OSCR

Hierarchical neural circuit theory of normalization and inter-areal communication.

Code ↔ Paper

9 matches between paragraphs of the paper and lines of its authors' code, computed by the harvester (lexical-v1). Click a colored paragraph or line to see its counterpart.

6 blocks of code here have a permanent identifier (oscr:b:…), listed with the matches below: citable for good, with a page of its own. Select a match and press Cite to copy it. What a block identifier is.

The 9 matches
  1. [1] § Methods › The model ↔ Utils/Create_weight_matrices.py, lines 84 to 181 · score 0.73 · encoding matrix, receptive fields, weight matrix, connectivity matrix, transpose, symmetric · block oscr:b:MTTSC9Q6QFCEN-P · Report a mismatch
  2. [2] § Methods › Communication subspaces › Interpretation of the results ↔ Utils/matrix_spectrum.py, lines 34 to 167 · score 0.72 · spectral density matrix, auto spectra, cross power spectrum, squared, coherence · block oscr:b:874K8K9NMX8SE-M · Report a mismatch
  3. [3] § Methods › The model ↔ Utils/Create_weight_matrices.py, lines 84 to 181 · score 0.70 · raised cosine, encoding matrix, ring model, stimulus, connectivity, recurrent · block oscr:b:MTTSC9Q6QFCEN-P · Report a mismatch
  4. [4] § Methods › The model ↔ Three_area/Parameters_3.py, lines 84 to 155 · score 0.70 · raised cosine, encoding matrix, ring model, stimulus, connectivity, recurrent · block oscr:b:HAXZQPNN8ZDV8-V · Report a mismatch
  5. [5] § Methods › The model ↔ Three_area/Parameters_3.py, lines 84 to 155 · score 0.69 · connectivity matrix, normalization matrix, inter areal, transpose, symmetric, Gaussian · block oscr:b:HAXZQPNN8ZDV8-V · Report a mismatch
  6. [6] § Methods › The model ↔ Three_area/Communication_3.py, lines 52 to 114 · score 0.62 · Gaussian rectification, membrane potentials, firing rates, variable, matrix, model · block oscr:b:DY2H28JZMAD67-T · Report a mismatch
  7. [7] § Methods › Communication subspaces › Interpretation of the results ↔ Utils/Communication.py, lines 351 to 379 · score 0.56 · left singular vectors, eigenvectors, Alignment · block oscr:b:HNDCJR8VKNFA9-1 · Report a mismatch
  8. [8] § Results › Theory › The core mechanism in brief ↔ Utils/Communication.py, lines 167 to 253 · score 0.55 · reduced rank regression, covariance matrix, cross · block oscr:b:76GTGT44CJDKN-Q · Report a mismatch
  9. [9] § Methods › Frequency decomposition of the communication subspace ↔ Utils/matrix_spectrum.py, lines 34 to 167 · score 0.54 · power spectral density, PSD, Wiener, inverse, matrix · block oscr:b:874K8K9NMX8SE-M · Report a mismatch

Paper

Loaded from Europe PMC by your browser; OSCR does not store it: doi.org · Europe PMC

The paper is loaded when this pane is shown.

The authors' code

Python · 182 lines · 6.4 KB · MIT · 2 matches

  1. import numpy as np
  2. import math
  3. def Norm_matrix(N,std_dev,gaussian_height):
  4. '''Gaussian Normalization pool within each layers'''
  5. # std_dev = std_dev
  6. convMat = np.zeros((N, N))
  7. for row in range(N):
  8. # Instead of generating a kernel upfront, we'll compute the Gaussian values with wrap-around in this loop
  9. for col in range(N):
  10. # Calculate the distance considering the wrap-around (circular condition)
  11. circular_distance = min(abs(col - row), N - abs(col - row))
  12. # Gaussian value calculation based on the circular distance
  13. value = (1 / (std_dev * np.sqrt(2 * np.pi))) * np.exp(-0.5 * (circular_distance ** 2) / (std_dev ** 2))
  14. convMat[row, col] = value
  15. # Normalize the row to make the sum of its elements equal to 1.
  16. convMat[row] /= np.max(convMat[row])
  17. convMat[row] = convMat[row] * gaussian_height
  18. return (convMat + convMat.T)/2.0
  19. def Recurrence_matrix(N, kernel):
  20. '''Inter-areal recurrent excitation matrix'''
  21. K = len(kernel)
  22. convMat = np.zeros((N, N))
  23. for row in range(N):
  24. i = 0
  25. for col in range(int(row - np.floor(K / 2)), int(row + np.ceil(K / 2))):
  26. # print(col)
  27. convMat[row, np.remainder(col + N, N)] = kernel[i]
  28. i = i + 1
  29. return convMat
  30. def RescaleEigenvalues(convMat):
  31. '''Rescale the eigenvalues recurrent excitation matrix'''
  32. Utmp, Stmp, Vh = np.linalg.svd(convMat)
  33. Stmp[Stmp > 1] = 1 # Stmp is 1d vector
  34. Stmp = Stmp * 1.0
  35. n_convMat = (Utmp @ np.diag(Stmp) @ Vh)
  36. return n_convMat
  37. def ReceptiveFields(N,theta,M):
  38. '''Receptive fields matrix'''
  39. d = int(N / 2)
  40. pint = d - 1
  41. if pint < 1:
  42. pint = 1
  43. const = np.sqrt(d / N) * np.sqrt((2 ** (2 * pint) * (math.factorial(pint)) ** 2)
  44. / (math.factorial(2 * pint) * (pint + 1)))
  45. RFs = np.zeros((N, M))
  46. for idx in range(N):
  47. thetaOffset = idx * 2 * np.pi / N
  48. thetaDiff = (theta - thetaOffset) / 2
  49. rf = const * np.cos(thetaDiff) ** pint
  50. RFs[idx, :] = abs(rf)
  51. return RFs
  52. def get_W14(N):
  53. '''Receptive fields matrix'''
  54. d = int(N / 2)
  55. pint = d - 1
  56. if pint < 1:
  57. pint = 1
  58. const = np.sqrt(d / N) * np.sqrt((2 ** (2 * pint) * (math.factorial(pint)) ** 2)
  59. / (math.factorial(2 * pint) * (pint + 1)))
  60. RFs = np.zeros((N, N))
  61. theta = np.linspace(0, 2 * np.pi, N, endpoint=False)
  62. for idx in range(N):
  63. thetaOffset = idx * 2 * np.pi / N
  64. thetaDiff = (theta - thetaOffset) / 2
  65. rf = const * np.cos(thetaDiff) ** pint
  66. RFs[idx, :] = abs(rf)
  67. return RFs
  68. def setup_parameters(config, tau=1e-3, kernel=None, N=36, M=None, tauPlus=1*1e-3, **kwargs):
  69. '''This function sets up the parameters for the ring model.
  70. Params:
  71. config: configuration dictionary from YAML
  72. kernel: kernel for the within area recurrent matrix
  73. N: number of neurons
  74. M: number of stimuli
  75. kwargs: additional arguments'''
  76. if kernel is None:
  77. kernel = [0.02807382, -0.060944743, -0.073386624, 0.41472545, 0.7973934,
  78. 0.41472545, -0.073386624, -0.060944743, 0.02807382] # qmf 9 kernel
  79. # within area recurrent matrix
  80. W11 = Recurrence_matrix(N, kernel)
  81. W11 = (W11.T + W11)/2.0 # making symmetric
  82. W11 = RescaleEigenvalues(W11) # rescaling eigenvalues to one
  83. # Wn1 = Norm_matrix(N, std_dev=N/2,gaussian_height=1) # Normalization pool within each layers
  84. # Wn2 = Norm_matrix(N, std_dev=N/2,gaussian_height=1) # Normalization pool within each layers
  85. Wn1 = np.ones((N,N))
  86. Wn2 = np.ones((N,N))
  87. # Wn1 = identity_matrix
  88. # Wn2 = identity_matrix
  89. # Wn1 = (Wn1 + Wn1.T)/2.0 # making symmetric
  90. # Wn2 = (Wn2 + Wn2.T)/2.0 # making symmetric
  91. W44 = W11 # currently setting same within area recurrent matrix
  92. # W44 = identity_matrix
  93. # theta1 = np.arange(0, 2 * np.pi, 2 * np.pi / 36)
  94. # Wy1y2 = Wy1y1**2 # For sabilization by feedback plot
  95. W14 = W11 @ W11 # connectivity matrix
  96. # W14 = identity_matrix
  97. # W14 = W11
  98. W41 = W14.T # Wy2y1 is transpose of Wy1y2
  99. W45 = W14
  100. # W11 = identity_matrix # identity matrix check if the results holds
  101. # W44 = identity_matrix
  102. # Changing the weight matrices
  103. # identity_matrix = np.eye(N) * 1e-6
  104. # Wn1 = identity_matrix
  105. # Wn2 = identity_matrix
  106. # W14 = 1/np.sqrt(N) * np.ones((N, N))
  107. # W41 = W14.T
  108. # W14 = get_W14(N)
  109. # W41 = W14.T
  110. # W45 = np.ones((N,N))
  111. if M is None:
  112. theta = np.arange(0, 2 * np.pi, 2 * np.pi / 360)
  113. # theta = np.arange(0, np.pi, np.pi / 360)
  114. M = len(theta)
  115. else:
  116. theta = np.linspace(0, 2 * np.pi, M)
  117. # Encoding matrix: Receptive fields are raised cosine
  118. Wzx = ReceptiveFields(N, theta, M)
  119. # Wzx = np.ones((N, M))/np.sqrt(N)
  120. pars = {
  121. 'N': N, 'M': M,
  122. 'tauY1': tau, 'tauY4': tau,
  123. 'tauYPlus1': tauPlus, 'tauYPlus4': tauPlus,
  124. 'tauP1': tau, 'tauP4': tau,
  125. 'tauPPlus1': tauPlus, 'tauPPlus4': tauPlus,
  126. 'tauU1': tau, 'tauU4': tau,
  127. 'tauUPlus1': tauPlus, 'tauUPlus4': tauPlus,
  128. 'tauS1': tau, 'tauS4': tau,
  129. 'tauSPlus1': tauPlus, 'tauSPlus4': tauPlus,
  130. 'tauBeta1': tau, 'tauBeta4': tau,
  131. 'tauGamma1': tau, 'tauGamma4': tau,
  132. 'tau': tau, 'tauPlus': tauPlus,
  133. 'tau_x': tau, 'tau_f': config['noise_params']['tau_f'],
  134. 'sigma1': config['model_params']['sigma1'], 'sigma4': config['model_params']['sigma4'],
  135. 'dt': tauPlus/3,
  136. 'W11': W11, 'W44': W44,
  137. 'W14': W14*0.7, 'W41': W41*1.0, 'W45': W45,
  138. 'Wn1': Wn1, 'Wn4': Wn2,
  139. 'Wzx': Wzx*0.7,
  140. # Load parameters from config
  141. 'sigma_f': config['noise_params']['sigma_f'],
  142. 'alpha1': config['model_params']['alpha1'],
  143. 'alpha4': config['model_params']['alpha4'],
  144. 'beta1': config['model_params']['beta1'],
  145. 'beta4': config['model_params']['beta4'],
  146. 'gamma1': config['model_params']['gamma1'],
  147. 'gamma4': config['model_params']['gamma4'],
  148. 'b1': config['model_params']['b1'],
  149. 'b4': config['model_params']['b4'],
  150. 'g1': config['model_params']['g1'],
  151. 'g4': config['model_params']['g4']
  152. }
  153. for k in kwargs:
  154. pars[k] = kwargs[k]
  155. return pars

Create_weight_matrices.py at commit 9ee11f9, under MIT · at the source

Overview

Authors: Asit Pal1,2, Shivang Rawat2,3, David J Heeger4,5, Stefano Martiniani1,2,3,5
  1. Simons Center for Computational Physical Chemistry, Department of Chemistry, New York University New York United States
  2. Center for Soft Matter Research, Department of Physics, New York University New York United States
  3. Courant Institute of Mathematical Sciences, New York University New York United States
  4. Department of Psychology, New York University New York United States
  5. Center for Neural Science, New York University New York United States
Institutions: New York University (United States)
Journal: eLife, volume 15, article RP111297
Dates: published online 17 November 2026
Type: Research article · Language: English
License: CC BY
Identifiers: DOI 10.7554/elife.111297 · PMID 43042781 · PMCID PMC13777271 · OpenAlex W4412501798
Open access: gold, a free copy (OpenAlex)
Status: code verified (how the code is found and verified)
Categories: systems (subfield)
Methods: Spectral & time-frequency, Connectivity, Single-unit activity, calcium imaging
Keywords: recurrent neural circuits, RNN, communication subspace, power spectra, feedback, functional connectivity, None
MeSH: Models, Neurological*, Nerve Net*, Visual Cortex*, Animals, Neurons (* major topic)
Journal subjects: Neuroscience
Topic: Neural dynamics and brain function (Cognitive Neuroscience, Neuroscience), according to OpenAlex
Funding: National Science Foundation (PHY-2210452); National Institute of Health (R01MH137669, R01EY035242); Simons Foundation (839534,MT)
Citations: not cited yet (Europe PMC); 105 references in the paper

Abstract

The primate brain exhibits a hierarchical, modular architecture with conserved microcircuits executing canonical computations across reciprocally connected cortical areas. We present a hierarchical neural circuit theory with feedback connections that dynamically implements divisive normalization across its hierarchy. In a two-stage instantiation (V1 ↔ V2), increasing feedback from V2 to V1 amplifies responses in both areas. We analytically derive power spectra (V1) and coherence spectra (V1-V2) as functions of frequency (f) and compare them with experimental observations: peaks in both spectra shift to higher frequencies with increased stimulus contrast, and power decays as 1/f4 at high frequencies. The closed-form spectra are validated against direct stochastic simulation of the full nonlinear circuit across contrasts. The theory further predicts distinctive spectral signatures of feedback and input gain modulation. Crucially, the theory offers a unified view of inter-areal communication, defined as the ability to linearly predict neural activity in one brain area from neural activity in another, with emergent features consistent with empirical observations of both communication subspaces and inter-areal coherence. It admits a low-dimensional communication subspace, where inter-areal communication is lower-dimensional than within-area communication. It further predicts that: (1) increasing feedback strength enhances inter-areal communication and diminishes within-area communication, without altering the subspace dimensionality; (2) communication is strongest, and the subspace dimensionality lowest, at the frequencies at which V1-V2 coherence peaks; (3) normalization reduces the subspace dimensionality. Finally, a three-area (V1 ↔ V4 and V1 ↔ V5) instantiation of the theory demonstrates that differential feedback from higher to lower cortical areas dictates their dynamic functional connectivity. Altogether, our theory provides a robust and analytically tractable framework for generating experimentally testable predictions about normalization, inter-areal communication, and functional connectivity.

Reproduced under the paper's license (CC BY), from the paper cited above.

Repository

Its files are read in the Code ↔ Paper reader above, with 9 matches between paragraphs and lines of code.

asit-pal/hierarchical-organics

License: MIT
State: the link answers, verified on 20 November 2026
Evidence: files inventoried
Commit: 9ee11f956299e846c32bce31c40660aacca0da16, 10 November 2026
Languages: Python (46), Shell (1)
Size: 83 files, 47 scripts
Software Heritage: not archived
Found in: the references
Holds: README, license file, environment (requirements.txt), tests
Not found: CITATION.cff, continuous integration, documentation
Tools: NumPy (35 files), Matplotlib (23 files), PyTorch (7 files), SciPy (4 files), SymPy (1 file)
Availability: 1 check, the latest on 20 November 2026: the link answers
  • 20 November 2026: the link answers
49 files

The paper's code and data availability statement is in the Data section.

Tracing map

Proposed by the machine: these links were found in the paper and verified at the source, without human review. The map will receive a Zenodo DOI once one of the paper's authors has validated it with their ORCID.

What the map holds:

  • 1 repository of the authors' code, each at its verified commit, with its license and how the link was found in the paper;
  • 47 scripts, each with its path and the digest of its content;
  • 9 matches between paragraphs of the paper and lines of the code (method lexical-v1);
  • neither the text of the paper nor the code itself.

Its JSON (tracing-map.json) is deposited on Zenodo with its DOI once the map is validated.

Data

No dataset and no data link were found in the paper.

Data availability

The code to generate all the figures can be found in the following GitHub repository: https://github.com/asit-pal/Hierarchical-ORGaNICs (copy archived at Pal, 2026).

Reproduced under the paper's license (CC BY), from the paper cited above.

Versions

The history of this record: each version stored by the harvester or made by a correction of its authors or of the maintainers of its code, and what changed in its facts. The texts of the paper (its abstract, its availability statements) are not part of it; versions that changed only those are not listed.

Version 1, 20 November 2026: the first record

Recorded: type, language, journal, volume, pages, dates, 4 authors, 7 keywords, 5 MeSH terms, 3 funders, 102 references.

Cite

This paper

Pal, A., Rawat, S., Heeger, D. J., & Martiniani, S. (2026). Hierarchical neural circuit theory of normalization and inter-areal communication. eLife, 15, RP111297. https://doi.org/10.7554/elife.111297

BibTeX

@article{pal2026hierarchical,
author = {Pal, Asit and Rawat, Shivang and Heeger, David J and Martiniani, Stefano},
title = {{Hierarchical neural circuit theory of normalization and inter-areal communication}},
journal = {eLife},
year = {2026},
month = nov,
volume = {15},
pages = {RP111297},
publisher = {eLife Sciences Publications, Ltd},
issn = {2050-084X},
doi = {10.7554/elife.111297},
url = {https://doi.org/10.7554/elife.111297},
pmid = {43042781},
pmcid = {PMC13777271}
}

RIS

TY - JOUR
AU - Pal, Asit
AU - Rawat, Shivang
AU - Heeger, David J
AU - Martiniani, Stefano
TI - Hierarchical neural circuit theory of normalization and inter-areal communication
T2 - eLife
J2 - eLife
PY - 2026
DA - 2026/11/17
VL - 15
SP - RP111297
SN - 2050-084X
PB - eLife Sciences Publications, Ltd
DO - 10.7554/elife.111297
UR - https://doi.org/10.7554/elife.111297
LA - en
ER -

CSL-JSON

{
"id": "10.7554/elife.111297",
"type": "article-journal",
"title": "Hierarchical neural circuit theory of normalization and inter-areal communication",
"container-title": "eLife",
"author": [
{
"family": "Pal",
"given": "Asit"
},
{
"family": "Rawat",
"given": "Shivang"
},
{
"family": "Heeger",
"given": "David J"
},
{
"family": "Martiniani",
"given": "Stefano"
}
],
"container-title-short": "eLife",
"volume": "15",
"page": "RP111297",
"DOI": "10.7554/elife.111297",
"PMID": "43042781",
"PMCID": "PMC13777271",
"ISSN": "2050-084X",
"publisher": "eLife Sciences Publications, Ltd",
"URL": "https://doi.org/10.7554/elife.111297",
"language": "en",
"issued": {
"date-parts": [
[
2026,
11,
17
]
]
}
}

The tracing map gets a citation of its own once an author has validated it and it has a DOI.

Similar papers

The papers with a page that share the most with this one: the tools found in their code, their categories, datasets, cited references and authors, the rarest counting most.

[1] doi:10.7554/elife.99808 [code]
Untangling stability and gain modulation in cortical circuits with multiple interneuron classes.
Journal: eLife
In common: systems, 12 references
[2] doi:10.1126/sciadv.adv9396 [code]
Neuronal heterogeneity of normalization strength in a circuit model.
Journal: Science advances
In common: systems, 10 references
[3] doi:10.1016/j.neuron.2024.04.020
Distinct feedforward and feedback pathways for cell-type specific attention effects.
Journal: Neuron
In common: 11 references
[4] doi:10.1016/j.neuron.2022.01.002 [code]
Predictive coding of natural images by V1 firing rates and rhythmic synchronization.
Journal: Neuron
In common: SciPy, Matplotlib, NumPy, 9 references
[5] doi:10.1371/journal.pbio.3003831 [code]
Disinhibitory signaling enables flexible coding of top-down information in cortical networks.
Journal: PLoS biology
In common: PyTorch, SciPy, Matplotlib, 1 other tool, systems, 7 references
[6] doi:10.1371/journal.pcbi.1013958 [code]
Computing the effects of excitatory-inhibitory balance on neuronal input-output properties.
Journal: PLoS computational biology
In common: 9 references
[7] doi:10.1371/journal.pcbi.1013081 [code]
Balanced state of networks of winner-take-all units.
Journal: PLoS computational biology
In common: SciPy, Matplotlib, NumPy, 8 references
[8] doi:10.7554/elife.71801 [code]
Gated recurrence enables simple and accurate sequence prediction in stochastic, changing, and structured environments.
Journal: eLife
In common: PyTorch, SciPy, Matplotlib, 1 other tool, 8 references
[9] doi:10.7554/elife.108408 [code]
Frequency and laminar profile of feature-specific visual activity revealed by interleaved EEG-fMRI.
Journal: eLife
In common: SciPy, Matplotlib, NumPy, systems, 8 references
[10] doi:10.1371/journal.pcbi.1011704 [code]
Normalization by orientation-tuned surround in human V1-V3.
Journal: PLoS computational biology
In common: systems, 8 references

Contribute

The authors of this paper can claim it, correct its record and validate its tracing map, and the maintainers of its code (its owner, or a public member of its organization) correct what it says of their repository; anyone signed in can ask for its removal. Every request goes to OSCR's own machine, which answers it; your account page follows them.

Sign in with ORCID to claim this paper as one of its authors, correct its record or validate its tracing map: when the paper's metadata lists your ORCID iD, you are recognized at once. Maintainers of its code: sign in with GitHub, then claim the repository on your account page.

Add the badge to the code's README

OSCR: the badge of this paper

If you are an author or a maintainer of this paper's code, you can add this badge. It says what OSCR knows of the paper (its code, its validated tracing map) and links the code to this page. Paste one of these snippets under the README's title. Keep the two marks around it, so that the badge can be replaced in place and is never added twice. OSCR changes nothing in your repository.

Markdown
<!-- oscr-badge:start -->
[![OSCR](https://openscicode.org/badge/doi/10.7554/elife.111297.svg)](https://openscicode.org/badge/doi/10.7554/elife.111297)
<!-- oscr-badge:end -->
reStructuredText
.. oscr-badge:start

.. image:: https://openscicode.org/badge/doi/10.7554/elife.111297.svg
   :target: https://openscicode.org/badge/doi/10.7554/elife.111297
   :alt: OSCR

.. oscr-badge:end
HTML
<!-- oscr-badge:start -->
<a href="https://openscicode.org/badge/doi/10.7554/elife.111297"><img src="https://openscicode.org/badge/doi/10.7554/elife.111297.svg" alt="OSCR"></a>
<!-- oscr-badge:end -->

If you choose to, you can have the badge kept up to date for you, with oscr badge --write from the command line or with the GitHub Action in your repository's workflows. Every other form and option (a card, AsciiDoc, LaTeX, shields.io) is on the badge page.

Request its removal

To ask OSCR to remove this record, the copies of its authors' scripts or its tracing map, use the removal request page: signed in, you say who you are, what to remove and why, then review and confirm the request. Published rules decide every request (how).

Discussion, reproductions, activity

Discussion: code errors and mismatches between the code and the paper about this paper and its code, from readers and its authors, beside the paper and its tracing map.

Report a mismatch between the code and the paper · Report a code error (signed in)

Reproductions: reports from readers who ran the authors' code and did not get the paper's result, what they ran, with which environment, commit and data.

Report a reproduction failure (signed in)

Versions of its code: the releases of its code tied to a version of this paper (the preprint, the accepted manuscript, the version of record…), each with the tracing map versioned with it.

Activity: what happens around this paper: new versions of its record, its map's validation, discussions and reproductions. It opens with sign-in.