1982

    1982

  49. [49]

    The self-tuning neuron: synaptic scaling of excitatory synapses.Cell, 135(3):422–435, 2008

    Gina G Turrigiano. The self-tuning neuron: synaptic scaling of excitatory synapses.Cell, 135(3):422–435, 2008. 19

    2008

  50. [50]

    Detect- ing rich-club ordering in complex networks.Nature physics, 2(2):110–115, 2006

    Vittoria Colizza, Alessandro Flammini, M Angeles Serrano, and Alessandro Vespignani. Detect- ing rich-club ordering in complex networks.Nature physics, 2(2):110–115, 2006

    2006

  51. [51]

    inter” and column “peri

    Martijn P Van Den Heuvel and Olaf Sporns. Rich-club organization of the human connectome. Journal of Neuroscience, 31(44):15775–15786, 2011. 20 core periphery interneurons t0 = 1 t0 = 5 (a) (b) (c) (d) (e) intra-triplet inter-triplet intra-triplet inter-triplet (f) Figure 1:The model system.•(a)Three neural sub-populations and their connections.•(b) Conne...

    2011