{"id":"d2a69421-9b90-43ce-ad3e-7f0a74175e86","arxiv_id":"2501.05576","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"A two-stage Hopf-oscillator model shows that detecting distortion-product frequencies instead of primary tones sharpens frequency selectivity, while cascading active detectors restores sensitivity, matching the architecture of the mosquito ear.","lead":"This paper uses a simple oscillator model to show that detecting 'distortion product' tones, rather than the original sound frequencies, can sharply improve frequency selectivity, and that chaining several such detectors can recover lost sensitivity. The findings suggest a general design principle for sensitive, sharply tuned hearing and may explain how male mosquitoes identify females in noisy swarms.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claimed detection advantage rests on deterministic, noise-free quality factors; QDP diverges as forcing vanishes while the DP response itself vanishes, so the gains may not survive in a noisy swarm.","rationale":"I read the paper in good faith and verified that the main quantitative results are derivable from the stated model: the effective-forcing reduction leading to Eq. 19 is algebraically coherent, the cascade simulations are plausible, and the paper does not rely on fitting to claim the QDP/QPT enhancement. The reader's weakest_assumption identifies the same load-bearing issue I find: the entire signal-detection argument is made in a deterministic setting, yet the stated biological context is a noisy swarm. My concern is not that the model is internally inconsistent, but that the central claim — that DP tuning and cascading yield 'immense benefits' for acoustic detection — is extrapolated from phase-locked response amplitudes and quality factors that diverge precisely in the limit where the response amplitude vanishes. A noise floor truncates the diverging selectivity and may erase the sensitivity advantage, because the DP signal is generated at order Ff^2 and is necessarily weaker than the primary-tone response. Since the reader already conditioned acceptance on adding noise and SNR analysis, I do not recommend changing the verdict: it should remain CONDITIONAL. The concrete test I propose — injecting noise into Eqs. (6)–(7) and computing SNR rather than deterministic response amplitude — would directly decide whether the concern lands. I deliberately did not raise objections to the absence of code or experimental data, because those are secondary to the logical gap between response amplitude and detection performance.","tokens_in":3,"tokens_out":7965,"duration_ms":206969,"concrete_test":"Augment Eqs. (6)–(7) with additive complex white noises η1(t), η2(t) of intensity D, e.g. ⟨ηi(t)ηi*(t′)⟩ = 2D δ(t − t′), and compute the signal-to-noise ratio for detecting the female tone at the relevant readout frequency (ωf for the PT detector, 2ωf − ωm for the DP detector, and 2ωf − ωm for the composite) from the phase-locked amplitude divided by the noise variance in that Fourier bin. Sweep D from 0 up to amplitudes comparable to Ff (or to measured mosquito swarm noise levels). If the composite and DP detectors cease to outperform the PT detector in SNR before D reaches biologically plausible values, the noise-free QDP / QPT advantage does not translate into a detection benefit. A simpler analytic check: compare Rdp ≈ Ff^2 Fm / (2(ωm − ωf)^3) against the noise floor; if Rdp falls below the floor at forcing levels where Eq. 21 predicts large QDP/QPT, the central claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that tuning a Hopf detector to a distortion product, and cascading such detectors, yields signal-detection benefits, with the strongest quantitative support being the quality-factor ratio QDP/QPT = (ωm − ωf)^2 / ((1/2 Ff Fm)^{2/3}) (Eq. 21) and the composite-model curves in Fig. 4. Every metric reported — phase-locked response amplitude, linear response function, and quality factor — is computed from the deterministic equations (1), (4)–(7), which contain no stochastic term. The abstract and discussion explicitly motivate the scheme as enabling a male mosquito to 'identify and pursue a particular female within a noisy swarm environment,' but no noise floor or signal-to-noise ratio is ever introduced. This matters because the selectivity enhancement is not independent of signal amplitude: the DP response amplitude grows as roughly Ff^2 (Eq. 19 and Fig. 2B), while QDP diverges as (FfFm)^{-2/3}. In any physical detector, thermal noise, intrinsic oscillator noise, or swarm-generated acoustic noise sets a minimum detectable phase-locked amplitude; once Rdp falls below that floor, the tuning curve cannot become arbitrarily narrow and the 'orders of magnitude' improvement in Q is not a real detection advantage. In addition, Ff and Fm are dimensionless and uncalibrated; the impressive ratios in Eq. 21 can be made arbitrarily large by decreasing the forcing amplitudes, but the corresponding DP signals become arbitrarily small. The model-level math is internally consistent, but the biological claim of 'immense benefits' for detection is supported only in the zero-noise, arbitrarily-weak-stimulus limit. The paper notes future experimental work will need to test the gains directly, but the Discussion nevertheless states the benefits as if established. Without noise analysis, the central claim is conditional at best.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes a generic model of mosquito auditory detection based on Hopf oscillators, with two main ingredients: (i) tuning a detector to a distortion product (e.g., 2ω_f − ω_m) rather than to the primary tone, and (ii) cascading the response of one oscillator into a second oscillator. Using deterministic simulations and analytic approximations, the authors find that distortion-product-tuned detectors exhibit greatly increased quality factors (Eq. 20–21), that cascades improve weak-signal amplification, compression, and frequency selectivity (Fig. 3), and that a two-oscillator model combining both features outperforms either oscillator alone on all three metrics (Fig. 4). The work is motivated by experimental observations that mosquito auditory neurons are tuned to distortion products rather than to the female wingbeat frequency.","tokens_in":11729,"tokens_out":15225,"duration_ms":128633,"significance":"If the central claims hold, the paper identifies a plausible general design principle for nonlinear auditory detection: distortion-product tuning sharpens frequency selectivity, and cascading active elements boosts weak-signal sensitivity. The analytic results, especially Eq. (21), are parameter-free in the sense that no fitted parameters are introduced; the enhancement is a direct consequence of the Hopf nonlinearity. The model is simple, reproducible, and makes explicit predictions (e.g., increasing quality factor with decreasing stimulus amplitude and with distortion-product order) that could be tested experimentally. A notable strength is the transparent appendix derivation of the single-tone quality factor and the effective forcing for the distortion-product response. However, the significance is limited by the complete absence of noise in the model, despite the stated motivation of detection in a noisy swarm environment; all reported measures are deterministic quantities, so the claimed detection advantages are not yet tied to signal-to-noise ratios or detection thresholds.","major_comments":[{"comment":"The central claim of improved signal detection is not supported by a noise analysis. All reported metrics — response amplitude, linear response function χ, and quality factor — are computed from the deterministic equations (1), (4)–(7), which contain no stochastic term. Yet the abstract and Discussion motivate the work as enabling detection \"within a noisy swarm environment.\" This matters because the distortion-product response amplitude grows as roughly Ff^2 (Eq. 19), while the claimed quality-factor enhancement diverges as (Ff Fm)^{-2/3} (Eq. 21); as the forcing tends to zero, the response and the quality factor move in opposite directions. In any physical detector a noise floor sets a minimum detectable phase-locked amplitude, and once the response falls below that floor the divergent Q is not a usable detection advantage. Please either add a noise analysis (e.g., additive or oscillator-intrinsic noise) demonstrating that the enhancement survives when operating above threshold, or explicitly restrict the claims to the noiseless limit and revise the motivating statements accordingly.","section":"Abstract and Discussion; Eqs. (1), (7), (19), (21)"},{"comment":"The derivation of Eq. (21) is not sufficiently transparent regarding the definition of QDP and its conversion to the female-stimulus-frequency domain. The text states that a factor of 2 gained from the derivative dδω/dωf = 2 cancels a factor of 2 lost from the detector being tuned to half the primary-tone frequency, but this cancellation is not immediate: the center of the tuning curve in terms of ωf is (ω0 + ωm)/2, not simply ω0, so the effective quality factor with respect to ωf may differ from QDP by a factor that depends on ωm/ω0. Please provide a step-by-step derivation of Eq. (21), including the precise definitions of QPT and QDP and the explicit conversion from δω to ωf, so that the quantitative claim can be verified.","section":"Appendix, Eqs. (20)–(21)"},{"comment":"The comparison in Fig. 4 may not isolate the effect of cascading. In the composite model the second oscillator receives the full nonlinear output z1(t) of the first oscillator, which contains primary tones, harmonics, and distortion products generated by the first stage; the \"second oscillator alone\" baseline is presumably driven directly by the two-tone stimulus. These inputs differ not only in amplitude but also in spectral content. Please specify the exact driving conditions for the single-oscillator baselines, and ideally drive the baselines with a signal of the same spectral composition (e.g., the linear response of the first stage) so that the reported improvement can be attributed to the cascade rather than to differences in the input spectrum.","section":"Results, Fig. 4"}],"minor_comments":[{"comment":"The text reads \"chemicals such as DTT\" — this should be \"DDT.\"","section":"Page 1, Introduction"},{"comment":"The Results state that cascading enhances frequency selectivity, while the Discussion says cascading \"negligibly affect[s] the quality factor.\" If these refer to different metrics (e.g., half-maximum width versus tail suppression), please define both explicitly to avoid the appearance of contradiction.","section":"Results, page 5 and Discussion, page 7"},{"comment":"The statement that distortion products lower than ω1 can be represented as ω_{p,q} = ω1 − (1/2)(|p|+|q|−1)Δω holds only for the branch p = q+1 (e.g., 2ω1−ω2, 3ω1−2ω2), not for all p and q. Please clarify the restriction.","section":"Page 3, Eq. (2)"},{"comment":"Please list the full parameter set for each panel in the caption; currently panel (D) uses F1 = F2 = 0.1 while panels (A) and (B) use F2 = 5, which is easy to miss.","section":"Fig. 2 caption"}],"recommendation":"major_revision","confidential_remarks":"The paper is clearly written and the analytic appendix is a genuine strength; the derivations are reproducible and the central mathematical structure is internally consistent. The main concern is the gap between the deterministic model and the noisy-swarm motivation. If the authors add a noise analysis or significantly temper the motivational wording, the paper could be published. The novelty is incremental relative to existing Hopf-oscillator models of hearing, but the specific focus on distortion-product tuning is a new and interesting angle. I found no citation or attribution issues."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: the math is honest and the core observation is worth having — a Hopf detector tuned to a distortion product has a quality factor that, for fixed forcing, can be orders of magnitude above the primary-tone detector, and cascading such elements compresses strong inputs. The appendix derivation is solid; I checked the steps and the QDP/QPT ratio follows from the stated weak-forcing, near-bifurcation approximations. The exact single-tone width also checks out. Credit where due: the model clearly separates the two ideas, gives a clean scaling law for the cascade (F power 1/3^j), and the parameters are all specified in the text. This is not a fitted model; the comparison is derived, not tuned.\n\nThe soft spots are the ones the stress test flags. Every metric — response amplitudes, linear response, quality factor — is computed from a deterministic equation with no stochastic term. The text motivates the whole thing as detecting a female 'within a noisy swarm environment,' but there is no noise floor anywhere. That matters because the quality-factor divergence is at small forcing: QDP/QPT grows as (FfFm)^(-2/3), but the DP response itself grows as Ff^2, so at the forcing levels where the predicted Q-ratio is huge, the actual signal is tiny. Until you put in a noise floor — thermal, mechanical, or acoustic — you cannot claim the selectivity gain translates into a detection advantage in a swarm. The authors do say future experiments must test the gains, but the abstract and discussion say 'immense benefits' as if the deterministic result already settles it. That overstatement should be toned down.\n\nMinor: no code or data, but the equations and parameters are given, so reproduction is straightforward. The ratio ωm/ωf ≈ 1.5 is taken from experiment, which is fine; the results don't depend sensitively on it.\n\nVerdict: the model-level framework is sound and novel enough to deserve a referee. The main revision the paper needs is a noise analysis — at minimum a simple sensitivity calculation with a representative noise spectral density — and a rewrite of the claims to match what the model actually shows. I'd send it out; the findings are worth having even if the biological conclusion ends up being weaker.","headline":"A clean, internally consistent Hopf-oscillator analysis of distortion-product-tuned ears; the model-level gains are real, but the 'noisy swarm' headline claim needs a noise floor before it holds.","tokens_in":12441,"tokens_out":2040,"would_cite":true,"duration_ms":20912,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A Hopf-oscillator detector tuned to a distortion product instead of the primary tone achieves sharply enhanced frequency selectivity, and cascading such amplifiers adds weak-signal sensitivity and compression, so the combined scheme…","keywords":["mosquito hearing","distortion products","Hopf oscillator","frequency selectivity","signal cascading","active auditory amplification","two-tone interference","quality factor"],"falsifier":"Add noise of realistic amplitude to Eqs. (1), (6), and (7) and measure the phase-locked response at $2\\omega_f - \\omega_m$: if intrinsic oscillator fluctuations or swarm noise wash out the sharpened tuning curve and the $Q_{\\mathrm{DP}}/Q_{\\mathrm{PT}}$ scaling of Eq. (21), the predicted detection advantage does not survive. A complementary experiment would record Johnston's organ neuron responses to two-tone stimuli and test whether neural tuning width narrows roughly as $F_f^{-2/3}$ as the female-tone amplitude drops.","tokens_in":11237,"feed_emoji":"🦟","tokens_out":9201,"duration_ms":79656,"temperature":0.7,"pith_summary":"The paper argues that an acoustic detector can sharpen its frequency selectivity by locking onto a distortion product—a combination tone generated by nonlinearity—rather than the primary tone itself. Working with the normal form of the Hopf oscillator, it derives the quality-factor ratio $Q_{\\mathrm{DP}}/Q_{\\mathrm{PT}} = (\\omega_m - \\omega_f)^2 / (\\tfrac{1}{2} F_f F_m)^{2/3}$ and shows the advantage grows without bound for weak stimuli. It then shows that cascading several oscillator stages—feeding the response of one into the next—enhances weak-signal sensitivity, compresses large signals, and narrows the response band. When both schemes are combined into a two-oscillator model of the male mosquito's flagellum and antennal sensory neurons, the composite detector outperforms either stage alone on all three metrics. If the model is right, it identifies a general design principle: active nonlinear detectors can profit from being tuned away from the stimulus frequency.","feed_headline":"Phantom-tone tuning sharply boosts frequency selectivity","feed_subtitle":"A two-stage Hopf model shows why male mosquitoes may tune to distortion products, not the wingbeat.","key_machinery":"The central object is the normal form equation for the supercritical Hopf bifurcation, $dz/dt = (\\mu + i\\omega_0)z - |z|^2z + \\mathrm{forcing}$, a canonical equation for an oscillator that begins to self-oscillate when the control parameter $\\mu$ crosses zero. The cubic term generates odd-order distortion products with amplitudes that fall off with order, and the identity $d\\omega_{p,q}/d\\omega_1 = p$ means frequency modulations of a primary tone are magnified at the distortion-product frequency. For distortion-product tuning, inserting a two-tone ansatz yields an effective equation of the same form with $F_{\\mathrm{eff}} = R_f^2 R_m$ and $\\mu_{\\mathrm{eff}} = \\mu - 2(R_f^2 + R_m^2)$, which explains the quadratic growth of weak responses and produces the quality-factor formula. For cascading, each subsequent stage $z_j$ is forced by the previous stage $z_{j-1}$, which accumulates a response exponent $(1/3)^j$ at high amplitudes. Together these two mechanisms carry the argument.","core_discovery":"The central discovery claim is that tuning a Hopf-oscillator detector to a cubic distortion product of a two-tone stimulus, such as $2\\omega_f - \\omega_m$, converts the detector's response from the familiar compressive growth of primary-tone detection into a regime with quadratic growth at weak inputs, strong compression at large inputs, and a quality factor that grows with distortion-product order. The analytic approximation gives $Q_{\\mathrm{DP}}/Q_{\\mathrm{PT}} = (\\omega_m - \\omega_f)^2 / (\\tfrac{1}{2} F_f F_m)^{2/3}$, so the selectivity gain diverges as the product of the stimulus amplitudes goes to zero. Cascading the output of one Hopf oscillator into another changes the high-amplitude response growth from $F^{1/3}$ to $F^{(1/3)^j}$, flattening the response and steepening the near-resonance window. The concrete combined model—a first oscillator representing the flagellum driven by female and male wingbeat tones, and a second oscillator tuned to $2\\omega_0 - \\omega_m$ and driven by the flagellum's response—achieves better sensitivity, compression, and frequency selectivity than either oscillator alone. This is offered as a general principle for active acoustic detection, inspired by and consistent with the observed mosquito ear configuration.","pith_inferences":["Because the model omits noise, the divergence of the quality-factor ratio as $F_f F_m \\to 0$ will in reality be cut off at the noise floor; the practically achievable selectivity gain is the ratio at the amplitude where the distortion product equals the noise level.","The same mechanism could be tested in other flagellar insects with active auditory mechanics, since only the normal-form nonlinearity and a two-tone stimulus are required.","The non-monotonic response amplitude of a single distortion-product detector leaves an input-level ambiguity; the paper mentions arrays of detectors as a fix, and a natural extension is to ask how population coding across multiple distortion-product orders resolves the ambiguity.","The control parameter $\\mu$ is identified as a plausible target of efferent neuromodulation; if that link is real, the ear could dynamically switch between primary-tone and distortion-product detection regimes, an idea the paper leaves implicit."],"forward_implications":["A detector tuned to a distortion product instead of the primary tone gains a frequency-selectivity advantage that grows as the stimulus amplitude weakens, rather than vanishing with it.","Cascading $n$ oscillators turns the large-amplitude response growth into $F^{(1/3)^n}$, so the response flattens to near-constant and the dynamic range widens.","The two-oscillator flagellum-plus-neuron configuration is better at weak-signal sensitivity, strong-signal compression, and frequency selectivity than either of its components alone.","The benefits are robust to imprecise tuning of the control parameters $\\mu_j$, since they persist with Gaussian spread around the bifurcation point.","Most of the cascade's gain arrives in the first few stages, with diminishing returns for additional layers."],"supporting_citations":[{"why":"Supplies the experimental observation that male mosquito sensory neurons are optimally tuned to a distortion product of male and female wingbeat tones.","marker":"[17]"},{"why":"Evidence that sex recognition in Culex mosquitoes is mediated by acoustic distortion.","marker":"[21]"},{"why":"Masking experiment showing male mosquitoes use distortion to detect females.","marker":"[23]"},{"why":"Provides the normal form equation for the supercritical Hopf bifurcation used as the model for each detector element.","marker":"[28]"},{"why":"Documents that distortion-product amplitudes fall off exponentially with order, justifying focus on low-order distortion products.","marker":"[29]"},{"why":"Establishes the power-law growth and compressive behavior of Hopf-oscillator detectors used for comparison.","marker":"[30]"},{"why":"Provides the two-tone suppression and combination-tone formulation used to write the distortion-product response in effective single-tone form leading to Eq. (21).","marker":"[39]"}],"fun_headline_variants":["Phantom-tone tuning boosts acoustic frequency selectivity","Cascading Hopf oscillators yield sharper acoustic detection","Distortion products sharpen mosquito hearing in noisy swarms","Two-stage oscillator model reveals acoustic detection advantages","Mosquito-inspired nonlinearity enhances frequency selectivity"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that detection performance can be judged from deterministic phase-locked responses without including noise; the model equations have no stochastic term, even though the motivating task is finding a female in a noisy swarm.","fun_headline_variants_meta":{"raw":{"variants":["Phantom-tone tuning boosts acoustic frequency selectivity","Cascading Hopf oscillators yield sharper acoustic detection","Distortion products sharpen mosquito hearing in noisy swarms","Two-stage oscillator model reveals acoustic detection advantages","Mosquito-inspired nonlinearity enhances frequency selectivity"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000425,"raw_usage":{"total_tokens":2184,"prompt_tokens":957,"completion_tokens":1227,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":573,"completion_tokens_details":{"reasoning_tokens":1155}},"tokens_in":573,"tokens_out":1227,"duration_ms":13233,"temperature":1.0,"reasoning_tokens":1155,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T21:15:01.450544+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Add noise of realistic amplitude to Eqs. (1), (6), and (7) and measure the phase-locked response at $2\\omega_f - \\omega_m$: if intrinsic oscillator fluctuations or swarm noise wash out the sharpened tuning curve and the $Q_{\\mathrm{DP}}/Q_{\\mathrm{PT}}$ scaling of Eq. (21), the predicted detection advantage does not survive. A complementary experiment would record Johnston's organ neuron responses to two-tone stimuli and test whether neural tuning width narrows roughly as $F_f^{-2/3}$ as the female-tone amplitude drops.","supporting_citations":[{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the experimental observation that male mosquito sensory neurons are optimally tuned to a distortion product of male and female wingbeat tones."},{"cited_title":"Warren, G","cited_arxiv_id":null,"evidence_quote":"Evidence that sex recognition in Culex mosquitoes is mediated by acoustic distortion."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Masking experiment showing male mosquitoes use distortion to detect females."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Provides the normal form equation for the supercritical Hopf bifurcation used as the model for each detector element."},{"cited_title":"J¨ ulicher, D","cited_arxiv_id":null,"evidence_quote":"Documents that distortion-product amplitudes fall off exponentially with order, justifying focus on low-order distortion products."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the power-law growth and compressive behavior of Hopf-oscillator detectors used for comparison."},{"cited_title":"Stoop and A","cited_arxiv_id":null,"evidence_quote":"Provides the two-tone suppression and combination-tone formulation used to write the distortion-product response in effective single-tone form leading to Eq. (21)."}],"review_version":1}