{"id":"cb985ab9-3231-4748-8f80-3a027b59d1de","arxiv_id":"2607.20699","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A rigorous spectral theory for Fokker-Planck population dynamics of spiking neurons with absolute refractory period is derived, including exact transfer functions and defective eigenvalues.","lead":"This paper develops a rigorous mathematical framework for networks of spiking neurons that includes the brief absolute refractory period after each spike, using operator theory to characterize the system's spectrum and exact response to input. It matters because it puts widely used spectral decomposition methods on solid footing and shows how refractoriness alone can push a network into rhythmic firing.","discovery_kind":"first_principles","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 3.3 hinges on Lemma 3.1, whose proof is only a citation to [28] and never verifies Bγ is regular; without that, Eq. (3.13) may not capture σ(Tγ).","rationale":"The paper's mathematical core is Theorem 3.3, and its proof consists of applying Lemma 3.1. Every subsequent result—transfer-function poles (Section 4), the defective-eigenvalue condition (3.20), and the Hopf criterion (4.15)—uses the identification σ(Tγ) = zeros of ∆γ. Lemma 3.1 is not proved; it cites two results from [28]. The authors construct Zγ and a right inverse Qγ, but they never check that Bγ satisfies the regularity assumptions under which the global equivalence (3.8) is valid. This is a genuine gap in exposition if not necessarily in truth: for boundary eigenvalue problems with nonlocal transmission conditions, regularity is not automatic. A direct verification—analytic or numerical—is therefore the minimal condition for the central claim to be considered rigorous. I do not see a demonstrated mathematical error elsewhere; the omitted determinant computation in Theorem 4.3 and the abstract's overstatement of the limit-cycle result are secondary. The reader's CONDITIONAL verdict is appropriate: the concern does not warrant rejection, but it does require an explicit check or a completed proof before the spectral decomposition can be taken as fully established.","tokens_in":21930,"tokens_out":27011,"duration_ms":200275,"concrete_test":"Use a high-order pseudospectral discretization of Tγ (Eqs. 2.5–2.11) for the LIF model with τ0=0.1 and the parameters of Fig. 3; impose the five boundary conditions exactly. Compute eigenvalues in Re λ≥-20, |Im λ|≤40 and compare with zeros of ∆γ(λ) from (3.13) obtained via the hypergeometric representation of f1. A match in location and algebraic multiplicity confirms the operative content of Lemma 3.1; any mismatch shows the global equivalence fails. Also verify Bγ|ker eT_D(λ) has constant rank 5 away from eigenvalues and that the right inverse Qγ(λ) from Remark 3.2 is holomorphic on W_p.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central spectral characterization is Theorem 3.3, and its proof depends entirely on Lemma 3.1, which asserts global equivalence eTγ(λ) = Cγ(λ) diag(Mγ(λ), Id_{Lp}) Dγ(λ) with invertible Cγ, Dγ (Eq. 3.8). The lemma is dismissed as 'a restatement of Theorem 1.11.1 and Lemma 1.11.2 in [28]', but the hypotheses of those results are not verified. In particular, the boundary operator Bγ in (2.11) involves non-local transmission conditions coupling p(θ) to pr(0) and p(H±) to pr(τ0). Regularity in the sense of [28] is not automatic for such non-self-adjoint boundary eigenvalue problems. If Bγ is not regular, the zeros of det Mγ need not coincide with σ(Tγ), and the algebraic/geometric multiplicities—used for the defective-eigenvalue condition (3.20), the biorthogonal root-function construction (Lemma 3.4), and the transfer-function pole condition (4.15)—could be misrepresented. Thus the whole spectral decomposition, and every downstream result, rests on an unverified external theorem.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops a spectral theory for the population-density Fokker-Planck operator of integrate-and-fire neurons with an absolute refractory period. The state space is augmented by a refractory-age density, and the evolution operator is formulated as a non-self-adjoint boundary eigenvalue problem on three compartments. The main claim (Theorem 3.3) is that the spectrum is purely discrete and coincides with the zeros of a characteristic function Δγ(λ) (Eq. 3.14), that 0 is always a simple eigenvalue, and that every eigenvalue has geometric multiplicity one. A transfer function is derived from the resolvent, including new boundary-modulation terms, and the theory is applied to linear stability of recurrent networks. The paper also identifies defective eigenvalues as exceptional points and presents numerical evidence that refractoriness can promote limit cycles.","tokens_in":22159,"tokens_out":7322,"duration_ms":62991,"significance":"If the central spectral theorem holds, this is a substantial contribution: it would put spectral decomposition methods for population density dynamics on a rigorous footing for arbitrary finite refractory periods, and it yields an explicit characteristic equation that reduces correctly to known results in the τ0→0 limit. The paper is self-contained from the FP model, introduces no free parameters, and makes falsifiable predictions about the location of eigenvalues and the structure of the transfer function. The treatment of defective eigenvalues and the corrected boundary contributions to the transfer function are also valuable. The numerical checks are consistent with the analytic expressions and the authors are candid about the non-rigorous nature of the nonlinear bifurcation claim.","major_comments":[{"comment":"The proof of the central spectral characterization is delegated to [28] without verifying the hypotheses. The boundary operator Bγ in Eq. (2.11) contains nonlocal transmission conditions (BC3, BC4) coupling p(θ) to p_r(0) and p(H±) to p_r(τ0). Regularity in the sense of [28] is not automatic for such non-self-adjoint boundary eigenvalue problems. Please verify the regularity assumptions of [28, Thm. 1.11.1 and Lemma 1.11.2] for Bγ, or state and prove them. In particular, show that the global equivalence (3.8) holds with invertible Cγ(λ), Dγ(λ). Without this, Theorem 3.3's identification of σ(Tγ) with the zeros of Δγ, and the algebraic/geometric multiplicity statements used in Lemma 3.4 and in Eq. (4.15), are not established.","section":"Lemma 3.1 and Theorem 3.3 (Eq. 3.8, Eq. 3.13)"},{"comment":"The proof that every eigenvalue has geometric multiplicity one is compressed: 'if some λ_n had geometric multiplicity all minors of Mγ(λn) would vanish, forcing f1(θ,λn)=f1(H,λn)=0; the boundary conditions then give f2(θ,λn)=f2(H,λn)=0'. This step is not immediate for a 5×5 characteristic matrix with the nonlocal structure of (3.12). Please provide the explicit minor computation, since this uniqueness is used in the root-function construction (Lemma 3.4) and in the normalization of the eigenfunctions.","section":"Theorem 3.3, geometric multiplicity argument"},{"comment":"The abstract states that refractoriness 'can facilitate the onset of limit cycles, that is, stable oscillations in the firing rate'. However, Section 4.3 explicitly says: 'the spectral analysis proves only the loss of linear stability of the fixed point; the emergence of a stable limit cycle is verified numerically and not proven here, as this would require a nonlinear bifurcation analysis beyond the present scope.' The abstract should be rephrased to distinguish the rigorous Hopf linear-stability result from the numerical observation of stable oscillations.","section":"Section 4.3 and Abstract"}],"minor_comments":[{"comment":"The second equation is written as ∂_t p^r_t(τ) = −∂_τ p^r_t(τ,t), but the argument (τ,t) is inconsistent with p^r_t(τ).","section":"Eq. (2.3)"},{"comment":"The reference to the Lumer-Phillips theorem is misspelled as '[13, Therem 3.15]'. Also, the authors should mention explicitly that they are using the spectrum result of Proposition 3.9 to obtain the range condition required by Lumer-Phillips.","section":"Section 3.4"},{"comment":"The displayed form of an eigenfunction for eigenvalue iω writes the refractory component as exp(iτ ω), while the construction in Eq. (3.17) gives exp(−iτ s). The subsequent integral uses cos(n2πτ/τ0), which is consistent with the latter convention; please fix the sign for clarity.","section":"Proposition 3.9 proof (Appendix B)"},{"comment":"The paper uses the phrase 'complete spectral characterization' in the abstract, but the discussion lists 'the rigorous proof of the completeness of the eigenbasis' as an open problem. Consider softening 'complete' to 'explicit' or 'full point-spectrum' to avoid overstating what is proven.","section":"Discussion, completeness"},{"comment":"There are some typos, e.g., 'redunction' in Section 3.2 and 'spiking neuron networks' in reference [27]. A careful proofreading pass is recommended.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The most important issue is whether the regularity conditions of [28] hold for Bγ. If they do, the paper is likely acceptable; if not, the spectral characterization may need to be weakened or made conditional. The limit-cycle claim in the abstract should be adjusted regardless. I see no novelty-disclosure concern: prior work by the same group is cited as a baseline and corrected, not used circularly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This paper deserves a serious referee. It does something real: it extends the population-density approach to include an absolute refractory period by adding a refractory compartment, derives a characteristic equation for the full operator, and gives a complete spectral characterization for τ0>0. The defective-eigenvalue analysis at real-to-complex transitions is genuinely new; previous work implicitly assumed simple eigenvalues. The transfer function derivation is explicit and corrects earlier heuristic treatments, and the τ0→0 limit reduces to known equations, which is a good sanity check.\n\nThe main theorem, however, depends on Lemma 3.1, which asserts that the boundary eigenvalue problem is globally equivalent to a 5×5 characteristic matrix via the Mennicken–Möller framework. The proof is a citation, and the paper does not verify that the boundary operator Bγ is regular in the sense of [28]. For a non-self-adjoint problem with non-local transmission conditions, that is not a trivial check. If regularity fails, the zero set of the characteristic determinant might not capture the whole spectrum, and the multiplicity data—used for the defective-eigenvalue condition, the root-function construction, and the transfer-function pole structure—could be wrong. I would want that hypothesis checked before betting on the theorem. This is a standard kind of gap, but it is load-bearing.\n\nThere is also a mismatch between the abstract and the body. The abstract says refractoriness 'facilitate[s] the onset of limit cycles'; the body explicitly says only loss of linear stability is proven, and stable limit cycles are verified numerically. That should be fixed.\n\nMinor issues: the determinant computation in Theorem 4.3 is left at 'computing this matrix determinant'; there is no code for the numerical figures; and the boundary-term corrections to the transfer function are motivated but not tested against direct simulation. None of these are fatal.\n\nWho is this for? Anyone using spectral decompositions of Fokker–Planck models in computational neuroscience, and mathematicians working on non-self-adjoint boundary eigenvalue problems. If the regularity hypothesis checks out, this is a solid contribution. I'd recommend sending it to peer review with a request to verify the hypotheses behind Lemma 3.1, soften the abstract, and expand the determinant step.","headline":"Serious spectral-theory paper with a real result on refractoriness, but the central theorem leans on an unverified regularity assumption and the abstract overclaims the limit-cycle result.","tokens_in":22713,"tokens_out":2090,"would_cite":true,"duration_ms":18016,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["47A10","35P05","47D06","92C20"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper establishes a complete spectral theory for the Fokker-Planck operator governing spiking neural populations with a finite absolute refractory period, showing that the spectrum is discrete and generated by a single characteristic e","keywords":["population density dynamics","spiking neurons","absolute refractory period","Fokker-Planck operator","non-self-adjoint spectral theory","boundary eigenvalue problem","linear response transfer function","limit cycles"],"falsifier":"Discretize Tγ for the LIF model with τ0>0 (finite-element or spectral method on the three compartments) and compute the largest eigenvalues; compare with zeros of Δγ(λ). Finding a spectral eigenvalue that is not a zero of Δγ, or a zero of Δγ with no matching eigenvalue, would refute Theorem 3.3. Also check whether Δγ'(0) can vanish for some admissible γ, which would contradict the asserted simplicity of the zero eigenvalue.","tokens_in":21743,"feed_emoji":"🧠","tokens_out":3914,"duration_ms":35253,"temperature":0.7,"pith_summary":"This paper proves that population density dynamics of spiking neurons with an absolute refractory period can be studied rigorously as a non-self-adjoint boundary eigenvalue problem. By adding a refractory compartment that stores the past firing rate, the dynamics become Markovian and the generator is shown to have a purely discrete spectrum: every eigenvalue solves an explicit characteristic equation, zero is always a simple eigenvalue, and every eigenspace is one-dimensional. This justifies spectral decomposition methods that were previously used heuristically, and yields an exact linear-response transfer function including new boundary-modulated terms. As an application, the paper shows that increasing refractoriness can drive a stable population into self-sustained limit-cycle oscillations by moving diffusion poles across the imaginary axis.","feed_headline":"Rigorous spectrum for spiking populations with refractoriness","feed_subtitle":"Augmented state space turns the refractory period into a tractable boundary problem; transfer function becomes exact.","key_machinery":"The central object is the evolution operator Tγ acting on L^p of three compartments: (α,H), (H,θ) and the refractory interval (0,τ0), with five linear boundary conditions coupling the membrane potential density to the stored firing-rate history. The argument runs through the theory of non-self-adjoint boundary eigenvalue problems: a holomorphic operator function eTγ(λ) is shown to be Fredholm and globally equivalent to a 5×5 characteristic matrix Mγ(λ); the determinant of that matrix divided by Wronskians yields the scalar characteristic equation Δγ(λ)=0 that completely determines the spectrum. Holomorphic root functions built from the adjoint fundamental solutions supply the eigenfunctions","core_discovery":"The central result is Theorem 3.3: for fixed input moments, the evolution operator Tγ has a discrete spectrum given by the zeros of Δγ(λ)=f1(θ,λ)/wγ(θ)−e^{−τ0λ}f1(H,λ)/wγ(H), with 0 always a simple eigenvalue and all eigenvalues of geometric multiplicity one. The characteristic equation emerges from the determinant of a 5×5 characteristic matrix built from fundamental solutions of the Fokker-Planck operator and an exponential factor encoding the refractory delay. Because the boundary eigenvalue operator function is globally equivalent to this matrix, eigenvalues, algebraic multiplicities, and generalized eigenfunctions of the operator are read off from the matrix. This converts previously he","pith_inferences":["If the boundary regularity hypothesis used to invoke the abstract boundary-eigenvalue machinery fails for some parameter range, the characteristic equation would not necessarily capture the full spectrum; this is a concrete target for numerical verification.","The new boundary-modulated terms in the transfer function should also appear in conductance-based neuron models where noise is state-dependent; the paper suggests but does not demonstrate this.","A natural next step is to prove completeness of the biorthogonal eigenbasis via resolvent asymptotics; if achieved, the spectral decomposition would become a fully rigorous replacement for time-stepping simulations in these networks.","The Hopf-bifurcation scenario for excitatory populations suggests that the refractory period is a singular perturbation that can change network state; testing whether this persists in finite-size networks would check robustness."],"forward_implications":["Spectral mode expansions of population activity for neurons with τ0>0 are justified, placing low-dimensional firing-rate models on a rigorous footing.","The exact transfer function reveals two previously missed contributions: noise-modulation at threshold and the presence of the reflecting barrier at α.","Defective eigenvalues are identified as exceptional points where relaxational modes coalesce into oscillatory modes; the full Jordan-chain description is required near them.","In recurrent networks with delay, increasing the refractory period can destabilize the fixed point via diffusion poles, producing limit-cycle firing rates at realistic parameters without transmission delays."],"fun_headline_variants":["Spectral theory for refractory spiking populations","Exact transfer function for spiking populations with refractoriness","Refractoriness can trigger oscillations in spiking networks","Rigorous spectrum reveals exceptional points in neuronal populations"],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The proof leans on an abstract regularity condition for the boundary conditions (from the theory of non-self-adjoint boundary eigenvalue problems) that the paper invokes without verifying explicitly.","fun_headline_variants_meta":{"raw":{"variants":["Spectral theory for refractory spiking populations","Exact transfer function for spiking populations with refractoriness","Refractoriness can trigger oscillations in spiking networks","Rigorous spectrum reveals exceptional points in neuronal populations"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000661,"raw_usage":{"total_tokens":2854,"prompt_tokens":738,"completion_tokens":2116,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":482,"completion_tokens_details":{"reasoning_tokens":2053}},"tokens_in":482,"tokens_out":2116,"duration_ms":15139,"temperature":1.0,"reasoning_tokens":2053,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-01T09:38:13.426302+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Discretize Tγ for the LIF model with τ0>0 (finite-element or spectral method on the three compartments) and compute the largest eigenvalues; compare with zeros of Δγ(λ). Finding a spectral eigenvalue that is not a zero of Δγ, or a zero of Δγ with no matching eigenvalue, would refute Theorem 3.3. Also check whether Δγ'(0) can vanish for some admissible γ, which would contradict the asserted simplicity of the zero eigenvalue.","supporting_citations":[],"review_version":1}