pith. sign in

arxiv: 2605.16761 · v1 · pith:BA7M27OQnew · submitted 2026-05-16 · 🧮 math.DS · q-bio.NC

A Mathematical Characterization of Neural Activation Induced by Temporal Interference Stimulation

Esteban Paduro , Antoine Chaillet , Mario Sigalotti This is my paper

Pith reviewed 2026-05-19 19:51 UTC · model grok-4.3

classification 🧮 math.DS q-bio.NC
keywords temporal interference stimulationFitzHugh-Nagumo modelgeometric singular perturbationphase-plane analysisneural activation thresholdsbeat frequencyneuromodulationtonic firing
0
0 comments X

The pith

Amplitudes and beat frequency together decide whether temporal interference stimulation leaves a neuron quiet, produces brief spikes, or drives continuous firing.

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

The paper builds a mathematical description of how temporal interference stimulation, created by two slightly offset high-frequency currents, interacts with a single neuron. It applies phase-plane methods and geometric singular perturbation to the FitzHugh-Nagumo equations to classify the resulting behaviors. A reader would care because the classification supplies explicit thresholds in the space of current amplitudes and frequency difference that separate no response from sustained activity. The work therefore supplies a predictive rule set for when this non-invasive technique will activate deep neurons rather than leave them unaffected.

Core claim

In the FitzHugh-Nagumo model driven by the envelope of two high-frequency sinusoids, the neuron remains quiescent below a lower threshold surface in amplitude-beat-frequency space, produces only transient responses between that surface and an upper threshold, and exhibits persistent tonic firing once both thresholds are crossed, with the precise location of the surfaces obtained by geometric singular perturbation reduction of the fast oscillations.

What carries the argument

The FitzHugh-Nagumo system reduced by geometric singular perturbation, which replaces the rapid carrier oscillations with a slow effective drive whose amplitude is set by the two current strengths and whose frequency is the beat frequency.

If this is right

  • Parameter regions can be plotted in the plane of the two current amplitudes for each fixed beat frequency, giving explicit boundaries between quiescent, transient, and tonic regimes.
  • Increasing the beat frequency while holding amplitudes fixed moves the system across the threshold surfaces and can switch the neuron from tonic firing back to quiescence.
  • The same reduction technique yields analytic approximations for the minimal current strengths needed to reach each regime.
  • The classification extends immediately to any slowly varying envelope whose shape is known, not only sinusoidal beats.

Where Pith is reading between the lines

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

  • The same envelope-driven threshold surfaces could be used to design current waveforms that selectively activate only a desired subset of neurons at a given depth.
  • Numerical continuation of the reduced slow system would give quantitative predictions for the minimal stimulation duration required to elicit at least one spike.
  • The framework suggests testing whether real cortical neurons obey the predicted ordering of thresholds when the carrier frequencies are scaled while the beat frequency is held constant.

Load-bearing premise

The FitzHugh-Nagumo equations plus the singular perturbation reduction faithfully reproduce the firing thresholds of real neurons exposed to the low-frequency envelope of temporal interference stimulation.

What would settle it

A direct voltage-clamp or patch-clamp recording showing that a biological neuron stays silent under parameter values the reduced model predicts will produce tonic firing would falsify the central claim.

Figures

Figures reproduced from arXiv: 2605.16761 by Antoine Chaillet, Esteban Paduro, Mario Sigalotti.

Figure 1
Figure 1. Figure 1: v-nullclines Cc of the frozen system (5) for three different values of c. The minimum of the cubics Cc evolves monotonically with c and belongs to the arc Jm. The segment Je contains the equilibria of (5) for c ∈ [−1, 1]. • E0: the set of all pairs (A, B) ∈ Eunique for which A + B < √ 2 and, given any c ∈ [−1, 1], ve(c) < vm(c). Then Eunique and E0 do not depend on ε and we have the following: Eunique = ( … view at source ↗
Figure 2
Figure 2. Figure 2: (a)-(e) Represent the regions described in the proof of Lemma 6 in the [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Description of the geometric elements used in the construction of the invariant region [PITH_FULL_IMAGE:figures/full_fig_p010_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Region H used in the proof of Theorem 9. The second picture is a zoomed-in picture near the point (ve(−1), we(−1)) • S6 is the arc of the integral curve of the vector field G−1 that connects (p, q) and P1. Next, we show that the region H satisfies our desired property. Recall that, by definition of t ∗ , s(t) = 1 for t ∈ (t ∗ , t∗ + π/η). Let t ∗ + T0 be the exit time from H of the integral line of G1 star… view at source ↗
Figure 5
Figure 5. Figure 5: Vector plot of X↓ and X↑ together with some trajectories of the system. (a)-(d) use γ = 0.5, β = 0.8, A = B = 0.3. (a) correspond to X↓ with κ = 1.3, (b) X↓ with κ = 2, (c) X↑ with κ = 1.3, (d) X↑ with κ = 2, 5.2 Relating trajectories of (21) and (22) 5.2.1 Tracking regular trajectories of (22) by trajectories of (21) A key fact that makes (22) an informative limit system for (21) is that the trajectories … view at source ↗
Figure 6
Figure 6. Figure 6: Regions Q and Qδ in the proof of Lemma 17 Consider w¯ in the interval (wm(cos(κs0)), wm(1)) and set s¯ := min{s > s0 | wm(cos(κs)) = ¯w}. Let v1 < v2 < 0 be such that (v1, w¯) and (v2, w¯) are the two intersection points of the cubic Ccos(κs0) with the horizontal line {w = ¯w} that lie in the half-plane {v < 0}. Let, moreover, δ be such that wm(cos(κs0)) + δ < w¯ (See [PITH_FULL_IMAGE:figures/full_fig_p01… view at source ↗
Figure 7
Figure 7. Figure 7: (a) - (d) Vector plot of the field X↑. All cases use γ = 0.5, β = 0.8. (a) A = B = 0.3, κ = 0.5, (b) A = B = 0.3, κ = 1.58, (c) A = B = 0.15, κ = 6.4, (d),(e) A = B = 0.3, κ = 3. (a) - (c) satisfy the conditions of Theorem 8, whereas (d) and (e) satisfy Assumption(E) Lemma 19. Let (v, w) be a maximal integral curve of the vector field X↑ with initial condition (ve(−1), we(−1)). Suppose that (v, w) reaches … view at source ↗
Figure 8
Figure 8. Figure 8: Plot of activation for fixed values of β = 0.8, γ = 0.5, and varying (κ, ε) with η = κε. The dark region indicates tonic spiking, defined as the presence of at least 2 action potentials. The maximum simulation time is T = 2000, chosen to observe activation even for the smallest ε. The vertical red line indicates the infimum of the values of κ for which there exists some s0 at which the escaping condition (… view at source ↗
Figure 9
Figure 9. Figure 9: Results for Experiment 2. In the first column, we verify Assumption (E) graphically, and the condition [PITH_FULL_IMAGE:figures/full_fig_p020_9.png] view at source ↗
read the original abstract

Temporal Interference Stimulation (TIS) is a non-invasive neuromodulation technique in which two high-frequency sinusoidal currents with slightly different frequencies generate a low-frequency envelope that can activate deep neural structures. This study investigates the conditions under which TIS elicits action potentials in a single neuron modeled by the FitzHugh-Nagumo system. This research integrates phase-plane analysis and geometric singular perturbation to develop a mathematical framework for analyzing TIS. By combining a mathematical analysis of differential equations with computer simulations, the study elucidates how the amplitudes and beat frequency jointly determine whether the neuron remains quiescent, exhibits only transient responses, or undergoes persistent (tonic) firing.

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 mathematical framework for Temporal Interference Stimulation (TIS) applied to a single neuron modeled by the FitzHugh-Nagumo (FHN) equations. It combines phase-plane analysis with geometric singular perturbation theory to reduce the system and characterize how the amplitudes of the two high-frequency currents and the beat frequency jointly determine three regimes: quiescent, transient responses only, or persistent tonic firing. The analysis is supplemented by numerical simulations that illustrate the predicted boundaries between these regimes.

Significance. If the central claims hold, the work supplies a systematic, analytically grounded approach to predicting activation thresholds under TIS envelopes, which is relevant for parameter selection in non-invasive deep-brain neuromodulation. The explicit use of geometric singular perturbation on the standard FHN model, together with the regime classification derived from the reduced slow system, offers a template that could be extended to other high-frequency stimulation protocols. The combination of rigorous reduction with illustrative simulations strengthens the practical utility of the results.

major comments (2)
  1. [§3] §3 (Geometric Singular Perturbation Reduction) and the subsequent regime classification: the central claim that amplitudes and beat frequency jointly set the quiescent/transient/tonic boundaries rests on the reduced slow system obtained by treating the beat frequency as asymptotically small. The manuscript does not state or verify the required separation-of-scales condition (beat frequency ≪ fast time scale of the FHN voltage variable) nor provide bounds on the beat frequency for which the envelope approximation remains accurate for tonic-firing classification. This omission is load-bearing because, outside the asymptotic regime, residual carrier-frequency effects can shift the effective thresholds.
  2. [§4] §4 (Numerical Simulations) and the reported regime boundaries: the simulations are used to confirm the analytically predicted boundaries, yet no systematic sweep or error analysis is presented that tests the reduction when the beat frequency is only moderately small (e.g., comparable to the FHN recovery time scale). Without such checks, it is unclear whether the reported boundaries remain reliable when the geometric singular perturbation hypothesis is only approximately satisfied.
minor comments (2)
  1. Notation for the two current amplitudes and the beat frequency is introduced without a consolidated table of symbols; a single reference table would improve readability.
  2. Figure captions for the phase-plane and time-series plots do not list the exact parameter values (including the small parameter ε and the specific beat frequencies) used in each panel.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and constructive comments on our manuscript. The points raised regarding the separation-of-scales assumption and the validation of the reduced system are well taken. We address each major comment below and have revised the manuscript to incorporate explicit statements of the asymptotic conditions and additional numerical checks.

read point-by-point responses

  1. Referee: §3 (Geometric Singular Perturbation Reduction) and the subsequent regime classification: the central claim that amplitudes and beat frequency jointly set the quiescent/transient/tonic boundaries rests on the reduced slow system obtained by treating the beat frequency as asymptotically small. The manuscript does not state or verify the required separation-of-scales condition (beat frequency ≪ fast time scale of the FHN voltage variable) nor provide bounds on the beat frequency for which the envelope approximation remains accurate for tonic-firing classification. This omission is load-bearing because, outside the asymptotic regime, residual carrier-frequency effects can shift the effective thresholds.

    Authors: We agree that the separation-of-scales condition should be stated explicitly. The geometric singular perturbation reduction in §3 is derived under the assumption that the beat frequency is asymptotically small relative to the fast voltage dynamics of the FitzHugh-Nagumo model. In the revised manuscript we have added a dedicated paragraph at the start of §3 that states this condition (beat frequency ≪ 1 in the normalized fast time scale) and explains its role in justifying the envelope approximation for the slow system. We also include a brief remark on the regime of validity for the tonic-firing classification, noting that the reduced slow system captures the leading-order behavior when the beat frequency remains sufficiently small compared with the recovery time scale. While we do not derive explicit numerical bounds for every parameter combination, the singular-perturbation framework itself delineates the asymptotic regime; outside it the classification may be perturbed by residual carrier effects, which we now acknowledge as a limitation in the discussion. revision: yes

  2. Referee: §4 (Numerical Simulations) and the reported regime boundaries: the simulations are used to confirm the analytically predicted boundaries, yet no systematic sweep or error analysis is presented that tests the reduction when the beat frequency is only moderately small (e.g., comparable to the FHN recovery time scale). Without such checks, it is unclear whether the reported boundaries remain reliable when the geometric singular perturbation hypothesis is only approximately satisfied.

    Authors: We acknowledge that a more systematic test of the reduction for moderately small beat frequencies would strengthen the numerical validation. In the revised §4 we have added a new set of simulations that sweep the beat frequency from asymptotically small values up to values comparable to the recovery time scale. For each value we compare the simulated activation thresholds against the analytically predicted boundaries and report the relative deviation. The results show that the qualitative regime boundaries remain reliable even when the separation-of-scales hypothesis is only approximately satisfied, with quantitative error increasing gradually as the beat frequency approaches the recovery time scale. These additional checks are summarized in an expanded figure and accompanying text. revision: yes

Circularity Check

0 steps flagged

No circularity: standard geometric singular perturbation applied to FHN model yields independent regime classification

full rationale

The derivation begins from the standard FitzHugh-Nagumo equations driven by the TIS envelope and applies geometric singular perturbation theory to obtain a reduced slow system whose equilibria and bifurcations classify the three firing regimes. This reduction is a general mathematical technique whose validity conditions (slow beat frequency relative to fast variable) are external to the result and not defined by the classification itself. No parameters are fitted to the target regimes, no self-citations carry the central claim, and the phase-plane analysis produces new statements about the joint dependence on amplitudes and beat frequency that are not tautological with the inputs. The paper is therefore self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the standard FitzHugh-Nagumo model and the applicability of geometric singular perturbation to separate fast spiking from slow envelope dynamics; no new free parameters or invented entities are introduced in the abstract.

axioms (2)
  • domain assumption The FitzHugh-Nagumo system is an adequate reduced model for the voltage and recovery dynamics of a neuron under high-frequency current injection.
    Invoked when the authors state they model a single neuron by the FitzHugh-Nagumo system (abstract).
  • standard math Geometric singular perturbation theory applies to the non-autonomous system obtained by adding the TIS envelope.
    Used to develop the mathematical framework for analyzing activation (abstract).

pith-pipeline@v0.9.0 · 5634 in / 1379 out tokens · 29576 ms · 2026-05-19T19:51:36.719247+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

29 extracted references · 29 canonical work pages

  1. [1]

    Blanchini

    F. Blanchini. Set invariance in control.Automatica J. IFAC, 35(11):1747–1767, 1999. doi:10.1016/S0005- 1098(99)00113-2

    work page doi:10.1016/s0005- 1999

  2. [2]

    Cerpa, N

    E. Cerpa, N. Corrales, M. Courdurier, L. E. Medina, and E. Paduro. The Impact of High-Frequency-Based Stability on the Onset of Action Potentials in Neuron Models.SIAM Journal on Applied Mathematics, 84(5):1910–1936, 2024, 2402.05886. doi:10.1137/24M1645632

    work page doi:10.1137/24m1645632 1910

  3. [3]

    Cerpa, M

    E. Cerpa, M. Courdurier, E. Hernández, L. E. Medina, and E. Paduro. A partially averaged system to model neuron responses to interferential current stimulation.J. Math. Biol., 86(1):8, 2023. doi:10.1007/s00285-022- 01839-8

    work page doi:10.1007/s00285-022- 2023

  4. [4]

    E.Cerpa, M.Courdurier, E.Hernández, L.E.Medina, andE.Paduro. Approximationandstabilityresultsfor the parabolic FitzHugh-Nagumo system with combined rapidly oscillating sources.Discrete and Continuous Dynamical Systems, 50(0):69–102, 2025, 2305.00123. doi:10.3934/dcds.2026001

    work page doi:10.3934/dcds.2026001 2025

  5. [5]

    Grossman, D

    N. Grossman, D. Bono, N. Dedic, S. B. Kodandaramaiah, A. Rudenko, H.-J. Suk, A. M. Cassara, E. Neufeld, N. Kuster, L.-H. Tsai, A. Pascual-Leone, and E. S. Boyden. Noninvasive Deep Brain Stimulation via Tem- porally Interfering Electric Fields.Cell, 169(6):1029–1041.e16, jun 2017. doi:10.1016/j.cell.2017.05.024

    work page doi:10.1016/j.cell.2017.05.024 2017

  6. [6]

    M. Hayashi. Global Asymptotic Stability of FitzHugh-Nagumo System. InProceedings of the Ninth Interna- tional Colloquium on Differential Equations, pages 191–196. De Gruyter, 1999. doi:10.1515/9783112318973- 029

    work page doi:10.1515/9783112318973- 1999

  7. [7]

    M. W. Hirsch, S. Smale, and R. L. Devaney.Differential equations, dynamical systems, and an introduction to chaos. Academic Press, Waltham, MA, 3rd edition, 2013

    work page 2013

  8. [8]

    Karimi, A

    F. Karimi, A. Attarpour, R. Amirfattahi, and A. Z. Nezhad. Computational analysis of non-invasive deep brain stimulation based on interfering electric fields.Physics in Medicine & Biology, 64(23):235010, dec 2019. doi:10.1088/1361-6560/ab5229

    work page doi:10.1088/1361-6560/ab5229 2019

  9. [9]

    Karimi, R

    N. Karimi, R. Amirfattahi, and A. Zeidaabadi Nezhad. Neuromodulation effect of temporal interference stimulation based on network computational model.Frontiers in Human Neuroscience, 18, Sept. 2024. doi:10.3389/fnhum.2024.1436205

    work page doi:10.3389/fnhum.2024.1436205 2024

  10. [10]

    Kaumann and U

    E. Kaumann and U. Staude. Uniqueness and nonexistence of limit cycles for the FitzHugh equation. In Equadiff 82 (Würzburg, 1982), volume 1017 ofLecture Notes in Math., pages 313–321. Springer, Berlin,

    work page 1982

  11. [11]

    doi:10.1007/BFb0103262

    work page doi:10.1007/bfb0103262

  12. [12]

    H. K. Khalil.Nonlinear Systems, 3rd Edition. Pearson, 2002

    work page 2002

  13. [13]

    Kostova, R

    T. Kostova, R. Ravindran, and M. Schonbek. FitzHugh-Nagumo revisited: types of bifurcations, periodical forcingand stabilityregions bya Lyapunovfunctional.Internat. J. Bifur. Chaos Appl. Sci. Engrg., 14(3):913– 925, 2004. doi:10.1142/S0218127404009685

    work page doi:10.1142/s0218127404009685 2004

  14. [14]

    BiophysicsofTemporalInterferenceStimulation

    E.Mirzakhalili, B.Barra, M.Capogrosso, andS.F.Lempka. BiophysicsofTemporalInterferenceStimulation. Cell Systems, 11(6):557–572.e5, 2020. doi:10.1016/j.cels.2020.10.004

    work page doi:10.1016/j.cels.2020.10.004 2020

  15. [15]

    Missey, E

    F. Missey, E. Acerbo, A. S. Dickey, J. Trajlinek, O. Studnička, C. Lubrano, M. d. A. e Silva, E. Brady, V. Všiansk` y, J. Szabo, et al. Non-invasive temporal interference stimulation of the hippocampus suppresses epileptic biomarkers in patients with epilepsy: Biophysical differences between kilohertz and amplitude modulated stimulation.Brain stimulation,...

    work page doi:10.1016/j.brs.2025.11.008 2025

  16. [16]

    Missey, E

    F. Missey, E. Rusina, E. Acerbo, B. Botzanowski, A. Trébuchon, F. Bartolomei, V. Jirsa, R. Carron, and A. Williamson. Orientation of Temporal Interference for Non-invasive Deep Brain Stimulation in Epilepsy. Frontiers in Neuroscience, 15(June):1–13, 2021. doi:10.3389/fnins.2021.633988

    work page doi:10.3389/fnins.2021.633988 2021

  17. [17]

    M. Nagumo. Über die Lage der Integralkurven gewöhnlicher Differentialgleichungen.Proc. Phys.-Math. Soc. Japan (3), 24:551–559, 1942. 23

    work page 1942

  18. [18]

    Opančar, P

    A. Opančar, P. Ondráčková, D. S. Rose, J. Trajlinek, V. Derek, and E. D. Głowacki. The same biophysical mechanism is involved in both temporal interference and direct kHz stimulation of peripheral nerves.Nature Communications, 16(1):9006, Oct. 2025. doi:10.1038/s41467-025-64059-w

    work page doi:10.1038/s41467-025-64059-w 2025

  19. [19]

    Plovie, R

    T. Plovie, R. Schoeters, T. Tarnaud, W. Joseph, and E. Tanghe. Nonlinearities and timescales in neural mod- els of temporal interference stimulation.Bioelectromagnetics, 46(1):e22522, 2025. doi:10.1002/bem.22522

    work page doi:10.1002/bem.22522 2025

  20. [20]

    NonexistenceofperiodicsolutionsfortheFitzHughnervesystem.Quarterly of Applied Mathematics, 49(3):543–554, Sept

    J.Sugie. NonexistenceofperiodicsolutionsfortheFitzHughnervesystem.Quarterly of Applied Mathematics, 49(3):543–554, Sept. 1991. doi:10.1090/qam/1121685

    work page doi:10.1090/qam/1121685 1991

  21. [21]

    A. R. Teel, J. Peuteman, and D. Aeyels. Semi-global practical asymptotic stability and averaging.Systems & control letters, 37(5):329–334, 1999. doi:10.1016/S0167-6911(99)00039-0

    work page doi:10.1016/s0167-6911(99)00039-0 1999

  22. [22]

    A. N. Tihonov. Systems of differential equations containing small parameters in the derivatives.Mat. Sbornik N.S., 31(73):575–586, 1952

    work page 1952

  23. [23]

    A. B. Vasil’eva. Asymptotic behaviour of solutions of certain problems for ordinary non-linear differential equations with a small parameter multiplying the highest derivatives.Uspehi Mat. Nauk, 18(3(111)):15–86,

    work page

  24. [24]

    doi:10.1070/RM1963v018n03ABEH001137

    work page doi:10.1070/rm1963v018n03abeh001137

  25. [25]

    I. R. Violante, K. Alania, A. M. Cassarà, E. Neufeld, E. Acerbo, R. Carron, A. Williamson, D. L. Kurtin, E. Rhodes, A. Hampshire, et al. Non-invasive temporal interference electrical stimulation of the human hippocampus.Nature neuroscience, 26(11):1994–2004, 2023. doi:10.1038/s41593-023-01456-8

    work page doi:10.1038/s41593-023-01456-8 1994

  26. [26]

    B. Wang, A. S. Aberra, W. M. Grill, and A. V. Peterchev. Responses of Model Cortical Neurons to Temporal Interference Stimulation and Related Transcranial Alternating Current Stimulation Modalities.Journal of neural engineering, 19(6):066047, Jan. 2022. doi:10.1088/1741-2552/acab30

    work page doi:10.1088/1741-2552/acab30 2022

  27. [27]

    A. R. Ward. Electrical stimulation using kilohertz-frequency alternating current.Physical Therapy, 89(2):181–190, 2009. doi:10.2522/ptj.20080060

    work page doi:10.2522/ptj.20080060 2009

  28. [28]

    S. Xu, H. Cui, X. Xiao, F. Manshaii, G. Hong, and J. Chen. Precision at Deep Brain: Noninvasive Temporal Interference Stimulation.ACS Nano, 19(46):39589–39614, Nov. 2025. doi:10.1021/acsnano.5c15238

    work page doi:10.1021/acsnano.5c15238 2025

  29. [29]

    C. Yang, Y. Xu, X. Feng, B. Wang, Y. Du, K. Wang, J. Lü, L. Huang, Z. Qian, Z. Wang, N. Chen, J. Zhou, C. Zhang, and Y. Liu. Transcranial Temporal Interference Stimulation of the Right Globus Pallidus in Parkinson’s Disease.Movement Disorders, 40(6):1061–1069, June 2025. doi:10.1002/mds.29967. 24

    work page doi:10.1002/mds.29967 2025