pith. sign in

arxiv: 2506.11272 · v2 · submitted 2025-06-12 · ⚛️ physics.bio-ph · cond-mat.soft· nlin.AO· q-bio.CB· q-bio.QM

Maximum-Entropy Model of Colored Noise in Superdiffusive Axonal Growth

Julian Sutaria , Cristian Staii This is my paper

Pith reviewed 2026-05-19 09:37 UTC · model grok-4.3

classification ⚛️ physics.bio-ph cond-mat.softnlin.AOq-bio.CBq-bio.QM
keywords axonal growthmaximum entropycolored noisesuperdiffusiongrowth conetraction forcesLangevin dynamicsvelocity autocorrelation
0
0 comments X

The pith

Maximum-entropy inference of traction force relaxation rates produces colored noise with velocity correlation exponent -1/2 in axonal growth.

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

The paper starts from a standard Langevin description of growth cone motion on patterned surfaces and uses the maximum entropy principle to determine the distribution of traction force relaxation rates. The constraints come from experimental features of the dynamics rather than from an assumed form for the noise. The resulting distribution produces a colored acceleration process whose correlations follow a power law. This leads to analytical expressions for the mean squared displacement of the axon tip and for the velocity autocorrelation function. For parameter values typical of cortical neurons the model gives a correlation exponent of -1/2, which is close to the measured value near -0.6, and it reproduces the observed crossover from short-time diffusion to long-time superdiffusion with mean squared displacement exponent 1.4.

Core claim

Applying the Shannon-Jaynes maximum entropy principle to experimentally motivated constraints on a Langevin model of growth cone motion yields an effective distribution of traction force relaxation rates. The slow-relaxation tail of this distribution generates a stationary colored acceleration noise with temporal correlations decaying as a power law with exponent α = -1/2. The same distribution produces superdiffusive axonal mean squared displacement that scales as t^1.4 at long times, a crossover from early diffusive behavior, and a power-law decay of the velocity autocorrelation, all in quantitative agreement with measurements on cortical neurons grown on micropatterned substrates.

What carries the argument

The maximum-entropy distribution of traction force relaxation rates inferred from constraints on the Langevin growth-cone dynamics, which sets the spectrum of the resulting colored acceleration noise.

If this is right

  • The model accounts for the observed crossover from short-time diffusive motion to long-time superdiffusive growth.
  • Velocity autocorrelation functions decay as a power law whose exponent matches the displacement scaling.
  • The slow part of the relaxation-rate distribution corresponds to broadly distributed adhesion engagement times.
  • Microscopic force-generation statistics are connected directly to macroscopic axonal growth laws.

Where Pith is reading between the lines

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

  • The same maximum-entropy construction could be applied to other active-matter systems whose fluctuations arise from broadly distributed relaxation processes.
  • Measurements of adhesion lifetimes on different substrates would provide an independent test of the inferred distribution.
  • Spatial variations in the substrate pattern could be incorporated by making the constraints position-dependent.

Load-bearing premise

The distribution of traction force relaxation rates can be obtained solely by maximizing entropy subject to the chosen constraints on the Langevin dynamics, without direct experimental measurement of the rates themselves.

What would settle it

A direct experimental histogram of clutch or adhesion engagement times in growth cones; significant deviation from the power-law tail required by the maximum-entropy solution would falsify the inference step.

Figures

Figures reproduced from arXiv: 2506.11272 by Cristian Staii, Julian Sutaria.

Figure 1
Figure 1. Figure 1: FIG. 1. (a) Schematic representation of a micropatterned PDMS surface. Periodic surface patterns are shown along the [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: FIG. 2. Time evolution of the mean-squared displacement [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: FIG. 3. Time dependence of the velocity–autocorrelation [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
read the original abstract

We develop a coarse-grained stochastic theory for axonal growth on micropatterned substrates using the Shannon--Jaynes maximum entropy principle. Starting from a Langevin description of growth cone motion, we infer the effective distribution of traction force relaxation rates from experimentally motivated constraints rather than postulating the colored noise directly. The resulting relaxation rate distribution generates a stationary colored acceleration process with power-law temporal correlations and yields analytical predictions for the axonal mean squared displacement and velocity autocorrelation. The long-time behavior is controlled by the slow-relaxation part of the inferred distribution, corresponding physically to broadly distributed clutch or adhesion engagement times. For biologically relevant parameters, the model predicts a negative correlation exponent $\alpha=-1/2$. This prediction is in close quantitative agreement with measurements on cortical neurons cultured on micropatterned poly-D-lysine-coated PDMS substrates, which are well described by $\alpha\simeq -0.6$ and exhibit superdiffusive mean squared displacement scaling with exponent $1.4$. The same framework accounts for the crossover from early diffusive behavior to long-time anomalous growth and for the corresponding power law decay of the velocity autocorrelation. These results show how entropy-constrained active fluctuations can connect microscopic force generation processes to emergent growth laws in neuronal systems and, more broadly, in active matter.

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 develops a coarse-grained stochastic theory for axonal growth on micropatterned substrates using the Shannon-Jaynes maximum entropy principle. Starting from a Langevin description of growth cone motion, the effective distribution of traction force relaxation rates is inferred from experimentally motivated constraints rather than postulated directly. This generates a stationary colored acceleration process with power-law temporal correlations, yielding analytical predictions for axonal mean squared displacement and velocity autocorrelation. For biologically relevant parameters, the model predicts a negative correlation exponent α=-1/2 in close agreement with cortical neuron measurements (α≃-0.6) that show superdiffusive MSD scaling with exponent 1.4. The framework also accounts for the crossover from early diffusive to long-time anomalous growth.

Significance. If the derivations are rigorous, the work provides a principled, non-ad hoc route to derive effective colored noise in active biological systems by applying maximum entropy to Langevin dynamics. The quantitative match to experimental exponents for both velocity correlations and MSD, plus the explanation of the diffusive-to-anomalous crossover, are strengths. This could influence modeling of anomalous transport in neuronal development and broader active-matter contexts, particularly if the constraints are shown to be independently justified.

major comments (2)
  1. [Abstract and maximum-entropy inference section] Abstract and the section deriving the relaxation-rate distribution: the claim that the max-ent procedure infers P(γ) whose slow-relaxation tail produces exactly α=-1/2 relies on the specific constraints chosen. Standard max-ent with fixed mean or variance yields an exponential P(γ), whose integrated noise gives exponential (not power-law) correlations and ordinary diffusion. The manuscript must explicitly list and justify the constraints (e.g., which moments or biologically motivated quantities are fixed) to demonstrate that the 1/γ or heavier tail emerges naturally rather than being selected to match the target phenomenology.
  2. [Parameter selection and long-time asymptotics] Section on parameter selection and long-time asymptotics: the prediction α=-1/2 is stated to hold 'for biologically relevant parameters,' yet these parameters are not defined independently of the experimental data being explained. Without a clear, a priori range or derivation of these parameters, the quantitative agreement with α≃-0.6 risks circularity, undermining the claim that the exponent is a robust output of the entropy-constrained model.
minor comments (2)
  1. [Notation throughout] Clarify notation for the correlation exponent α versus the MSD scaling exponent to avoid any ambiguity in the text and figures.
  2. [Introduction or discussion] Add a brief comparison table or paragraph contrasting the max-ent approach with direct postulation of colored noise in prior axonal-growth models.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for their constructive and insightful comments, which have helped us identify areas where the presentation of our maximum-entropy derivation can be clarified. We address each major comment below and indicate the revisions we will make to strengthen the manuscript.

read point-by-point responses

  1. Referee: [Abstract and maximum-entropy inference section] Abstract and the section deriving the relaxation-rate distribution: the claim that the max-ent procedure infers P(γ) whose slow-relaxation tail produces exactly α=-1/2 relies on the specific constraints chosen. Standard max-ent with fixed mean or variance yields an exponential P(γ), whose integrated noise gives exponential (not power-law) correlations and ordinary diffusion. The manuscript must explicitly list and justify the constraints (e.g., which moments or biologically motivated quantities are fixed) to demonstrate that the 1/γ or heavier tail emerges naturally rather than being selected to match the target phenomenology.

    Authors: We agree that the specific constraints are essential to the emergence of the power-law tail and that this must be made fully explicit. In the current manuscript the constraints consist of the mean traction force magnitude, the mean logarithmic relaxation rate (reflecting the geometric mean of adhesion lifetimes), and a second-moment constraint on force fluctuations; these are motivated by independent experimental reports on growth-cone traction forces and clutch-engagement statistics. Subject to these constraints the maximum-entropy procedure analytically yields P(γ) ∝ 1/γ for small γ. We will revise the maximum-entropy inference section to list every constraint with its experimental justification, reproduce the Lagrange-multiplier derivation in full, and demonstrate that the 1/γ tail is a direct consequence of the chosen constraints rather than an input chosen to reproduce the target exponent. revision: yes

  2. Referee: [Parameter selection and long-time asymptotics] Section on parameter selection and long-time asymptotics: the prediction α=-1/2 is stated to hold 'for biologically relevant parameters,' yet these parameters are not defined independently of the experimental data being explained. Without a clear, a priori range or derivation of these parameters, the quantitative agreement with α≃-0.6 risks circularity, undermining the claim that the exponent is a robust output of the entropy-constrained model.

    Authors: We acknowledge that the current wording leaves open the possibility of circularity. The parameters entering the long-time asymptotics (the lower cutoff γ_min of the relaxation-rate distribution and the overall noise amplitude) are taken from independent literature values for cortical-neuron adhesion lifetimes and force magnitudes measured on similar micropatterned substrates; they are not fitted to the velocity-autocorrelation data reported in the paper. The exponent α = −1/2 follows analytically from the 1/γ tail once γ_min is small enough that the integral for the velocity autocorrelation is dominated by the slow-relaxation regime; this holds throughout the biologically plausible interval 10^{-3} s^{-1} < γ_min < 10^{-1} s^{-1}. We will add an explicit subsection that tabulates the independent experimental sources for each parameter, derives the analytic condition under which α = −1/2 is obtained, and includes a brief robustness check showing that the exponent remains −1/2 across the cited range. revision: yes

Circularity Check

1 steps flagged

Parameter selection for α=-1/2 relies on 'biologically relevant' choices tuned to observed phenomenology

specific steps
  1. fitted input called prediction [Abstract]
    "For biologically relevant parameters, the model predicts a negative correlation exponent α=-1/2. This prediction is in close quantitative agreement with measurements on cortical neurons cultured on micropatterned poly-D-lysine-coated PDMS substrates, which are well described by α≃-0.6 and exhibit superdiffusive mean squared displacement scaling with exponent 1.4."

    The long-time α=-1/2 is controlled by the slow-γ tail of the inferred P(γ). Selecting the 'biologically relevant parameters' that enforce this tail and then presenting the resulting exponent as a prediction that agrees with the same class of measurements reduces the claim to a fit of the target phenomenology rather than an independent derivation.

full rationale

The derivation applies max-ent to infer P(γ) from constraints on the Langevin equation and obtains analytical expressions for MSD and velocity correlations. The long-time exponent α=-1/2 emerges specifically from the slow-relaxation tail of that distribution. While the constraints are described as experimentally motivated, the paper selects 'biologically relevant parameters' to produce exactly this tail and reports close agreement with the same cortical-neuron data (α≃-0.6, MSD exponent 1.4). This makes the headline prediction dependent on parameter choices that encode the target phenomenology rather than being an independent output of the max-ent procedure alone. No self-citation load-bearing or explicit self-definition is present; the circularity is limited to the final parameter step.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 1 invented entities

The central claim rests on the maximum entropy principle applied to a Langevin starting point plus selection of biologically relevant parameters that pick out the slow-relaxation tail.

free parameters (1)
  • biologically relevant parameters
    Used to isolate the slow-relaxation part of the inferred distribution that sets the long-time α=-1/2 behavior
axioms (2)
  • domain assumption Langevin description of growth cone motion
    Starting point from which the effective distribution is inferred
  • domain assumption Shannon-Jaynes maximum entropy principle
    Used to obtain the distribution of relaxation rates from constraints
invented entities (1)
  • distribution of traction force relaxation rates no independent evidence
    purpose: Generates the stationary colored acceleration process with power-law correlations
    Inferred via max entropy rather than measured; no independent falsifiable handle given in abstract

pith-pipeline@v0.9.0 · 5773 in / 1482 out tokens · 39350 ms · 2026-05-19T09:37:34.338375+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

34 extracted references · 34 canonical work pages

  1. [1]

    Franze and J

    K. Franze and J. Guck, The biophysics of neuronal growth, Reports on Progress in Physics 73, 094601 (2010)

    work page 2010

  2. [2]

    L. A. Lowery and D. Van Vactor, The trip of the tip: un- derstanding the growth cone machinery, Nature Reviews Molecular Cell Biology 10, 332 (2009)

    work page 2009

  3. [3]

    A. B. Huber, A. L. Kolodkin, D. D. Ginty, and J. F. Cloutier, Signaling at the growth cone: ligand-receptor complexes and the control of axon growth and guidance, Annual Review of Neuroscience 26, 509 (2003)

    work page 2003

  4. [4]

    de Gennes, Collective neuronal growth and self organization of axons, Proceedings of the National Academy of Sciences 104, 4904 (2007)

    P.-G. de Gennes, Collective neuronal growth and self organization of axons, Proceedings of the National Academy of Sciences 104, 4904 (2007)

    work page 2007

  5. [5]

    Peruani and L

    F. Peruani and L. G. Morelli, Self-propelled particles with fluctuating speed and direction of motion in two dimen- sions, Phys. Rev. Lett. 99, 010602 (2007)

    work page 2007

  6. [6]

    P. F. C. Tilles, S. V. Petrovskii, and P. L. Natti, A ran- 6 dom acceleration model of individual animal movement allowing for diffusive, superdiffusive and superballistic regimes, Scientific Reports 7, 14364 (2017)

    work page 2017

  7. [7]

    Oliveri and A

    H. Oliveri and A. Goriely, Mathematical models of neu- ronal growth, Biomechanics and Modeling in Mechanobi- ology 21, 89 (2022)

    work page 2022

  8. [8]

    Alert and X

    R. Alert and X. Trepat, Physical models of collective cell migration, Annual Review of Condensed Matter Physics 11, 10.1146/annurev-conmatphys-031218-013516 (2020)

    work page doi:10.1146/annurev-conmatphys-031218-013516 2020

  9. [9]

    Tkaˇ cik, O

    G. Tkaˇ cik, O. Marre, T. Mora, D. Amodei, M. J. Berry II, and W. Bialek, The simplest maximum entropy model for collective behavior in a neural network, Journal of Statis- tical Mechanics: Theory and Experiment 2013, P03011 (2013)

    work page 2013

  10. [10]

    Amselem, M

    G. Amselem, M. Theves, A. Bae, E. Bodenschatz, and C. Beta, A stochastic description of dictyostelium chemo- taxis, PLoS One 7, e37213 (2012)

    work page 2012

  11. [11]

    Y. Song, D. Zhou, and S. Li, Maximum entropy principle underlies wiring length distribution in brain networks, Cerebral Cortex 31, 4628 (2021)

    work page 2021

  12. [12]

    Lamberti, M

    M. Lamberti, M. Hess, I. Dias, M. van Putten, J. le Feber, and S. Marzen, Maximum entropy models provide func- tional connectivity estimates in neural networks, Scien- tific Reports 12, 9656 (2022)

    work page 2022

  13. [13]

    Franze, Integrating chemistry and mechanics: The forces driving axon growth, Annual Review of Cell and Developmental Biology 36, 61 (2020)

    K. Franze, Integrating chemistry and mechanics: The forces driving axon growth, Annual Review of Cell and Developmental Biology 36, 61 (2020)

    work page 2020

  14. [14]

    Oliveri, K

    H. Oliveri, K. Franze, and A. Goriely, Theory for durotac- tic axon guidance, Phys. Rev. Lett. 126, 118101 (2021)

    work page 2021

  15. [15]

    de Rooij, E

    R. de Rooij, E. Kuhl, and K. E. Miller, Modeling the axon as an active partner with the growth cone in axonal elongation, Biophysical Journal 115, 1783 (2018)

    work page 2018

  16. [16]

    Kundu, L

    A. Kundu, L. Micholt, S. Friedrich, D. R. Rand, C. Bar- tic, D. Braeken, and A. Levchenko, Superimposed topo- graphic and chemical cues synergistically guide neurite outgrowth, Lab on a Chip 13, 3070 (2013)

    work page 2013

  17. [17]

    Ishihara, N

    M. Ishihara, N. Mochizuki-Oda, K. Iwatsuki, H. Kishima, Y. Iwamoto, Y. Ohnishi, M. Umegaki, and T. Yoshimine, A new three-dimensional axonal outgrowth assay for central nervous system regeneration, Journal of Neuro- science Methods 198, 181 (2011)

    work page 2011

  18. [18]

    Kumarasinghe, L

    U. Kumarasinghe, L. N. Fox, and C. Staii, Combined traction force–atomic force microscopy measurements of neuronal cells, Biomimetics 7, 10.3390/biomimet- ics7040157 (2022)

    work page doi:10.3390/biomimet- 2022

  19. [19]

    J. M. Vensi Basso, I. Yurchenko, M. Simon, D. J. Rizzo, and C. Staii, Role of geometrical cues in neuronal growth, Physical Review E 99, 022408 (2019)

    work page 2019

  20. [20]

    Descoteaux, J

    M. Descoteaux, J. P. Sunnerberg, D. D. Brady, and C. Staii, Feedback-controlled dynamics of neuronal cells on directional surfaces, Biophysical Journal 121, 769 (2022)

    work page 2022

  21. [21]

    Yurchenko, J

    I. Yurchenko, J. M. Vensi Basso, V. S. Syrotenko, and C. Staii, Anomalous diffusion for neuronal growth on sur- faces with controlled geometries, PLoS One 14, e0216181 (2019)

    work page 2019

  22. [22]

    J. M. V. Basso, I. Yurchenko, M. R. Wiens, and C. Staii, Neuron dynamics on directional surfaces, Soft Matter15, 9931 (2019)

    work page 2019

  23. [23]

    [18-22], for experimental details, solution to the Maximum Entropy problem, and derivations for mean- square displacement and velocity autocorrelation func- tions

    See Supplemental Material at [URL], which includes Refs. [18-22], for experimental details, solution to the Maximum Entropy problem, and derivations for mean- square displacement and velocity autocorrelation func- tions

    work page

  24. [24]

    Swanson and N

    D. Swanson and N. S. Wingreen, Active biopolymers con- fer fast reorganization kinetics, Phys. Rev. Lett. 107, 218103 (2011)

    work page 2011

  25. [25]

    Franze, J

    K. Franze, J. Gerdelmann, M. Weick, T. Betz, S. Pawl- izak, M. Lakadamyali, J. Bayer, K. Rillich, M. G¨ ogler, Y. B. Lu, A. Reichenbach, P. Janmey, and J. K¨ as, Neu- rite branch retraction is caused by a threshold-dependent mechanical impact, Biophysical Journal 97, 1883 (2009)

    work page 2009

  26. [26]

    D. Koch, W. J. Rosoff, J. Jiang, H. M. Geller, and J. S. Urbach, Strength in the periphery: growth cone biome- chanics and substrate rigidity response in peripheral and central nervous system neurons, Biophysical Journal102, 452 (2012)

    work page 2012

  27. [27]

    Hyland, A

    C. Hyland, A. F. Mertz, P. Forscher, and E. Dufresne, Dynamic peripheral traction forces balance stable neurite tension in regenerating aplysia bag cell neurons, Scientific Reports 4, 4961 (2014)

    work page 2014

  28. [28]

    E. T. Jaynes, Information theory and statistical mechan- ics, Phys. Rev. 106, 620 (1957)

    work page 1957

  29. [29]

    Press´ e, K

    S. Press´ e, K. Ghosh, J. Lee, and K. A. Dill, Principles of maximum entropy and maximum caliber in statistical physics, Rev. Mod. Phys. 85, 1115 (2013)

    work page 2013

  30. [30]

    Zhang, J

    X.-Y. Zhang, J. M. Moore, X. Ru, and G. Yan, Geometric scaling law in real neuronal networks, Phys. Rev. Lett. 133, 138401 (2024)

    work page 2024

  31. [31]

    A. M. Reynolds, Truncated l´ evy walks are expected be- yond the scale of data collection when correlated random walks embody observed movement patterns, J. R. Soc. Interface 9, 528 (2012)

    work page 2012

  32. [32]

    A. M. Edwards, R. A. Phillips, N. W. Watkins, M. P. Freeman, E. J. Murphy, V. Afanasyev, S. V. Buldyrev, M. G. E. da Luz, E. P. Raposo, H. E. Stanley, and G. M. Viswanathan, Revisiting l´ evy flight search patterns of wandering albatrosses, bumblebees and deer, Nature 449, 1044 (2007)

    work page 2007

  33. [33]

    Dieterich, R

    P. Dieterich, R. Klages, R. Preuss, and A. Schwab, Anomalous dynamics of cell migration, Proc. Natl. Acad. Sci. U.S.A. 105, 459 (2008)

    work page 2008

  34. [34]

    Metzler and J

    R. Metzler and J. Klafter, The random walk’s guide to anomalous diffusion: A fractional dynamics approach, Phys. Rep. 339, 1 (2000). 7 Supplemental Material for Entropy-Constrained Noise Yields Superdiffusive Dynamics in Axonal Growth This PDF file includes: Experimental Details Solution to the variational problem Analysis of the power law dependence for ...

    work page 2000