{"id":"6fd7be44-f5e4-4cd3-a313-091f74ad8388","arxiv_id":"2412.06804","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A pair of delay-coupled excitable theta neurons has explicit synchronous and alternating periodic solutions whose stability changes at the minimum period and at saddle-node bifurcations.","lead":"This paper works out exact formulas for the firing rhythms of two brain-like neurons that excite each other after a fixed delay, and pinpoints when each rhythm is stable. Because the results are analytic, they give a clean standard case for understanding delay-coupled excitable systems in neuroscience, optics, and chemistry.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Stability results hinge on an unproved equivalence between eigenvalues of the linearized firing-time map and Floquet multipliers of the impulsive DDE; border-collision events could invalidate the linearization.","rationale":"The reader's weakest assumption correctly identifies the lack of proof that the firing-time-map eigenvalues are the Floquet multipliers of the impulsive DDE. This is the central load-bearing step: without it, the analytic stability results and the bifurcation set are not rigorously established. I agree with the reader that this makes the paper conditionally acceptable rather than fully definitive. The concern is a gap in justification, not a demonstrated error; numerical evidence in Sec. 4 for the smooth system supports the qualitative picture, which mitigates the risk. Secondary issues include the introduction's claim that 'all periodic solutions' are found (contradicted by Sec. 5, which reports additional stable periodic solutions) and the abstract's statement that symmetry-broken solutions are 'all unstable' (whereas Appendix A.2 shows the τ=0 alternating symmetry-broken family is neutrally stable). These are overstatements that should be corrected, but they do not affect the analysis of the synchronous and alternating families themselves. The concrete test above would settle whether the main gap is real; until then, CONDITIONAL remains the appropriate verdict, so no change to the reader's decision is recommended.","tokens_in":14,"tokens_out":21935,"duration_ms":479171,"concrete_test":"For a representative case (e.g., n=1, κ=5, τ=2), construct the monodromy matrix of the periodic orbit of (2)-(3) directly: linearize the explicit flow (5)-(7), apply the saltation matrix for each Dirac-delta reset (4), and multiply the pieces around the period. Compare the eigenvalues of this matrix with the roots of Ds(λ) (Eq. 19) and Da(λ) (Eq. 45). If the nontrivial eigenvalues match, the reduction is validated; if extra eigenvalues of modulus ≥ 1 appear, the stability claims in Propositions 1-2 are unsupported. An independent cross-check: continue the smooth system (55)-(56) with DDE-BIFTOOL for increasingly sharp pulses (m=10,20,50) and verify that the computed Floquet multipliers converge to the roots of Ds/Da, quantifying the discrepancy at finite m.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Sections 2.1 and 3.1.3 linearize the firing-time map (Eqs. 11-12, 34-35) and assert that the roots of the resulting characteristic equations Ds(λ)=0 (Eq. 19) and Da(λ)=0 (Eq. 45) are the Floquet multipliers of the periodic solutions of the discontinuous DDE (2)-(3). This equivalence is not proved. System (2)-(3) is an impulsive delay differential equation with an infinite-dimensional state (the history segment); its stability is governed by the monodromy operator, whose spectrum generally contains infinitely many eigenvalues, most of them contractive. The paper gives no argument that the finite-dimensional firing-time reduction captures every multiplier of modulus ≥ 1. In addition, the reduction assumes the event count n and the order of events within the delay window are fixed under small perturbations. If a perturbed firing time crosses a boundary of the window (i.e., τ - nT or τ - (n-1/2)T passes through 0 or T), the firing-time map is non-differentiable (border collision) and the linearization misses the resulting stability change. The paper neither proves the event structure is locally rigid nor checks that the solutions stay away from these boundaries. Because the main stability conclusions and the claimed bifurcation set (symmetry-breaking at γ=1, saddle-node at γ=(n+1)/n or (2n+1)/(2n-1)) rest entirely on this equivalence, it is the load-bearing step of the central claim.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript studies two identical excitable theta neurons mutually coupled by delayed Dirac delta pulses, system (2)-(3). It constructs synchronous periodic solutions via the single-neuron existence condition (8), and alternating solutions via the substitution tau -> tau - T/2 leading to (29). For each family the authors linearize the firing-time return map and obtain characteristic equations Ds(lambda)=0 [Eq. (19)] and Da(lambda)=0 [Eq. (45)] in terms of a single parameter gamma. They conclude that both families are stable for 0<gamma<1, lose stability in a symmetry-breaking bifurcation at gamma=1 located at the minimum of T(tau), and undergo saddle-node bifurcations at gamma=(n+1)/n or gamma=(2n+1)/(2n-1). Symmetry-broken branches are constructed explicitly (Eqs. (25) and (53)) and shown unstable except for the neutrally stable tau=0 alternating continuum. The final section verifies the qualitative picture by numerical continuation for smooth pulses.","tokens_in":22962,"tokens_out":9031,"duration_ms":78494,"significance":"If the spectral reduction is accepted, the paper provides a complete, explicit two-parameter bifurcation description of two central periodic solution families of a canonical two-neuron network. The formulas are derived rather than fitted, and the smooth-pulse continuation gives falsifiable predictions about persistence and additional stability changes. The n <-> n+1/2 duality between synchronous and alternating solutions is elegant, and the tau=0 reversible family is well explained. The main caveat is that the equivalence between roots of the firing-time-map characteristic equation and Floquet multipliers of the infinite-dimensional impulsive DDE is asserted rather than proved; the paper's stability conclusions are conditional on that equivalence.","major_comments":[{"comment":"The paper states without proof that the roots of Ds(lambda) and Da(lambda) are the Floquet multipliers of the corresponding periodic solutions. System (2)-(3) is an impulsive delay differential equation whose phase space includes a history segment, so its monodromy operator is infinite-dimensional; the linearized firing-time map (13)-(16) and (38)-(39) only tracks event times. The authors need either a theorem or a precise reference showing that all multipliers except those computed from the event map are strictly inside the unit circle, and that the event count n and the order of events inside the delay window are locally rigid under small perturbations. Without this, the stability intervals, the symmetry-breaking bifurcation at gamma=1, and the saddle-node statements rest on an unproved reduction. This is the load-bearing step of the paper.","section":"Sec. 2.1 and Sec. 3.1.3 (Eqs. (19) and (45))"},{"comment":"The abstract and the discussion state that all symmetry-broken periodic solutions are unstable, but Appendix A.2 explicitly finds that the tau=0 family of symmetry-broken alternating solutions is neutrally stable: the characteristic equation (64) has lambda=1 as a double root, and Sec. 3.3 describes a one-parameter continuum. The claim should be qualified to 'all branches with tau>0' or 'all except the neutrally stable tau=0 family'; as written, the summary of results is not accurate.","section":"Abstract and Appendix A.2"},{"comment":"Proposition 2(4) proves only that no root of Da lies outside the unit circle for 0<gamma<1. Since Da always has the trivial root lambda=1, stability additionally requires that no nontrivial root lies on the unit circle. This can be fixed with a short argument: for |lambda|=1 and gamma>0, the modulus equation |lambda^{2n-1}||lambda-gamma|^2=(1-gamma)^2 forces |lambda-gamma|=1-gamma and hence lambda=1. As written, the proof is incomplete for the central stability claim.","section":"Sec. 3.1.3, Proposition 2(4)"}],"minor_comments":[{"comment":"The stability criterion should read 'all nontrivial multipliers lie inside the unit circle', because Ds and Da always contain the trivial multiplier lambda=1 corresponding to time translation; the quotation of [19, Proposition 1] should carry the same caveat.","section":"Sec. 2.1 and Sec. 3.1.3"},{"comment":"The sentence 'we find it more convenient to consider Da(lambda) in (19)' should refer to Eq. (45), not Eq. (19).","section":"Sec. 3.1.3"},{"comment":"The phrase 'find all periodic solutions of system (2)-(3)' overstates the scope, since only synchronous and alternating families are analyzed and Sec. 5 exhibits other stable periodic solutions in Fig. 12; suggest rephrasing to 'find all solutions of the considered symmetry classes'.","section":"Sec. 1"},{"comment":"The text contains numerous broken cross-references (e.g., 'Sec. 33.13.1.1', 'Sec. 22.1', 'Sec. 33.3', 'Appendix AA.1') and typos ('symmtry-broken', 'sytem', 'additonal', 'minumum'); these should be corrected before publication.","section":"Throughout"},{"comment":"The derivation divides by sinh(2 phi T), which implicitly assumes phi != 0; this is fine for the symmetry-broken branches but should be stated explicitly.","section":"Sec. 2.2, Eq. (24)"}],"recommendation":"major_revision","confidential_remarks":"The paper is a solid analytic study of a canonical pulse-coupled network, and the reported gap between event-map eigenvalues and Floquet multipliers is common in this literature; a careful proof or a precise reference would likely resolve it. The abstract overstatement about instability of symmetry-broken solutions should be corrected. If the spectral reduction is supplied, I would support acceptance."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a good paper. The genuinely new content is the transverse stability of synchronous periodic solutions (the g~s factor and Proposition 1), the alternating solution family (Sec. 3), and the instability proof for symmetry-broken branches (Appendix). Existence of synchronous branches is imported from your own [19], but that is an exact reduction, not a circular argument; alternating existence is obtained by the clean substitution tau -> tau - T/2. I re-derived Eqs. (8), (29), (19), and (45) and they are consistent. The stability regions and bifurcation set (symmetry-breaking at gamma=1, saddle-node at (n+1)/n or (2n+1)/(2n-1)) follow from the stated characteristic polynomials. No fitted parameters. The DDE-BIFTOOL continuation in Sec. 4 is a genuine check: for sharp pulses the stability of the delta-function case is reproduced, and the extra Hopf/other stability changes for m=5 are honestly reported.\n\nSoft spots, in order of importance.\n\nFirst, the linearised firing-time map is asserted to give Floquet multipliers of the impulsive DDE without proof. The stress-test note is right about the missing step. The event order and firing count are assumed fixed under perturbation, and no argument is given that the infinite-dimensional part of the spectrum is contractive. I don't think this sinks the paper; for these piecewise-analytic flows the finite-dimensional reduction is the standard and almost certainly correct object. But a referee should ask for a lemma (or a reference) stating the equivalence, and a check that the branches stay away from the borders tau=nT and tau=(n+1/2)T where the map is non-differentiable. The paper actually uses those borders to switch cases, so the solutions of interest are not on them, but this should be made explicit.\n\nSecond, the abstract and intro overclaim. 'We show that our approach ... can be used to find all periodic solutions' is contradicted by Fig. 12, where more complicated stable periodic solutions are shown. And 'all unstable' for symmetry-broken branches quietly excludes the neutrally stable tau=0 family. These are wording problems, but in a paper whose selling point is analytic completeness they matter.\n\nThird, minor: Proposition 2(2) says Ha has root 0 with multiplicity 2n-1 when n>=1. Fine. The proof of Proposition 1(1) has a small typo-looking inequality? Actually let's not manufacture. Keep minor.\n\nWho benefits: this is a benchmark for pulse-coupled excitable systems; anyone working on theta-neuron networks or delayed pulsatile coupling will cite it. It deserves a serious referee and, after the equivalence question is addressed and the claims are trimmed, I'd take it. Send to peer review.","headline":"Solid analytic extension of the single-neuron delay-coupled theta neuron results; the firing-map-to-Floquet equivalence is unproved but likely patchable, and the abstract overstates the scope.","tokens_in":23474,"tokens_out":2635,"would_cite":true,"duration_ms":26804,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["34K13","34K18","34C15","37C27"],"pacs":[],"model":"deepseek-v4-flash","headline":"For a pair of identical excitable theta neurons coupled by delayed pulses, this paper derives exact synchronous and alternating periodic solutions, proves that stability of each branch reduces to one parameter, and locates the…","keywords":["theta neuron","delayed pulsatile coupling","synchronous periodic solutions","alternating periodic solutions","spike-time map","symmetry-breaking bifurcation","saddle-node bifurcation","normal form"],"falsifier":"Compute the Floquet multipliers of the full discontinuous system by direct numerical integration with small pulse perturbations for a chosen branch, and compare them with the roots of the characteristic polynomial; the analytic criterion is wrong if any multiplier exits the unit circle at a parameter value different from the stated symmetry-breaking or saddle-node values, or if stability changes when a perturbed firing time crosses the edge of the delay window without a corresponding root crossing.","tokens_in":1585,"feed_emoji":"🧠","tokens_out":2112,"duration_ms":107928,"temperature":0.7,"pith_summary":"Two identical excitable theta neurons that kick each other with delayed Dirac-delta pulses are shown to support exact periodic firing rhythms of two types: perfectly synchronous and alternating. The paper derives closed-form existence equations for both types, and proves that stability of every branch is controlled by one positive number: all multipliers lie inside the unit circle in one range, and the first instability is a symmetry-breaking bifurcation at the minimum of the period as a function of delay. Branches of symmetry-broken solutions created at that point are proven unstable, with one neutrally stable family at zero delay. The result matters because it turns the simplest delay-coupled excitable network into a fully analytic normal form whose branch skeleton persists when the sharp pulses are replaced by smooth ones.","feed_headline":"Two delay-coupled theta neurons have exact periodic solutions","feed_subtitle":"Synchronous and alternating firing branches are solved in closed form; stability turns on a single number gamma.","key_machinery":"The carrying mechanism is the firing-time map: because the theta neuron integrates explicitly between pulses, the next firing time of each neuron is a closed-form function of earlier firing times. Linearising this map about a periodic solution gives characteristic polynomials whose roots are treated as the Floquet multipliers, and all dependence on coupling strength, delay, period, and firing count is condensed into the single parameter gamma. Factoring these polynomials separates dynamics within the symmetry subspace from transverse dynamics, which is what makes the symmetry-breaking bifurcation appear as a clean analytic condition.","core_discovery":"The paper establishes that, for coupling strength greater than 2, the delay-coupled pair has infinitely many branches of synchronous periodic solutions satisfying an exact coth equation and infinitely many branches of alternating periodic solutions satisfying the analogous half-integer equation, where the integer counts past firings inside the delay window; the primary synchronous branch has an explicit period formula. Stability of both families is governed by a single parameter: for values between 0 and 1 every Floquet multiplier is inside the unit circle, at one it crosses at the trivial multiplier, which is a symmetry-breaking bifurcation at the minimum of the period-versus-delay curve, and it exits at stated saddle-node values for higher parameter values. The alternating branch with zero firings in the delay window is stable for all positive delays and persists at zero delay, where it coexists with a one-parameter family of neutrally stable symmetry-broken solutions; all other symmetry-broken branches are unstable. Numerical continuation with smooth pulse coupling reproduces the same branch geometry, with extra stability changes appearing only when the pulse is wide.","pith_inferences":["Extension: the stability criterion predicts a robust experimental fingerprint: any realisation of two delay-coupled excitable elements with pulsatile coupling should lose its in-phase or anti-phase rhythm exactly at the turning point of the period-versus-delay curve, independent of pulse shape.","Extension: because the existence equations are pure coth identities, the same branch skeleton should hold for any excitable unit whose pulse-triggered phase advance approximates the tangent shift of the saddle-node-on-invariant-circle normal form, such as class-1 neurons with suitable phase-resetting curves.","Extension: the paper mentions more complex periodic solutions with multiple firings per neuron per period; the same spike-time-map machinery should be able to derive their existence and stability, although the event order inside the delay window is no longer fixed.","Extension: for wide smooth pulses, the extra stability changes invite the question of whether they are always Hopf bifurcations and whether a pulse-width perturbation expansion around the delta-function limit could predict them."],"forward_implications":["On any branch with at least one firing in the delay window, the stable region is exactly the part of the period-versus-delay curve to the right of its minimum, so the minimum period doubles as a symmetry-breaking boundary.","The synchronous and alternating families interleave: substituting a half-integer shift turns the synchronous existence equation and characteristic equation into the alternating ones, so the two rhythm types form one analytic structure.","In the strong-coupling limit the system becomes an infinite fan of neutrally stable rays, with synchronous periods and alternating periods scaling as fixed fractions of the delay, so all nontrivial stability in this model comes from finite coupling strength.","All symmetry-broken solutions that branch from the synchronous and alternating families are unstable except the zero-delay alternating family, which is neutrally stable, so the network cannot settle into a slightly desynchronised periodic state.","The Dirac-delta system acts as a normal form: with smooth pulse coupling of sharp width the computed branches and stability agree with the analytics, and only wider pulses introduce additional bifurcations such as Hopf points."],"supporting_citations":[{"why":"Gives the explicit theta-neuron integration, the positivity of the stability parameter, and the root properties used for the synchronous case; the synchronous existence equation is the self-coupling result.","marker":"[19]"},{"why":"Supplies the reappearance principle used to build secondary synchronous branches and the symmetry-broken alternating branches for positive delay.","marker":"[36]"},{"why":"Provides the linearised firing-time-map stability analysis for time-delay coupled pulse oscillators that the present derivation follows.","marker":"[14]"},{"why":"Provides the numerical continuation software used to verify that smooth pulse coupling preserves the branch structure and stability predictions for sharp pulses.","marker":"[29]"},{"why":"Identifies the theta neuron as the normal form of the saddle-node-on-invariant-circle bifurcation, which justifies the model equations.","marker":"[7]"}],"fun_headline_variants":["Exact solutions for two delay-coupled theta neurons","Infinitely many sync and anti-sync solutions for theta pair","Closed-form periodic solutions from delay-coupled theta neurons","Stability of theta neuron pair set by one parameter","Delay-coupled theta neurons solved for sync and anti-sync"],"cache_read_input_tokens":25600,"weakest_assumption_plain":"The stability analysis assumes that stability of the periodic orbit of the discontinuous system is exactly the stability of the linearised firing-time map, and that small perturbations never change the number of firings inside the delay window or the order in which pulses arrive; if a perturbation makes a firing cross the boundary of the delay interval, the analytic stability prediction could miss a border-collision change of stability.","fun_headline_variants_meta":{"raw":{"variants":["Exact solutions for two delay-coupled theta neurons","Infinitely many sync and anti-sync solutions for theta pair","Closed-form periodic solutions from delay-coupled theta neurons","Stability of theta neuron pair set by one parameter","Delay-coupled theta neurons solved for sync and anti-sync"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000842,"raw_usage":{"total_tokens":3668,"prompt_tokens":945,"completion_tokens":2723,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":561,"completion_tokens_details":{"reasoning_tokens":2646}},"tokens_in":561,"tokens_out":2723,"duration_ms":18640,"temperature":1.0,"reasoning_tokens":2646,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T13:41:08.650723+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the Floquet multipliers of the full discontinuous system by direct numerical integration with small pulse perturbations for a chosen branch, and compare them with the roots of the characteristic polynomial; the analytic criterion is wrong if any multiplier exits the unit circle at a parameter value different from the stated symmetry-breaking or saddle-node values, or if stability changes when a perturbed firing time crosses the edge of the delay window without a corresponding root crossing.","supporting_citations":[{"cited_title":"Theta neuron subject to delayed feedback: a prototypical model for self-sustained pulsing","cited_arxiv_id":null,"evidence_quote":"Gives the explicit theta-neuron integration, the positivity of the stability parameter, and the root properties used for the synchronous case; the synchronous existence equation is the self-coupling result."},{"cited_title":"Delay and periodicity.Phys Rev E, 79(4):046221, 2009","cited_arxiv_id":null,"evidence_quote":"Supplies the reappearance principle used to build secondary synchronous branches and the symmetry-broken alternating branches for positive delay."},{"cited_title":"Synchronization of time-delay coupled pulse oscillators","cited_arxiv_id":null,"evidence_quote":"Provides the linearised firing-time-map stability analysis for time-delay coupled pulse oscillators that the present derivation follows."},{"cited_title":"Parabolic bursting in an excitable system coupled with a slow oscillation","cited_arxiv_id":null,"evidence_quote":"Identifies the theta neuron as the normal form of the saddle-node-on-invariant-circle bifurcation, which justifies the model equations."}],"review_version":1}