{"id":"b41118d2-71cf-4eb8-a78a-c21c2412e641","arxiv_id":"2509.08533","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"In a nonlocal aggregation model, increasing system size creates saddle-point transients that exponentially slow pattern formation, a prediction qualitatively confirmed in mouse cell experiments.","lead":"This paper studies a mathematical model of particles that attract each other inside a confined space and shows that larger systems create slow, unstable patterns that delay the final clustering, making patterning time grow exponentially with system size. The authors test the model against mouse cells on patterned slides and find the same size-dependent behavior.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Experimental confirmation is conditional on an unmeasured neutral boundary β=αρ0; non-neutral β changes the predicted pattern classes, so the Fig. 5b match could be coincidental.","rationale":"The central argument has two pillars: a theoretical mechanism (saddle-induced slow modes) and an experimental confirmation. The theory is internally consistent given its stated assumptions; the eigenvalue reduction and instability curves are derived carefully for exponential kernels and neutral boundaries. The pillar that has to carry the 'confirm our predictions' claim is the experiment. The reader's strongest claim explicitly includes experimental confirmation. That is where the load-bearing weak point lies: the model-data comparison is made on the neutral-boundary line, β=αρ0, which is an unmeasured special case, and α is fitted from the same data. Deviations from this line change the qualitative stable states (Fig. 3) and therefore the predicted transient categories. A joint fit of α and β would provide a decisive test. I agree with the reader's identification of this assumption, and I do not see another concern that is more directly tied to the paper's central claim. The exponential-law issue (numerical only, narrow ℓ range) is secondary: even if the exact functional form is not exponential for all ℓ, the theoretical mechanism can still be correct; the boundary-neutral assumption, by contrast, determines whether the experimental comparison tests the mechanism at all. For this reason I keep the reader's CONDITIONAL verdict; the paper would need the β-free fit before the experimental confirmation can be accepted.","tokens_in":9779,"tokens_out":8669,"duration_ms":92598,"concrete_test":"Fit β (and α) as free parameters to the early-time experimental density profiles used in Fig. 5a, e.g., by minimizing the linearized Eq. (5) prediction error over both parameters; then compute the 95% confidence region. If this region excludes β=αρ0, or if the best-fit β yields a predicted pattern-fraction curve in Fig. 5b that differs by more than the experimental sampling error, the experimental confirmation is not robust to boundary interactions. Alternatively, use leave-one-out cross-validation: fit α,β on a subset of experimental bars and predict the held-out pattern types; compare the predictive performance of the neutral-constrained model against the free-β model.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's headline experimental confirmation (Fig. 5) compares the distribution of transient patterns to simulations of Eq. (5) with β=αρ0, the 'neutral' boundary condition. This value is not measured; it is imposed. The entire analytical structure that generates the predicted mode sequence—Eqs. (7)–(12), the pitchfork-bifurcation sequence, and the saddle branches in Fig. 4—is derived only for this neutral case (uniform ρ=ρ0 is a steady state only when β=αρ0). Fig. 3 shows that β is not a minor parameter: β<0 gives a single center peak, β>α gives symmetric boundary peaks, and only 0<β<α gives polarized final states. Thus the specific categories (polarized, middle peak, two peaks) used in Fig. 5b are tied to the neutral assumption. Fitting α from the same experimental early-time data can further compensate for a non-neutral β, making the apparent agreement circular. If the true boundary interaction deviates from β=αρ0, the experimental pattern fractions may match the neutral prediction purely by chance or may fail, and the 'confirmation' would not test the proposed saddle-induced slow modes.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This Letter studies pattern formation in a one-dimensional nonlocal advection-diffusion model of self-attracting active particles confined by boundaries. For exponential interaction kernels of equal range for particle-particle and particle-boundary coupling, the authors derive an exact fourth-order ODE reduction of the linear-stability eigenproblem under the neutral boundary choice β=αρ0, yielding analytic instability curves in the (ℓ,α) plane and accurate perturbative thresholds. Numerical bifurcation analysis shows that stable polarized states coexist with a hierarchy of unstable saddle branches that appear as the system size grows; these saddles guide slow transients, producing exponentially growing polarization times. Experiments on primary mouse cells patterned on rectangular bars are used to fit α from early-time data and then compared with model simulations for the distribution of transient pattern types across system sizes. The authors claim the experimental data confirm the predicted size-controlled patterning dynamics.","tokens_in":10107,"tokens_out":7806,"duration_ms":92512,"significance":"If the theoretical results stand, the paper makes a valuable conceptual contribution: confinement size controls not only whether patterns form but also the dynamical pathway and timescale of patterning, via saddle-induced slow transients. The exact eigenvalue reduction for exponential kernels is elegant and provides a rare analytical window into nonlocal problems on bounded domains; the instability curves of Eq. (12) and the perturbative thresholds appear internally consistent and are checked against numerics. The availability of code and the explicit comparison with experiments are also strengths. However, the experimental confirmation is currently much weaker than the theoretical core: it relies on an unmeasured boundary-interaction strength, on a calibration/validation using the same dataset, and on a small number of replicates at the sizes that carry the main claim. The theoretical part is likely to be of lasting value; the experimental claim needs substantial reframing or additional evidence.","major_comments":[{"comment":"The experimental confirmation is built on the neutral-boundary assumption β=αρ0, which is imposed rather than measured. The analytical saddle-branch analysis (Eqs. (7)-(12), Fig. 4) is derived only for this case, and the simulations in Fig. 5b set β=αρ0. Figure 3 shows that the stable pattern class depends strongly on β: β<0 gives a center peak, β>α gives boundary peaks, and only 0<β<α gives polarized states. A non-neutral β would therefore change the predicted fractions of 'polarized', 'middle peak', and 'two peaks' categories. The fit of α from the early experimental timepoints does not constrain β, so the agreement in Fig. 5b does not provide a stand-alone test of the saddle-induced slow-mode mechanism. Please estimate β independently, or demonstrate that the Fig. 5b predictions are robust to β over an experimentally plausible range, or explicitly reframe the experimental claim as con","section":"Experimentally observed patterns... / Fig. 5"},{"comment":"The exponential dependence of the polarization time τ_pol on system size is one of the two headline results (abstract and Conclusions), but it is asserted from simulations over a narrow range of ℓ without quantitative support. The lower panel of Fig. 4a appears to cover roughly ℓ∈[2,5] (up to the second saddle-branch threshold), and no fitted exponent, confidence interval, or comparison with a predicted scaling is reported. Over such a short range, an apparent exponential law can be a generic finite-size effect. Please report the fitted exponential rates for the mean and standard deviation, the number of ℓ values used, and a check that the rate is consistent with the number of saddle branches or the associated eigenvalue gaps.","section":"Fig. 4a (lower panel) and Conclusions"},{"comment":"The statistical support for the size-controlled suppression of multi-peak patterns is weak. In Fig. 5b, the two smallest domain sizes, which underpin the claim that confinement below 160 µm suppresses multiple peaks, appear to have only N=3 and N=2 replicates (assuming the N labels are in the same order as the bar lengths). In addition, the comparison time is set by mapping 16 h to 2.4τ, but the fitted value of τ is not reported; the predicted pattern fractions are time-dependent, so the comparison is sensitive to this mapping. Please provide the fitted τ and its uncertainty, and show how the fractions in Fig. 5b change if the classification time is varied within the uncertainty.","section":"Fig. 5b-c"},{"comment":"Figure 5 tests only the spatial pattern-type distribution at a single timepoint. It does not directly test the predicted exponential growth of the polarization time or the slow transient dynamics, which are the mechanism-specific predictions of the paper. The title and abstract claim that 'experimental measurements confirm our predictions' about patterning dynamics. If the experimental evidence is to support the dynamical claim, a time-resolved observable (e.g., the time at which the dominant mode emerges, or a comparison of τ_pol across sizes) is needed. Otherwise the conclusions should be limited to 'the observed spatial pattern statistics are consistent with the model under the neutral-boundary assumption'.","section":"General: scope of the experimental claim"}],"minor_comments":[{"comment":"The perturbative expansion in Eq. (9) does not explicitly identify the small parameter. Please state that ar α is assumed small and define the regime of validity of the approximation beyond the statement that it works for ω_n<0.","section":"Equation (9)"},{"comment":"The inset in Fig. 2a, which illustrates the eigenvalue veering/cross-over, is too small to read. Please enlarge it or provide a separate panel.","section":"Figure 2"},{"comment":"The definition of μ(ρ)=A(ρ)||ρ-ρ0||^2 is confusing because A(ρ) is already an integral functional of ρ. Rewriting A as a separate functional and then defining μ as its product with the L2 norm would improve clarity.","section":"Order parameter (Fig. 3)"},{"comment":"Reference [34] is cited as 'in preparation' for the experimental system, and the full imaging/classification methods appear only in the supplementary material. Since Fig. 5 depends on these methods, please provide more details in the main text or update the reference if a preprint is available.","section":"References"},{"comment":"The dashed lines in Fig. 3c are described as 'multiple unstable multi-peak states' but are not defined as saddle-node or pitchfork branches. Please label the bifurcation types or refer explicitly to the supplement.","section":"Figure 3c caption"}],"recommendation":"major_revision","confidential_remarks":"The theoretical analysis is strong and likely publishable; the main gap is the overclaim in the experimental confirmation. The authors could satisfy the major revision by reframing the experimental section as a consistency check rather than a confirmation, or by adding a direct estimate/sensitivity analysis of β. I would also flag that the key experimental dataset relies on an unpublished reference [34]; this should be resolved before publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Here's my take. The paper has a genuinely useful analytical core: exact eigenvalue reduction for exponential kernels on a bounded domain, closed-form instability curves (Eq. 12), and a clean demonstration that confinement generates a sequence of pitchfork bifurcations whose saddle branches control transient dynamics. That part is new and sound. The experimental part is suggestive but not conclusive: the match between simulation and cell patterns in Fig. 5b hinges on the neutral-boundary assumption beta = alpha*rho0, which is imposed rather than measured, and alpha is fit to the same data. I'd send this to review rather than desk reject, but I'd ask for robustness checks.\n\nWhat the paper does well: the analytical machinery is real. The reduction of the nonlocal eigenproblem to a fourth-order ODE for exponential kernels is elegant, and the instability curves (Eq. 12) are exact within the stated assumptions. The numerical bifurcation analysis is careful, and the phase-plane picture of saddle-guided transients (Fig. 4) is a nice conceptual contribution that could generalize beyond this specific model. The exponential growth of polarization time is presented as a simulation result over a limited range of system sizes; that is a minor weakness since no analytical scaling argument is given, but it is clearly labeled as numerical.\n\nThe soft spots are real but addressable. First, the entire theoretical structure—uniform steady state, eigenmodes, saddle branches, and the transient pattern classification in Fig. 5b—is derived for the neutral boundary beta = alpha*rho0. This is not a small technical restriction: Fig. 3 shows that different beta gives different final states (center peak, boundary peaks, polarized), so the predicted distribution of transient patterns is tied to that neutral choice. The stress-test note is right that beta is not measured; if the real boundary interaction deviates, the experimental agreement could be coincidence or due to beta compensation while fitting alpha. Second, alpha is estimated from the same experiments used for validation, which introduces circularity, and the experimental sample sizes are small (N ranges from 2 to 26 per bar length) with no error bars. Third, the assumption that particle-particle and particle-boundary kernels have equal shape and range is stated but not tested.\n\nThese are not fatal flaws. The analytical results stand on their own, independent of the biology. The experimental claim is conditional, but the authors are appropriately cautious elsewhere—they call the patterns \"transients\" and the comparison \"signatures.\" I would like to see the authors test the sensitivity to beta, report error bars, and either release the experimental data or point to a public version. The patterns themselves are from a companion paper (Zhao et al., in prep.), so some details are legitimately deferred.\n\nWho should read this: anyone working on nonlocal aggregation models, active matter on finite domains, or biological self-organization on micropatterns. It deserves a serious referee—the theoretical payoff is enough, and the experimental claim, while not airtight, is worth scrutiny. I would accept it for peer review and push for a revision that makes the beta dependence explicit.","headline":"Solid analytical core on bounded-domain nonlocal advection-diffusion; the experimental confirmation is suggestive but rests on a neutral-boundary assumption that is not directly measured.","tokens_in":10491,"tokens_out":1942,"would_cite":true,"duration_ms":24034,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["35B36","35K57","92C15"],"pacs":[],"model":"deepseek-v4-flash","headline":"This paper shows that confinement size controls not just where patterns form, but how long they take: beyond a critical length, slow multi-peaked transient states cause an exponential delay, and mouse-cell experiments confirm the predicted","keywords":["nonlocal advection-diffusion","pattern formation","confinement","active particles","bifurcation analysis","slow transients","multicellular aggregation","eigenvalue veering"],"falsifier":"Measure the time to reach a polarized profile in the same cell assay across a continuous range of bar lengths spanning the predicted first instability; if the mean and variance of that time do not grow exponentially once the bar exceeds the threshold, or if no multi-peaked transients appear, the saddle-controlled slowing mechanism is falsified. A second check is to vary boundary adhesiveness: the model predicts neutral walls polarize fastest, so making walls more attractive or repulsive should both slow polarization, and a flat timing curve would rule out the proposed boundary mechanism.","tokens_in":9718,"feed_emoji":"🧫","tokens_out":10516,"duration_ms":116990,"temperature":0.7,"pith_summary":"The paper asks whether the size and boundaries of a confined space change how a collective of self-attracting particles forms patterns—not just where the patterns sit, but how long they take to form. It studies a minimal nonlocal advection-diffusion model in which particles attract each other and also feel the walls, and shows that increasing the domain size triggers successive bifurcations that add unstable multi-peaked states. Trajectories pass near these saddles and leave slowly, so the time needed to reach the final polarized pattern, and its shot-to-shot variability, grow exponentially with system size. The boundary interaction determines which final pattern appears: repelling walls give a central peak, strongly attracting walls give boundary peaks, and weakly attracting walls give asymmetric polarized states. Experiments with mouse embryo cells on microprinted bars show the predicted shift: larger bars display more symmetric and multi-peaked transient patterns at early times, while bars below roughly 160 µm suppress multi-peak formation.","feed_headline":"Larger confined systems slow pattern formation exponentially","feed_subtitle":"Beyond a critical size, unstable multi-peaked states slow polarization; mouse-cell data match the predicted shifts.","key_machinery":"The central object is the exact eigenvalue reduction for exponential interaction kernels: an eigenfunction u of the nonlocal problem (∂xx−ᾱ∂xA)u=ωu with no-flux boundary conditions also solves the constant-coefficient ODE u''''+(2ᾱ−1−ω)u''+ωu=0. This reduction turns the nonlocal stability problem into algebra and yields the instability locus e^{ℓz}=(−1)^n(z−1)/(z+1) with z=√(1−2ᾱ), giving the size thresholds at which each spatial mode destabilizes. The second element is the sequence of pitchfork bifurcations that adds unstable saddle steady states; their unstable manifolds carry the slow transients that delay polarization.","core_discovery":"For a generic class of nonlocal advection-diffusion equations describing self-attracting active particles in a confined one-dimensional domain, the paper derives that boundaries and system size are active controls of the dynamics. Assuming interaction kernels that are equal exponentials for particle-particle and particle-boundary forces, every eigenfunction of the linearized stability problem also satisfies a constant-coefficient fourth-order ODE, which yields algebraic equations for the eigenvalues and exact instability curves e^{ℓz}=(-1)^n(z-1)/(z+1), z=√(1−2ᾱ). These curves show that successive spatial modes become unstable as the domain length increases, each transition adding an unstab","pith_inferences":["If the saddle-induced slowing is generic, tissues that grow across the critical size should suddenly display long, variable patterning times—an effect that could matter for interpreting developmental timing.","The exact reduction for exponential kernels likely extends to finite sums of the form s^m e^{-s}, yielding closed-form thresholds for more complex interaction kernels without further approximation.","Varying wall adhesiveness at fixed domain size would isolate the boundary's role: the model predicts neutral walls polarize fastest, and repulsive or strongly attractive walls both slow the approach to the final state."],"forward_implications":["Systems confined below the first critical length polarize quickly and robustly; above it, polarization slows because multi-peaked saddle states trap trajectories.","The mean and standard deviation of the polarization time increase exponentially with system size, so larger systems are both slower and less reproducible in their patterning time.","Wall interaction type selects the final pattern: repelling boundaries yield a central peak, strongly attracting boundaries yield boundary peaks, and weakly attracting boundaries yield asymmetric polarized states.","The number of unstable multi-peak states is controlled by system size and attraction strength, so the sequence of transient patterns can be predicted from the analytical instability thresholds.","The eigenvalue-to-ODE reduction and bifurcation approach extend to other nonlocal kernels, applying to neural fields, ecological dispersal, and kernel-based reaction-diffusion models."],"supporting_citations":[{"why":"The companion derivation and methods file: contains the eigenvalue-to-ODE reduction, algebraic eigenvalue equations, instability thresholds, numerical bifurcation analysis, and experimental data analysis.","marker":"[22]"},{"why":"Establishes nonlocal advection-diffusion equations as a generic framework for collective dynamics of self-attracting constituents, motivating the model class.","marker":"[13]"},{"why":"Provides the microscopic random-walk derivation of the density-dependent mobility used in the model.","marker":"[16]"},{"why":"Links chemotaxis-type mechanisms to exponentially decaying nonlocal kernels, justifying the kernel shape.","marker":"[19]"},{"why":"Gives the infinite/periodic-domain patterning condition that the finite-domain thresholds are compared against.","marker":"[23]"},{"why":"Introduces the chemotaxis model that the exponential attraction kernel represents, grounding the interpretation of particle-particle attraction.","marker":"[17]"},{"why":"Supplies the experimental dataset of reaggregated mouse cells on size-controlled microprinted bars used to test the predicted transient pattern distribution.","marker":"[34]"},{"why":"Documents eigenvalue veering in structural dynamics, the phenomenon used to interpret the crossing and turning of the eigenvalue loci.","marker":"[24]"}],"fun_headline_variants":["Size-triggered slow modes stretch patterning exponentially","Boundaries control how slow patterns form","Confined active particles hit exponential pattern delay","System size governs exponential rise in patterning time","Pattern timescales blow up with confined size"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The quantitative predictions assume the particle-particle and particle-boundary forces have the same exponential range, and the experimental match assumes the wall is neutral (β=αρ0) with α fitted from the same data, so any departure from this interaction balance shifts the predicted thresholds and transient fractions.","fun_headline_variants_meta":{"raw":{"variants":["Size-triggered slow modes stretch patterning exponentially","Boundaries control how slow patterns form","Confined active particles hit exponential pattern delay","System size governs exponential rise in patterning time","Pattern timescales blow up with confined size"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000537,"raw_usage":{"total_tokens":2330,"prompt_tokens":572,"completion_tokens":1758,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":316,"completion_tokens_details":{"reasoning_tokens":1700}},"tokens_in":316,"tokens_out":1758,"duration_ms":16869,"temperature":1.0,"reasoning_tokens":1700,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-04T20:27:07.785411+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the time to reach a polarized profile in the same cell assay across a continuous range of bar lengths spanning the predicted first instability; if the mean and variance of that time do not grow exponentially once the bar exceeds the threshold, or if no multi-peaked transients appear, the saddle-controlled slowing mechanism is falsified. A second check is to vary boundary adhesiveness: the model predicts neutral walls polarize fastest, so making walls more attractive or repulsive should both slow polarization, and a flat timing curve would rule out the proposed boundary mechanism.","supporting_citations":[],"review_version":1}