{"id":"145c6c40-45a8-456a-989b-b5ab09496cc4","arxiv_id":"2505.19665","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":5,"one_line_summary":"Starling flocks can propagate a collective turn as a wave while spontaneous velocity fluctuations stay overdamped; adding a quartic FPUT term to the alignment interaction reconciles both observations.","lead":"New panning-camera experiments on starling flocks show that collective turns travel as underdamped waves while spontaneous velocity fluctuations look overdamped, a combination linear theory says is impossible. The authors show that adding a quartic (Fermi-Pasta-Ulam-Tsingou) term to the standard alignment model can produce both behaviors at once.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Inequality (8) only enforces overdamping at k~1/L; at the k values measured experimentally the linearized ISM predicts underdamped spin-wave peaks, so the model may not reproduce the full Lorentzian correlation.","rationale":"The paper identifies a real paradox and the FPUT mechanism is a plausible resolution; the new panning-camera experiments and the explicit out-of-equilibrium field-theory calculation are genuine contributions. The simulations demonstrate that coexistence of Lorentzian correlation and propagating finite perturbations is possible in one parameter regime. However, the load-bearing premise is that spontaneous fluctuations are overdamped while finite perturbations are underdamped, and the paper's inequality (8) only guarantees overdamping at the smallest wavevector. Since the quartic term is dormant for spontaneous fluctuations, those fluctuations obey the linear ISM equation; at any measured k with (J/χ)a²k² > γ² the undriven correlation should be non-Lorentzian unless damping hides the peaks. The paper does not check the ISM+FPUT correlation above the linear crossover, and its k∼1/L criterion is therefore insufficient for the experimental k range. This concern is more specific than the reader's 'parameters not measured' objection: it shows that the window (8) is not even the correct condition for the claimed Lorentzian correlation. The proposed simulation check would settle this directly. It does not change the overall verdict, which remains CONDITIONAL: the mechanism is plausible but the parameter window and k-resolved behavior must be validated before the reconciliation can be accepted.","tokens_in":2015,"tokens_out":1236,"duration_ms":179963,"concrete_test":"Run the ISM+FPUT simulation with the quoted parameters (J=20, J4=10⁵, η=7, χ=1.25, T=10⁻⁶) and compute C(k,ω) over a frequency range covering ±c_s k for k values above the linear crossover k_c≈0.7, e.g. k a = 1.26, 1.88, and the experimental k a ≈ 3 and 5. If non-Lorentzian spin-wave structure appears at any of these k, the model fails to reproduce the full experimental correlation and inequality (8) must be replaced by a k-resolved bound; also report 1−Φ for the run and check whether J4 ≤ J/(1−Φ) still holds.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Spontaneous fluctuations in ISM+FPUT see only the quadratic stiffness J, because the quartic term is dormant by construction (J4 δφ² ≲ J); their correlation is therefore governed by the same linear ISM dispersion as Eq. (4). The paper's window (8) imposes overdamping only at k ∼ 1/L, but the experimental C(k,ω) is reported at k = 1.5–5 m⁻¹, far above 1/L. For any mode with (J/χ)a²k² > γ² the ISM is underdamped. With the quoted simulation parameters (J=20, χ=1.25, η=7, γ=2.8), the linear crossover is k_c ≈ 0.7, so the displayed k = 1.26 and 1.88 in Fig. 3 are nominally underdamped: if the Lorentzian shape there is only a consequence of strong damping, the truly reproducible condition is much stronger, roughly (J/χ)a²k_max² ≲ 2γ². Imposing this at the experimental k_max tightens the allowed J by a factor ∼ (k_max L)² relative to (8). Combining the tightened bound with J4 ≫ χγ²(L/a)² and J4 ≲ J/(1−Φ) yields 1−Φ ≲ 2/(k_max L)²; for L/a ∼ 10 and k_max a ∼ 5 this requires Φ ≳ 0.992, far above typical starling polarizations. The paper provides no data establishing that real flocks satisfy this k-resolved version of the window, so the central reconciliation is unverified precisely at the wavevectors where the absence of spin-wave peaks is claimed.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports new panning-camera observations of starling flocks showing that collective turns propagate linearly with little damping while the unperturbed velocity correlation function is Lorentzian with half-width scaling as omega_L ~ k^2 and no spin-wave peaks. The authors argue that the Inertial Spin Model (ISM) cannot account for this combination, because its underdamped regime implies spin-wave peaks in the unperturbed correlation. They propose adding a quartic alignment term to the pseudo-Hamiltonian, defining an effective stiffness J_eff = J + J4 (delta_phi)^2, and derive the parameter window in inequality (8): J less than or comparable to chi gamma^2 (L/a)^2 much less than J4 less than or comparable to J/(1-Phi). The claim is that spontaneous fluctuations see J and are overdamped, while strong perturbations activate J4 and become underdamped. Simulations with J=20, J4=10^5, eta=7, T=10^-6 show both a Lorentzian correlation and linear propagation after a 90-degree initiator turn. Finally, in the planar, noiseless, dissipation-free fixed-network limit the equations reduce to the Fermi-Pasta-Ulam-Tsingou (FPUT) form, and the authors suggest that the observed traveling turns are FPUT solitons.","tokens_in":20178,"tokens_out":18100,"duration_ms":237753,"significance":"If the proposed mechanism holds, it resolves a genuine paradox in a natural active system: the separation of the linear response to strong perturbations from the spontaneous fluctuation spectrum through a quartic stiffness. The new experimental dataset, with tracking times up to 12.5 s, is itself a valuable contribution, and the explicit connection between the model and the FPUT equation is elegant. However, the central reconciliation currently rests on a single simulation with hand-picked parameters, a parameter window validated only at k ~ 1/L, and an interpretive leap to solitons that the authors themselves state requires further work. No public code or experimental error bars accompany the analysis, and the quartic stiffness J4 is never estimated from the data. The paper is best read as a promising proof of principle rather than a validated reconciliation.","major_comments":[{"comment":"The overdamped side of inequality (8) is imposed only at the scale k ~ 1/L, but the Lorentzian correlation functions are measured at k = 1.5-5 m^-1 in the experiments (Figs. S1-S2) and displayed at k = 0.63, 1.26, and 1.88 in the ISM+FPUT simulation (Fig. 3b). For the linearized ISM, the spectral function of transverse fluctuations changes from a central Lorentzian to a spectrum with resolved side peaks when (J/chi)a^2 k^2 exceeds 2 gamma^2, not merely gamma^2. With the quoted simulation parameters (J=20, chi=1.25, eta=7, hence gamma=2.8 and gamma^2=7.84), the mode at k=1.88 has (J/chi)a^2 k^2 between roughly 42 (using the exact cubic-lattice eigenvalue 2J(1-cos k)/chi) and 56 (using the continuum approximation), well above 2 gamma^2=15.7; this mode should therefore show resolvable spin-wave peaks, contradicting the claimed Lorentzian correlation. The paper must specify the frequency range and the peak-resolution criterion used in the analysis, and either correct the quoted parameters or demonstrate that the full anisotropic active correlation remains Lorentzian at these wavevectors.","section":""},{"comment":"The soliton interpretation is not supported by any calculation or simulation in the manuscript. Equation (9) is the FPUT equation only in the planar, noise-free, dissipation-free, fixed-network limit; the paper itself states that the problem 'requires to study FPUT solitons on an adiabatically evolving 3d interaction network' and to include weak noise and dissipation, but no such study is presented. The simulations show a propagating disturbance after a 90-degree turn, but no evidence is given that this disturbance is a soliton (for example, an amplitude-width-speed relation, shape persistence, or soliton collision tests) rather than a weakly damped linear wave in the J4-dominated regime. Since the title and abstract advertise 'a case for soliton propagation', this load-bearing interpretive step either needs to be substantiated or the claims should be softened to nonlinear propagating pulses.","section":""},{"comment":"The reconciliation is not tested against the new experimental data. Inequality (8) involves J, J4, gamma, chi, L, and Phi, but the paper reports no estimates of these parameters from the starling measurements and does not show that measured polarizations and turn speeds fall inside the window. In particular, if the overdamped condition is imposed at the measured wavevectors rather than at k ~ 1/L, the combination of J4 much greater than chi gamma^2 (L/a)^2, J4 less than or comparable to J/(1-Phi), and J less than or comparable to chi gamma^2/(a^2 k_max^2) implies (1-Phi) less than or approximately 1/(k_max L)^2. For L/a ~ 10 and k_max a ~ 5 this requires Phi greater than about 0.9996, a much stronger constraint than the paper's invocation of the margin 1/(1-Phi) much greater than 1. No evidence is provided that real flocks satisfy this k-resolved version of the window, so the central claim that the FPUT term reconciles theory with experiment remains a proof of principle rather than a validated explanation.","section":""}],"minor_comments":[{"comment":"The expansion of (v_i - v_j)^2 contains the term (pi_i^2 - pi_j^2)^2/(4 v_0^2), which is subsequently dropped using J << J4; the paper should state why this term is negligible relative to the retained quartic term, since the two terms have different dependence on the phase differences.","section":""},{"comment":"The normalization of C(k, omega) and the frequency binning are not defined; please specify the normalization convention and the effective frequency resolution so that the absence of spin-wave peaks can be assessed quantitatively.","section":""},{"comment":"The text uses 'Lorentzian' and 'quasi-Lorentzian' interchangeably; it would help to state the fitted functional form and the criterion used to discriminate Lorentzian from spin-wave spectra (for example, the ratio of the spectral value at the expected spin-wave frequency to the central value).","section":""}],"recommendation":"major_revision","confidential_remarks":"The new experimental dataset and the non-linear stiffness mechanism are potentially interesting, and the route to a full paper is clear. However, the simulation evidence appears internally inconsistent at the largest displayed wavevector, and the parameter window is not checked against the measured k range or against real-fly polarization values. I would not accept before these load-bearing points are fixed. The soliton claim should also be either demonstrated or removed from the title and abstract."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"One thing to know: this paper reports a genuinely surprising experimental finding—underdamped traveling turns coexisting with overdamped Lorentzian correlations—and offers a clever FPUT-style mechanism to reconcile them. The new panning-camera data are a real advance, and the idea that a quartic stiffness term makes response stronger than spontaneous fluctuation is physically interesting. The paper is also honest: the soliton interpretation is labeled speculative, and the supplementary linear analysis is careful.\n\nThe soft spot is the quantitative window. Inequality (8) enforces overdamping only at k~1/L, but the correlation functions are shown at k = 1.26 and 1.88, where the linearized ISM with the stated parameters (J=20, chi=1.25, eta=7) is underdamped (crossover at k≈0.7). The authors do not explain why the quartic term suppresses spin-wave peaks at those k when it is, by construction, dormant for spontaneous fluctuations (J4 δφ² ≲ J). Tightening the condition to cover the measured k range shrinks the allowed J so much that the upper bound J4 ≲ J/(1−Φ) forces polarization Φ ≳ 0.999 for L/a~10 and k_max a~5—far above observed starling values. So the central reconciliation is unverified precisely at the wavevectors where the absence of spin-wave peaks is claimed.\n\nOther weaknesses are man-sized: J4 is never measured from data, no error bars on the experimental correlations, and no public code or data. The simulations demonstrate coexistence in one parameter set but not a robust regime.\n\nThat said, the flaws are addressable, not fatal. The experimental paradox is real regardless of the model, and the FPUT proposal is a plausible direction worth testing. I would not desk-reject this. A serious referee should ask for a k-resolved version of the window, error estimates, and public artifacts, or a softened claim. The paper deserves revision, not rejection.","headline":"A real experimental puzzle and a plausible nonlinear mechanism, but the claimed parameter window is not verified at the measured wavevectors and requires implausibly high polarization if tightened.","tokens_in":20738,"tokens_out":3600,"would_cite":true,"duration_ms":37867,"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":"Starling turns may be solitons, not spin waves.","keywords":["starling flocks","Inertial Spin Model","Fermi-Pasta-Ulam-Tsingou","collective turns","spin waves","solitons","Lorentzian correlation","effective stiffness"],"falsifier":"Turn one bird by a large angle and measure the restoring torque on its neighbours: the quartic mechanism requires a cubic nonlinearity in the force as the phase distortion approaches order one, so a purely linear restoring force at all amplitudes would close the parameter window.","tokens_in":19555,"feed_emoji":"🐦","tokens_out":9237,"duration_ms":54885,"temperature":0.7,"pith_summary":"Starling flocks turn collectively in a way that looks like a wave: the change of heading sweeps across the group with little damping and a speed that grows with polarization. But new long recordings, made with panning cameras that follow flocks for up to 12.5 seconds, show that the birds' spontaneous velocity fluctuations are overdamped, with a Lorentzian correlation function and no spin-wave peaks. The paper argues that this apparent paradox disappears once the alignment interaction contains a quartic Fermi–Pasta–Ulam–Tsingou term, so the effective stiffness grows with the local phase distortion. Small spontaneous fluctuations feel the small quadratic stiffness and stay overdamped, while a strong turn activates the large quartic stiffness and propagates linearly. If the argument is right, collective turns in starling flocks are solitary waves.","feed_headline":"Starling turns may be solitons, not spin waves","feed_subtitle":"A nonlinear stiffness makes turns propagate while spontaneous fluctuations stay overdamped, matching new long recordings.","key_machinery":"The central object is the effective stiffness $J_{\\mathrm{eff}} = J + J_4\\,\\delta\\varphi^2$, a named identity (equation (7) of the paper) that replaces the constant alignment coupling of the Inertial Spin Model. It carries the argument because it lets weak fluctuations and strong perturbations see different stiffnesses, thereby evading the linear-response tie between propagation and correlation. The planar phase dynamics, once noise and dissipation are dropped, become the Fermi–Pasta–Ulam–Tsingou equations, providing the soliton mechanism for the traveling turns.","core_discovery":"On the paper's own terms, the discovery is that the two experimental facts—underdamped linear propagation of collective turns and overdamped Lorentzian spontaneous correlation—are not contradictory once response and unperturbed correlation are governed by different effective stiffnesses. Adding the quartic term $\\frac{J_4}{4v_0^4}\\sum_{ij} n_{ij}(\\mathbf{v}_i-\\mathbf{v}_j)^4$ to the ISM Hamiltonian yields an effective stiffness $J_{\\mathrm{eff}} = J + J_4\\,\\delta\\varphi^2$, where $\\delta\\varphi$ is the local phase distortion. In the parameter window $J \\lesssim \\chi\\gamma^2(L/a)^2 \\ll J_4 \\lesssim J/(1-\\Phi)$, spontaneous fluctuations (small $\\delta\\varphi$) are overdamped, producing the Lorentzian $C(k,\\omega)$ with half-width $\\omega_L \\sim k^2$, while a perturbation with $\\delta\\varphi \\sim O(1)$ becomes underdamped and propagates with speed $c_s \\sim 1/\\sqrt{1-\\Phi}$, matching the observed turns. Numerically, the three-dimensional ISM+FPUT model reproduces both features, and the planar phase dynamics reduces to the Fermi–Pasta–Ulam–Tsingou equations, whose solitary waves the paper proposes as the mechanism of collective turns.","pith_inferences":["If the FPUT soliton picture is right, the propagation speed of a turn should depend on the amplitude of the heading change—a signature that can be tested on existing recordings without new experiments.","The stiffness-switch mechanism is generic: any system whose interaction strength grows with local misalignment will decouple small fluctuations from large perturbations, so similar 'wave propagation without spin-wave correlation' behavior might appear in other collective animal groups or active matter.","The paper assumes the interaction network evolves adiabatically during a turn; a sharper test would check whether the turning wave's shape and speed distort measurably as the flock's topology changes."],"forward_implications":["The new panning-camera data from nine flocks show that the Lorentzian correlation with $\\omega_L \\sim k^2$ is a robust feature, not an artifact of short recordings.","The ISM+FPUT model reproduces in simulation both the overdamped spontaneous correlation and the underdamped linear propagation of a collective turn.","The predicted relation $c_s \\sim 1/\\sqrt{1-\\Phi}$ between turn speed and polarization survives the addition of the quartic term, so the new model keeps the agreement with earlier measurements.","The quartic term is formally similar to the speed-control mechanism used to explain scale-free speed fluctuations, suggesting a common nonlinear response to large stimuli in bird flocks.","The traveling waves observed in flocks are interpreted as FPUT solitons propagating on an adiabatically evolving three-dimensional interaction network."],"supporting_citations":[{"why":"Supplies the original turning-flock measurements and the Inertial Spin Model, including the speed–polarization relation the new model must preserve.","marker":"[1]"},{"why":"Places the ISM equilibrium dynamics in Model G universality, grounding the expectation of spin-wave peaks.","marker":"[4]"},{"why":"The Fermi–Pasta–Ulam–Tsingou problem that motivates the quartic term and names the solitary-wave interpretation.","marker":"[16]"},{"why":"Demonstrates solitary-wave solutions on FPUT lattices, underpinning the soliton reading of the observed turns.","marker":"[18]"},{"why":"Relates polarization to fluctuation strength through $(1-\\Phi)\\sim T/J$, feeding the inequality window (8).","marker":"[8]"},{"why":"Documents the panning stereo system that produced the long recordings central to the new experimental result.","marker":"[23]"},{"why":"The GReTA tracking software adapted for the panning cameras used in the new experiments.","marker":"[24]"},{"why":"Provides the exact equilibrium dispersion used to compare the spin-wave peak positions against the observed Lorentzian correlation.","marker":"[26]"}],"fun_headline_variants":["Solitons explain starling flock turns at last","Starling turns: solitons instead of spin waves","Nonlinear stiffness turns flock theory to solitons","How starling turns stay fast while correlations lag","FPUT term turns starling turns into traveling waves"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The whole resolution rests on real flocks having a hidden quartic stiffness $J_4$ that is simultaneously large enough to dominate strong turns and small enough to stay dormant under spontaneous fluctuations, yet $J_4$ is never measured directly in the paper.","fun_headline_variants_meta":{"raw":{"variants":["Solitons explain starling flock turns at last","Starling turns: solitons instead of spin waves","Nonlinear stiffness turns flock theory to solitons","How starling turns stay fast while correlations lag","FPUT term turns starling turns into traveling waves"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000266,"raw_usage":{"total_tokens":1590,"prompt_tokens":903,"completion_tokens":687,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":519,"completion_tokens_details":{"reasoning_tokens":611}},"tokens_in":519,"tokens_out":687,"duration_ms":6561,"temperature":1.0,"reasoning_tokens":611,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-07T14:09:09.690299+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Turn one bird by a large angle and measure the restoring torque on its neighbours: the quartic mechanism requires a cubic nonlinearity in the force as the phase distortion approaches order one, so a purely linear restoring force at all amplitudes would close the parameter window.","supporting_citations":[{"cited_title":"Attanasi, A","cited_arxiv_id":null,"evidence_quote":"Supplies the original turning-flock measurements and the Inertial Spin Model, including the speed–polarization relation the new model must preserve."},{"cited_title":"Fermi, P","cited_arxiv_id":null,"evidence_quote":"The Fermi–Pasta–Ulam–Tsingou problem that motivates the quartic term and names the solitary-wave interpretation."},{"cited_title":"Friesecke and J","cited_arxiv_id":null,"evidence_quote":"Demonstrates solitary-wave solutions on FPUT lattices, underpinning the soliton reading of the observed turns."},{"cited_title":"Bialek, A","cited_arxiv_id":null,"evidence_quote":"Relates polarization to fluctuation strength through $(1-\\Phi)\\sim T/J$, feeding the inequality window (8)."},{"cited_title":"Cavagna, X","cited_arxiv_id":null,"evidence_quote":"Documents the panning stereo system that produced the long recordings central to the new experimental result."},{"cited_title":"Attanasi, A","cited_arxiv_id":null,"evidence_quote":"The GReTA tracking software adapted for the panning cameras used in the new experiments."},{"cited_title":"Cavagna, J","cited_arxiv_id":null,"evidence_quote":"Provides the exact equilibrium dispersion used to compare the spin-wave peak positions against the observed Lorentzian correlation."}],"review_version":1}