arxiv: 2605.07475 · v1 · submitted 2026-05-08 · 💻 cs.NE · cs.ET· eess.SP

Recognition: no theorem link

Broken-symmetry shape discrimination on a driven Duffing ring

Kaspar Anton Schindler

Authors on Pith no claims yet

Pith reviewed 2026-05-11 02:05 UTC · model grok-4.3

classification 💻 cs.NE cs.ETeess.SP

keywords Duffing oscillatorshape discriminationbroken symmetrycycle graphdistributed computationbinding operationnonlinear dynamicsmode mixing

0 comments

The pith

A driven Duffing ring on a cycle graph breaks time-reversal symmetry to let one observable discriminate input shapes.

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

The paper examines two elementary operations required for distributed computation—bundling independent components into a shared medium and binding them into relation-dependent outputs—on the simplest closed geometry that carries a continuous symmetry, a cycle graph of N nodes. A single master equation of motion is solved in two regimes: the linear regime distributes a temporal input across the substrate's U(1)-organized eigenmodes to produce a feature map that behaves like a windowed Fourier transform, while the Duffing regime activates a cubic nonlinearity whose mode-mixing is constrained by the cycle symmetry into a sparse selection rule on integer wavenumbers. This mixing generates harmonic content whose amplitude and phase depend on the input waveform's shape in a manner the linear regime cannot produce. The authors extract a single scalar observable, φ₀, whose value across the quotient domain of shape parameters records the combined action of this binding operation and the substrate's dissipation; an exact π-periodicity in the shape parameter is preserved, but a time-reversal symmetry that would otherwise make φ₀ degenerate is broken by dissipation, rendering φ₀ informative. Numerical trials show that the seed-averaged value of φ₀ stays separated from its symmetric-attractor baseline even when additive band-limited noise reaches 0 dB input SNR.

Core claim

On a cycle graph substrate governed by a master equation, the linear regime sorts temporal inputs into symmetry-organized eigenmodes while the Duffing regime activates a cubic mode-mixing operation that obeys a sparse wavenumber selection rule imposed by the cycle symmetry; the resulting shape-dependent harmonics are summarised by a single observable φ₀ whose trajectory encodes the joint response of binding and dissipation, with exact π-periodicity in the shape parameter but broken time-reversal degeneracy due to dissipation, so that φ₀ remains a meaningful discriminator.

What carries the argument

The scalar observable φ₀ that summarises the Duffing ring's bound response to input shape; its symmetry structure (exact π-periodicity combined with dissipation-broken time-reversal invariance) is what makes the single number carry information about shape.

If this is right

The linear regime produces a feature representation that matches windowed-FFT performance at high SNR and exceeds it for transient signals at low SNR.
The Duffing regime supplies shape-dependent harmonic content that linear evolution on the same substrate cannot generate.
φ₀ remains separated from its symmetric-attractor value under additive band-limited noise down to 0 dB input SNR.
The framework is stated for synthetic signals; the same symmetry-breaking mechanism is left open for richer drives and biological waveforms.

Where Pith is reading between the lines

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

If the symmetry structure of φ₀ generalises, similar single-number readouts could be extracted from other closed nonlinear lattices without requiring explicit Fourier analysis.
Hardware realisations using physical Duffing-like elements could implement low-overhead shape classification by monitoring only the phase and amplitude of the fundamental mode.
The exact π-periodicity suggests that shape discrimination on this substrate is intrinsically insensitive to 180-degree inversions of the input waveform.

Load-bearing premise

The master equation and its two parameter regimes on a cycle graph of N nodes capture the essential bundling and binding behavior of more general distributed computational substrates.

What would settle it

Direct measurement of φ₀ across a range of input shapes at high SNR showing that its values collapse to a single degenerate level consistent with unbroken time-reversal symmetry, or seed-averaged means falling to the symmetric-attractor baseline at SNRs clearly above 0 dB.

Figures

Figures reproduced from arXiv: 2605.07475 by Kaspar Anton Schindler.

**Figure 1.** Figure 1: Cycle-graph ring substrate (𝑁 = 32, linear regime). (A) Topology: 𝑁 nodes in periodic nearest-neighbour connectivity, with drive 𝑠(𝑡) injected at node 𝑗 = 0. (B) Representative realvalued Fourier eigenmodes for 𝑛 = 0 (uniform, orange) and 𝑛 = 1, 2, 4, 8 (graded blue); cosine basis shown. (C) Linear-regime dispersion 𝜔𝑛/𝜔0 = √𝜆𝑛 = 2|sin(𝜋𝑛/𝑁)|, with 𝜔0 ≡ √𝐾𝑐 in this regime. The dispersion vanishes for the … view at source ↗

**Figure 2.** Figure 2: Eigenmode reservoir matches windowed FFT on stationary signals and exceeds it on transients (linear regime, 𝑁 = 32, 𝛾 = 0.5 rad/s). Each row shows a canonical drive signal (left), the reservoir’s real-valued mode amplitude |𝑎𝑛(𝑡)| (centre), and the windowed-FFT spectrogram |𝑆(𝑡, 𝑓)| in dB (right). Top to bottom: pure tone at 𝜔5 ; chirp from 𝜔1 to 𝜔12 ; Gaussian burst at 𝜔8 ; FM around 𝜔8 . The reservoir tr… view at source ↗

**Figure 3.** Figure 3: Weak-signal classification: reservoir vs windowed-FFT baseline (linear regime). Classification accuracy across the four canonical drive classes plus a noise class as a function of input SNR. Reservoir features (blue circles) maintain > 90% accuracy down to SNR ≈ −12 dB; the windowed-FFT baseline (orange squares) crosses the same threshold ∼ 3 dB earlier. Shaded bands are ±1 standard deviation across cross… view at source ↗

**Figure 4.** Figure 4: compares the harmonic-energy response of the linear ring (panel B) and the Duffing ring (panel C) to two two-tone drives that differ only in the relative phase of their second-harmonic component, Δ𝜙2 = 0 versus Δ𝜙2 = 𝜋/2 (panel A). The two drives have identical magnitude spectra: each contains a fundamental at 𝑓drive and a second harmonic at 2𝑓drive, in the same proportions. Their waveforms are visibly dif… view at source ↗

**Figure 5.** Figure 5: extends this two-point comparison to a continuous sweep of the shape phase across the full Δ𝜙2 ∈ [0, 2𝜋) domain, exposing both waveform symmetries that the framework predicts. Panel A shows 𝐸5 (Δ𝜙2 ) at the working-point nonlinearity 𝛼 = 1.5, sampled at 64 phase points and interpolated by the trigonometric polynomial of Methods §2.6. The curve has two prominent peaks separated by exactly 𝜋, with identical… view at source ↗

**Figure 6.** Figure 6: reports the seed-to-seed mean and standard deviation at each SNR, with individual seeds shown as faint markers behind the aggregate. Two horizontal reference [PITH_FULL_IMAGE:figures/full_fig_p022_6.png] view at source ↗

read the original abstract

Distributed computational substrates rely on two elementary operations: bundling, the act of populating a shared physical medium with independently retrievable components, and binding, the act of composing components into outputs whose identity depends on their relations. We study these two primitives on the simplest closed substrate carrying a continuous symmetry, a cycle graph of N nodes, in two parameter regimes of a single master equation of motion. The linear regime sorts a temporal input across the substrate's U(1)-organised eigenmodes, providing a feature representation that matches a windowed-FFT baseline at high signal-to-noise ratio and modestly outperforms it for transient signals at low SNR. The Duffing regime activates a cubic mode-mixing operation constrained by the substrate's symmetry into a sparse selection rule on integer wavenumbers, generating shape-dependent harmonic content that the linear regime cannot produce. We identify a single-number observable, $\phi_0$, that summarises the bound representation's response to input shape, and we analyse its symmetry structure: a $\pi$-periodicity in the shape parameter is exact, while a time-reversal symmetry that would render $\phi_0$ degenerate is broken by the substrate's dissipation. The asymmetric status of these two symmetries is what licenses $\phi_0$ as a meaningful single-number observable; its trajectory across the quotient domain encodes the joint response of binding and dissipation to the input shape. Numerical experiments confirm that $\phi_0$ retains its information content under additive band-limited noise, with seed-averaged means staying clearly above the symmetric-attractor value down to 0 dB input SNR. The framework is developed on synthetic signals only; extensions to richer substrates, more elaborate drives, and real biological signals are open questions for the work that follows.

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.

Desk Editor's Note private letter to a colleague

The paper gives a clean symmetry argument for a single scalar φ₀ from driven Duffing dynamics on a cycle graph that stays informative down to 0 dB SNR.

read the letter

The main takeaway is that dissipation in the Duffing regime breaks time-reversal symmetry on the cycle while preserving exact π-periodicity in the shape parameter, which lets them define a usable single-number observable φ₀. The linear regime just sorts the input across U(1) eigenmodes and roughly matches a windowed FFT, with a modest edge on transients at low SNR. The cubic term then produces shape-dependent harmonics via a symmetry-constrained selection rule on wavenumbers, and φ₀ summarizes the result. Numerical runs show the seed-averaged φ₀ stays clearly separated from the symmetric-attractor value under additive band-limited noise down to 0 dB input SNR. That check is straightforward and falsifiable on the stated model. The construction is internally consistent and the scope is explicitly limited to synthetic signals on this substrate, so there is no overreach visible in the abstract or the described experiments. The free parameters (N, drive amplitude/frequency, damping) are acknowledged but the reported tests focus on noise rather than exhaustive sweeps, which is reasonable for a first pass. This is for readers working on physical or analog substrates for bundling and binding operations, or anyone looking at nonlinear mode mixing on graphs. It supplies a concrete, symmetry-grounded example rather than a broad claim. The symmetry analysis and the noise result are sharp enough to deserve a serious referee, even if later work will need to test richer drives and real signals.

Referee Report

0 major / 2 minor

Summary. The manuscript studies bundling and binding primitives on a cycle graph of N nodes governed by a single driven Duffing master equation in linear and nonlinear regimes. The linear regime yields eigenmode-based feature representations that match or modestly exceed windowed-FFT performance at low SNR for transients. The Duffing regime introduces symmetry-constrained cubic mode mixing that produces shape-dependent harmonic content. A scalar observable φ₀ is defined whose response to input shape is licensed by exact π-periodicity in the shape parameter together with dissipation-induced breaking of time-reversal symmetry. Numerical experiments on synthetic signals demonstrate that seed-averaged φ₀ remains statistically distinguishable from its symmetric-attractor value under additive band-limited noise down to 0 dB input SNR.

Significance. If the central numerical claim holds, the work supplies a symmetry-based, largely parameter-light mechanism for shape discrimination that exploits the interplay of nonlinearity, dissipation, and substrate topology. The explicit separation of exact versus broken symmetries and the restriction to synthetic signals provide a clean foundation for subsequent extensions to biological or hardware substrates. The approach is distinctive within physical-computing literature for grounding the observable directly in the quotient structure of the driven system rather than in learned or fitted features.

minor comments (2)

The numerical section would benefit from an explicit statement of the precise definition of input SNR, the bandwidth of the added noise, the number of independent seeds, and the precise threshold used to declare 'clearly above' the symmetric-attractor value, ideally accompanied by error bars or confidence intervals on the reported means.
Notation for the master equation and the definition of φ₀ should be collected in a single early section or appendix to improve readability; currently the symmetry analysis and the observable appear to be introduced across multiple paragraphs without a consolidated equation block.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for their positive summary, significance assessment, and recommendation of minor revision. The report contains no enumerated major comments, so we have no specific points to address point-by-point. We will incorporate minor editorial and clarification changes consistent with the recommendation in the revised manuscript.

Circularity Check

0 steps flagged

No significant circularity; derivation is self-contained

full rationale

The paper defines a master equation on a cycle graph of N nodes, splits it into linear and Duffing regimes, derives the single-number observable φ₀ directly from the model's U(1) symmetry (exact π-periodicity in the shape parameter) and dissipation-induced breaking of time-reversal symmetry, then confirms its noise robustness through separate numerical experiments. No equation reduces φ₀ to a fitted parameter by construction, no load-bearing claim rests on self-citation, and the symmetry analysis follows from the stated equations without importing uniqueness theorems or ansatzes from prior work by the same authors. The framework is explicitly limited to synthetic signals on this minimal substrate.

Axiom & Free-Parameter Ledger

3 free parameters · 2 axioms · 0 invented entities

The framework rests on a single master equation with linear and cubic terms on a cycle graph possessing U(1) symmetry. Free parameters include system size N, drive amplitude and frequency, and damping coefficient. Axioms are the existence of the continuous symmetry and the validity of the Duffing nonlinearity for the chosen substrate. No new particles or forces are postulated; φ₀ is an observable constructed from the dynamics.

free parameters (3)

N (number of nodes)
Cycle graph size; controls the discrete wavenumbers available for mode mixing.
Drive amplitude and frequency
Parameters that select between linear and Duffing regimes and set the operating point.
Damping coefficient
Controls dissipation that breaks time-reversal symmetry.

axioms (2)

domain assumption The substrate is a cycle graph with exact U(1) rotational symmetry.
Invoked to organize eigenmodes and constrain the cubic mode-mixing selection rule.
domain assumption The master equation accurately models the physical or computational substrate under study.
Central modeling choice separating linear sorting from nonlinear binding.

pith-pipeline@v0.9.0 · 5614 in / 1616 out tokens · 40425 ms · 2026-05-11T02:05:49.044005+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

27 extracted references · 27 canonical work pages

[1]

All Together Now: Analogies between Chimera State Collapses and Epileptic Seizures

“All Together Now: Analogies between Chimera State Collapses and Epileptic Seizures.” Scien- tific Reports 6 : 23000. https:/ /doi.org/10.1038/srep23000. Boyd, R. W

work page doi:10.1038/srep23000
[3]

Trends in Cognitive Sciences (2018) https://doi.org/10.1016/j.tics

“Brain Oscillations and the Importance of Waveform Shape.” Trends in Cognitive Sciences 21 (2): 137–49. https:/ /doi.org/10.1016/j.tics. 2016.12.008. Cole, S. R., and B. Voytek

work page doi:10.1016/j.tics 2016
[4]

Cycle-by-Cycle Analysis of Neural Oscillations

“Cycle-by-Cycle Analysis of Neural Oscillations.” Journal of Neurophysiology 122 (2): 849–61. https:/ /doi.org/10.1152/jn.00273.2019. Cole, S. R., R. van der Meij, E. J. Peterson, C. de Hemptinne, P. A. Starr, and B. Voytek

work page doi:10.1152/jn.00273.2019 2019
[5]

Nonsinusoidal Beta Oscillations Reflect Cortical Pathophysiology in Parkinson's Disease

“Nonsinusoidal Beta Oscillations Reflect Cortical Pathophysiology in Parkinson's Disease.” The Journal of Neuroscience 37 (18): 4830–40. https:/ /doi. org/10.1523/JNEUROSCI.2208-16.2017. Donoghue, T., N. Schaworonkow, and B. Voytek

work page doi:10.1523/jneurosci.2208-16.2017 2017
[6]

Methodological Considera - tions for Studying Neural Oscillations

“Methodological Considera - tions for Studying Neural Oscillations.” European Journal of Neuroscience 55 (11–12): 3502–27. https:/ /doi.org/10.1111/ejn.15361. Enel, P., E. Procyk, R. Quilodran, and P. F. Dominey

work page doi:10.1111/ejn.15361
[7]

Reservoir Computing Properties of Neural Dynamics in Prefrontal Cortex

“Reservoir Computing Properties of Neural Dynamics in Prefrontal Cortex.” Plos Computational Biology 12 (6): e1004967. https:/ /doi.org/10.1371/journal.pcbi.1004967. Gallou, O., J. Bartels, S. Ghosh, K. A. Schindler, J. Sarnthein, and G. Indiveri

work page doi:10.1371/journal.pcbi.1004967
[8]

Online Epileptic Seizure Detection in Long-Term iEEG Recordings Using Mixed- Signal Neuromorphic Circuits

“Online Epileptic Seizure Detection in Long-Term iEEG Recordings Using Mixed- Signal Neuromorphic Circuits.” In “2024 IEEE Biomedical Circuits and Systems 32 Conference (Biocas).” Special issue, 2024 IEEE Biomedical Circuits and Systems Con- ference (Biocas), 2024, 1–5. https:/ /doi.org/10.1109/biocas61083.2024.10798236. Glomb, K., J. Rué Queralt, D. Pasc...

work page doi:10.1109/biocas61083.2024.10798236 2024
[9]

Connectome Spectral Analysis to Track EEG Task Dynamics on a Subsecond Scale

“Connectome Spectral Analysis to Track EEG Task Dynamics on a Subsecond Scale.” Neuroimage 221 : 117137. https:/ /doi.org/10.1016/j.neuroimage.2020.117137. Holmes, P

work page doi:10.1016/j.neuroimage.2020.117137 2020
[10]

A Nonlinear Oscillator with a Strange Attractor

“A Nonlinear Oscillator with a Strange Attractor.” Philosophical Trans- actions of the Royal Society of London a 292 (1394): 419–48. https:/ /doi.org/10.1098/ rsta.1979.0068. Hoppensteadt, F. C., and E. M. Izhikevich

work page arXiv 1979
[11]

Oscillatory Neurocomputers with Dynamic Connectivity

“Oscillatory Neurocomputers with Dynamic Connectivity.” Physical Review Letters 82 (14): 2983–86. https:/ /doi.org/ 10.1103/PhysRevLett.82.2983. Hyafil, A

work page doi:10.1103/physrevlett.82.2983
[12]

2015.00370

https:/ /doi.org/10.3389/fnins. 2015.00370. Jaeger, H

work page doi:10.3389/fnins 2015
[13]

Reservoir computing approaches to recurrent neural network training

“Reservoir Computing Approaches to Recurrent Neural Network Training.” Computer Science Review 3 (3): 127–49. https:/ /doi.org/ 10.1016/j.cosrev.2009.03.005. Maass, W., T. Natschläger, and H. Markram

work page doi:10.1016/j.cosrev.2009.03.005 2009
[14]

Physical Reservoir Computing Enabled by Solitary Waves and Biologically Inspired Nonlinear Transformation of Input Data

“Physical Reservoir Computing Enabled by Solitary Waves and Biologically Inspired Nonlinear Transformation of Input Data.” Dynamics 4 (1): 119–34. https:/ /doi.org/10.3390/dynamics4010007. Müller, E. J., B. R. Munn, K. M. Aquino, J. M. Shine, and P. A. Robinson

work page doi:10.3390/dynamics4010007
[15]

The Music of the Hemispheres: Cortical Eigenmodes as a Physical Basis for Large- Scale Brain Activity and Connectivity Patterns

“The Music of the Hemispheres: Cortical Eigenmodes as a Physical Basis for Large- Scale Brain Activity and Connectivity Patterns.” Frontiers in Human Neuroscience 16 : 1062487. https:/ /doi.org/10.3389/fnhum.2022.1062487. Müller, M. F., C. Rummel, M. Goodfellow, and K. A. Schindler

work page doi:10.3389/fnhum.2022.1062487 2022
[16]

Standing Waves as an Explanation for Generic Stationary Correlation Patterns in Noninvasive EEG of Focal Onset Seizures

“Standing Waves as an Explanation for Generic Stationary Correlation Patterns in Noninvasive EEG of Focal Onset Seizures.” Brain Connectivity 4 (2): 131–44. https:/ /doi.org/10.1089/ brain.2013.0192. Nayfeh, A. H., and D. T. Mook

work page arXiv 2013
[17]

The Brain Wave Equation: A Model for the EEG

“The Brain Wave Equation: A Model for the EEG.” Mathematical Biosciences 21 (3–4): 279–97. https:/ /doi.org/10.1016/0025-5564(74)90020-0. Pang, J. C., K. M. Aquino, M. Oldehinkel, P. A. Robinson, B. D. Fulcher, M. Breakspear, and A. Fornito

work page doi:10.1016/0025-5564(74)90020-0
[18]

Geometric Constraints on Human Brain Function

“Geometric Constraints on Human Brain Function.” Nature 618 (7965): 566–74. https:/ /doi.org/10.1038/s41586-023-06098-1. 33 Plate, T. A

work page doi:10.1038/s41586-023-06098-1
[19]

Propagation and Stability of Waves of Electrical Activity in the Cerebral Cortex

“Propagation and Stability of Waves of Electrical Activity in the Cerebral Cortex.” Physical Review E 56 (1): 826–40. https:/ /doi.org/10.1103/PhysRevE.56.826. Schindler, K. A., and A. Rahimi

work page doi:10.1103/physreve.56.826
[20]

A Primer on Hyperdimensional Computing for iEEG Seizure Detection

“A Primer on Hyperdimensional Computing for iEEG Seizure Detection.” Frontiers in Neurology 12 : 701791. https:/ /doi.org/10.3389/ fneur.2021.701791. Schindler, K. A., C. E. Elger, and K. Lehnertz

work page arXiv 2021
[21]

Increasing Synchronization May Promote Seizure Termination: Evidence from Status Epilepticus

“Increasing Synchronization May Promote Seizure Termination: Evidence from Status Epilepticus.” Clinical Neuro- physiology 118 (9): 1955–68. https:/ /doi.org/10.1016/j.clinph.2007.06.006. Schindler, K. A., H. Gast, M. Goodfellow, and C. Rummel

work page doi:10.1016/j.clinph.2007.06.006 1955
[22]

On Seeing the Trees and the Forest: Single-Signal and Multisignal Analysis of Periictal Intracranial EEG

“On Seeing the Trees and the Forest: Single-Signal and Multisignal Analysis of Periictal Intracranial EEG.” Epilepsia 53 (9): 1658–68. https:/ /doi.org/10.1111/j.1528-1167.2012.03588.x. Schindler, K. A., H. Leung, C. E. Elger, and K. Lehnertz

work page doi:10.1111/j.1528-1167.2012.03588.x 2012
[23]

Assessing Seizure Dynamics by Analysing the Correlation Structure of Multichannel Intracranial EEG

“Assessing Seizure Dynamics by Analysing the Correlation Structure of Multichannel Intracranial EEG.” Brain 130 (1): 65–77. https:/ /doi.org/10.1093/brain/awl304. Schindler, K. A., C. Rummel, R. G. Andrzejak, M. Goodfellow, F. Zubler, E. Abela, R. Wiest, C. Pollo, A. Steimer, and H. Gast

work page doi:10.1093/brain/awl304
[24]

Ictal Time-Irreversible Intracranial EEG Signals as Markers of the Epileptogenic Zone

“Ictal Time-Irreversible Intracranial EEG Signals as Markers of the Epileptogenic Zone.” Clinical Neurophysiology 127 (9): 3051–58. https:/ /doi.org/10.1016/j.clinph.2016.07.001. Shuman, D. I., S. K. Narang, P. Frossard, A. Ortega, and P. Vandergheynst

work page doi:10.1016/j.clinph.2016.07.001 2016
[25]

The Emerging Field of Signal Processing on Graphs: Extending High-Dimensional Data Analysis to Networks and Other Irregular Domains

“The Emerging Field of Signal Processing on Graphs: Extending High-Dimensional Data Analysis to Networks and Other Irregular Domains.” IEEE Signal Processing Magazine 30 (3): 83–98. https:/ /doi.org/10.1109/MSP.2012.2235192. Smolensky, P

work page doi:10.1109/msp.2012.2235192 2012
[26]

Tensor Product Variable Binding and the Representation of Symbolic Structures in Connectionist Systems , url =

https:/ /doi.org/10.1016/0004-3702(90)90007-M. Todri-Sanial, A., C. Delacour, M. Abernot, and F. Sabo

work page doi:10.1016/0004-3702(90)90007-m
[27]

Xie, X., C

https:/ /doi.org/10.1038/s44335-024-00015-z. Xie, X., C. Cai, P. F. Damasceno, S. S. Nagarajan, and A. Raj

work page doi:10.1038/s44335-024-00015-z
[28]

Emergence of Canonical Functional Networks from the Structural Connectome

“Emergence of Canonical Functional Networks from the Structural Connectome.” Neuroimage 237 : 118190. https:/ /doi.org/10.1016/j.neuroimage.2021.118190. Yilmaz, G., A. Seiler, O. Chételat, and K. A. Schindler

work page doi:10.1016/j.neuroimage.2021.118190 2021