pith. sign in

arxiv: 2606.21432 · v1 · pith:YKVJ4R4Wnew · submitted 2026-06-19 · 💻 cs.NE · q-bio.NC

Soliton-like Waves in a Two-Dimensional Recurrent Spiking Neural Network with Weighted Spike-Timing-Dependent Plasticity

Ch. Meessen This is my paper

Pith reviewed 2026-06-26 12:46 UTC · model grok-4.3

classification 💻 cs.NE q-bio.NC
keywords spiking neural networksweighted STDPdissipative solitonstraveling wavesrecurrent networksexcitatory-inhibitory asymmetrysynaptic plasticitycortical waves
0
0 comments X

The pith

A two-dimensional spiking neural network with weighted STDP spontaneously generates stable dissipative soliton waves from local rules.

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

The paper builds a minimal spiking neuron model that combines weighted spike-timing-dependent plasticity, divisive normalization, homeostatic threshold adaptation, and a one-step refractory period. When excitatory-inhibitory neuron pairs form a recurrent two-dimensional network under periodic localized stimulation, stable self-propagating wave packets emerge. These packets keep a fixed spatial profile, move at constant speed, and annihilate on head-on collision. Emergence depends on asymmetric excitatory and inhibitory connection radii plus initially stronger inhibitory synapses. Weighted STDP imprints propagation direction into the weights, allowing the network to sustain one-way travel and to form phase-encoding boundaries between waves from multiple sources.

Core claim

Assembling excitatory-inhibitory pairs of such neurons into a two-dimensional recurrent network and applying periodic localized stimulation, the network spontaneously gives rise to stable, self-propagating wave packets with the properties of dissipative solitons: they maintain a stable spatial profile, propagate at constant speed, and annihilate upon frontal collision. Their emergence requires a geometric asymmetry between excitatory and inhibitory connection radii, and initial inhibitory synapses stronger than excitatory ones. WSTDP engraves the direction of propagation into the synaptic weight profile, so that the network learns by itself to sustain propagation in one direction while suppr

What carries the argument

Two-dimensional recurrent network of excitatory-inhibitory neuron pairs with weighted spike-timing-dependent plasticity and asymmetric connection radii, which together produce spontaneous soliton-like waves and directional learning.

If this is right

  • The waves maintain a stable spatial profile and travel at constant speed.
  • Frontal collisions cause the waves to annihilate.
  • WSTDP imprints unidirectional propagation into the synaptic weights.
  • Simultaneous activity from two sources produces a semi-persistent boundary whose location encodes relative phase and frequency.
  • The construction supplies a minimal framework for cortical traveling waves, activity-zone boundaries, and spatial memory arising from local plasticity rules.

Where Pith is reading between the lines

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

  • Similar local rules might allow biological cortex to generate and maintain directional traveling waves without external guidance or global signals.
  • The collision-created boundary could function as a computational primitive for encoding spatial relations or timing differences in sensory processing.
  • Adding realistic noise or scaling the network to larger sizes would test whether the soliton behavior persists beyond the minimal setting.
  • The same asymmetry-plus-plasticity combination might be transplanted into other recurrent architectures to produce stable propagating patterns for computational tasks.

Load-bearing premise

The emergence of these waves depends on a geometric asymmetry between the radii of excitatory and inhibitory connections together with initially stronger inhibitory synapses.

What would settle it

Simulating the network after removing the connection-radius asymmetry or setting initial excitatory and inhibitory synapse strengths equal, and finding no stable propagating waves, would falsify the necessity of those conditions.

Figures

Figures reproduced from arXiv: 2606.21432 by Ch. Meessen.

Figure 1
Figure 1. Figure 1: Snapshots of network activity during a single-source simulation (200 [PITH_FULL_IMAGE:figures/full_fig_p007_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Evolution of synaptic weights over 10,000 time steps for neuron (150 [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Converged synaptic weight profile along y = 50 for neuron (150, 50), after t = 10,000 time steps. The strong left-right asymmetry reflects the causal structure of WSTDP: synapses in the direction of soliton propagation (∆x > 0) undergo overall LTP, while those in the opposite direction undergo overall LTD. ∆x = 0 corresponds to the autapse (E→E self-connection). Zero weights indicate absent connections and… view at source ↗
Figure 4
Figure 4. Figure 4: Adaptive firing threshold profile along y = 50 after t = 10,000 time steps. Most neurons converge near the expected steady-state value θmean = 0.275 (dotted line). Neurons in the stimu￾lation zone (x ≈ 91–109) show thresholds at θmin = 0.05, reflecting sustained direct activation. 9 [PITH_FULL_IMAGE:figures/full_fig_p009_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Collision experiment: same frequency, same phase. [PITH_FULL_IMAGE:figures/full_fig_p010_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Collision experiment: same frequency, phase offset of 3 steps ( [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Network activity at frame 98 when the initial inhibitory weight is reduced to [PITH_FULL_IMAGE:figures/full_fig_p011_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Kymographs of excitatory neuron activity along [PITH_FULL_IMAGE:figures/full_fig_p013_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Effect of the inhibitory connection radii [PITH_FULL_IMAGE:figures/full_fig_p014_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Kymographs of single-source stimulation for four combinations of [PITH_FULL_IMAGE:figures/full_fig_p016_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Kymographs of dual-source stimulation with stim [PITH_FULL_IMAGE:figures/full_fig_p017_11.png] view at source ↗
read the original abstract

We construct a minimal but biologically plausible spiking neuron model operating in discrete time, combining multiplicative spike-timing-dependent plasticity (WSTDP), divisive normalization of synaptic integration, homeostatic threshold adaptation, and a one-step refractory period. We show that this normalization admits a biologically plausible dendritic implementation in which each binary junction operates using only locally available information. Assembling excitatory-inhibitory pairs of such neurons into a two-dimensional recurrent network and applying periodic localized stimulation, we find that the network spontaneously gives rise to stable, self-propagating wave packets with the properties of dissipative solitons: they maintain a stable spatial profile, propagate at constant speed, and annihilate upon frontal collision. Their emergence requires a geometric asymmetry between excitatory and inhibitory connection radii, and initial inhibitory synapses stronger than excitatory ones. WSTDP engraves the direction of propagation into the synaptic weight profile, so that the network learns by itself to sustain propagation in one direction while suppressing the reverse. When two sources are active simultaneously, the resulting waves annihilate upon collision, defining a semi-persistent boundary whose position encodes the relative phase and frequency of the two sources. These results provide a minimal computational framework for studying the emergence of cortical traveling waves, activity zone delimitation, and spatial memory from local plasticity rules alone.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

0 major / 3 minor

Summary. The manuscript constructs a minimal discrete-time spiking neuron model combining weighted spike-timing-dependent plasticity (WSTDP), divisive normalization of synaptic integration (with a proposed local dendritic implementation), homeostatic threshold adaptation, and a one-step refractory period. Assembling excitatory-inhibitory neuron pairs into a 2D recurrent network with asymmetric excitatory versus inhibitory connection radii and stronger initial inhibitory synaptic weights, the authors apply periodic localized stimulation and report the spontaneous emergence of stable, self-propagating wave packets. These packets exhibit dissipative-soliton properties (stable spatial profile, constant propagation speed, frontal annihilation on collision), with WSTDP engraving propagation direction and multi-source collisions producing semi-persistent boundaries that encode relative phase and frequency.

Significance. If the reported numerical observations are robust, the work supplies a minimal, fully specified computational framework in which local, biologically motivated plasticity rules alone suffice to produce macroscopic traveling waves, direction learning, and activity-zone delimitation. The explicit disclosure of the two required asymmetries (connection radii and initial weight imbalance) and the dendritic normalization proposal are constructive features that facilitate reproducibility and biological interpretation.

minor comments (3)
  1. The manuscript should include a dedicated methods subsection (or appendix) tabulating all fixed parameter values, network size, stimulation protocol details, and the precise numerical criteria used to classify a wave packet as 'stable' and 'self-propagating' (e.g., profile correlation threshold, speed variance bound).
  2. Figure captions and main text should explicitly state the number of independent runs, random seeds, and any statistical tests confirming that the soliton properties persist across initial conditions once the required asymmetries are imposed.
  3. A brief comparison paragraph placing the observed wave speed and annihilation behavior against existing continuum or mean-field models of cortical traveling waves would help readers assess novelty.

Simulated Author's Rebuttal

0 responses · 0 unresolved

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

Circularity Check

0 steps flagged

No significant circularity detected

full rationale

The paper constructs a discrete-time spiking model from explicit local rules (multiplicative WSTDP, divisive normalization, homeostatic threshold, refractory period) and reports direct numerical outcomes under stated initial conditions (E-I radius asymmetry and stronger initial inhibitory weights). No equations are presented that derive a target quantity from a fitted parameter whose value is itself taken from the target; no self-citation chain is invoked to justify uniqueness or an ansatz; and the soliton-like properties are described as observed simulation results rather than predictions forced by construction. The derivation chain is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The abstract states that the normalization admits a biologically plausible dendritic implementation using only locally available information; this is treated as a domain assumption rather than a derived result. The requirement for asymmetric connection radii and stronger initial inhibitory weights is presented as a necessary condition for wave emergence and functions as an ad-hoc modeling choice. No free parameters are numerically specified and no new physical entities are postulated.

free parameters (2)
  • excitatory versus inhibitory connection radii
    Geometric asymmetry between the two radii is required for soliton emergence; the specific ratio is not given in the abstract.
  • initial inhibitory versus excitatory synaptic strengths
    Initial inhibitory synapses must be stronger than excitatory ones; the exact ratio is not reported.
axioms (1)
  • domain assumption The chosen normalization admits a biologically plausible dendritic implementation in which each binary junction operates using only locally available information.
    Stated directly in the abstract as a property of the model.

pith-pipeline@v0.9.1-grok · 5761 in / 1529 out tokens · 25743 ms · 2026-06-26T12:46:02.970673+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

24 extracted references

  1. [1]

    , title =

    Amari, S. , title =. Biological Cybernetics , year =

  2. [2]

    Bennett, J. E. M. and Bair, W. , title =. PLOS Computational Biology , year =

  3. [3]

    Bi, G. Q. and Poo, M. M. , title =. Journal of Neuroscience , year =

  4. [4]

    Bressloff, P. C. , title =. Journal of Physics A: Mathematical and Theoretical , year =

  5. [5]

    Ermentrout, G. B. and Kleinfeld, D. , title =. Neuron , year =

  6. [6]

    Markram, H. and L\". Regulation of synaptic efficacy by coincidence of postsynaptic. Science , year =

  7. [7]

    and Thorpe, S

    Masquelier, T. and Thorpe, S. J. , title =. PLOS Computational Biology , year =

  8. [8]

    and Diesmann, M

    Morrison, A. and Diesmann, M. and Gerstner, W. , title =. Biological Cybernetics , year =

  9. [9]

    and Chavane, F

    Muller, L. and Chavane, F. and Reynolds, J. and Sejnowski, T. J. , title =. Nature Reviews Neuroscience , year =

  10. [10]

    van Rossum, M. C. W. and Bi, G. Q. and Turrigiano, G. G. , title =. Journal of Neuroscience , year =

  11. [11]

    and Miller, K

    Song, S. and Miller, K. D. and Abbott, L. F. , title =. Nature Neuroscience , year =

  12. [12]

    and Cruz, L

    Butler, K. and Cruz, L. , title =. Journal of Computational Neuroscience , year =

  13. [13]

    Davis, Z. W. and Muller, L. and Martinez-Trujillo, J. and Sejnowski, T. and Reynolds, J. H. , title =. Nature , year =

  14. [14]

    and Inglebert, Y

    Debanne, D. and Inglebert, Y. , title =. Current Opinion in Neurobiology , year =

  15. [15]

    and Garrido, J

    Debanne, D. and Garrido, J. J. , title =. Encyclopedia of Life Sciences , publisher =. 2018 , doi =

    2018

  16. [16]

    and Heeger, D

    Carandini, M. and Heeger, D. J. , title =. Nature Reviews Neuroscience , year =

  17. [17]

    Kole, M. H. P. and Stuart, G. J. , title =. Neuron , year =

  18. [18]

    Trachtenberg, J. T. and Chen, B. E. and Knott, G. W. and Feng, G. and Sanes, J. R. and Welker, E. and Svoboda, K. , title =. Nature , year =

  19. [19]

    Douglas, R. J. and Martin, K. A. C. , title =. Annual Review of Neuroscience , year =

  20. [20]

    Laughlin, S. B. and Sejnowski, T. J. , title =. Science , year =

  21. [21]

    and Klausberger, T

    Somogyi, P. and Klausberger, T. , title =. Journal of Physiology , year =

  22. [22]

    and Trappenberg, T

    Standage, D. and Trappenberg, T. , title =. Proceedings of the 2007 International Joint Conference on Neural Networks (IJCNN) , pages =. 2007 , doi =

    2007

  23. [23]

    Abraham, W. C. , title =. Nature Reviews Neuroscience , year =

  24. [24]

    Multimicroelectrode investigation of monkey striate cortex: spike train correlations in the infragranular layers , journal =

    Kr\". Multimicroelectrode investigation of monkey striate cortex: spike train correlations in the infragranular layers , journal =. 1988 , volume =

    1988