{"id":"21a3d063-2e94-4330-9ca0-205f8ac84c91","arxiv_id":"1907.11075","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":7.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":1,"one_line_summary":"A whole-connectome stochastic simulator of C. elegans is used with SMC to impute membrane potentials from calcium fluorescence on synthetic data.","lead":"The authors create a stochastic simulator of the entire C. elegans nervous system and body based on its connectome. They apply sequential Monte Carlo to impute hidden membrane potentials from partial calcium imaging observations on synthetic data.","discovery_kind":"new_application","skeptic_critique":{"model":"grok-4.3","headline":"Imputation performance is demonstrated only on data synthesized from the identical simulator; no test of robustness under model misspecification or real calcium traces is reported.","rationale":"The reader's weakest_assumption already flags reliance on synthetic data; the load-bearing issue is precisely that this data is generated from the model being tested, making the experiment a consistency check rather than an external validation of biological regularization.","tokens_in":1743,"tokens_out":321,"duration_ms":11518,"concrete_test":"Generate a held-out test set by running the simulator with 20% of synaptic weights randomly perturbed by ±30% (or by substituting an independent published C. elegans connectome variant), then apply the trained SMC filter to the resulting calcium traces; if median per-neuron voltage RMSE rises by more than a factor of two relative to the matched-model case, the regularization claim does not transfer beyond self-consistent data.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim requires that the connectome-derived stochastic simulator supplies enough regularization for SMC to recover membrane potentials from partial observations. All reported experiments use synthetic trajectories drawn from the same generative model (with parameters either known or variationally optimized). This setup cannot distinguish whether recovery succeeds because the simulator is biologically faithful or simply because the inference procedure is inverting its own forward process. When the true dynamics deviate from the assumed connectome rules, ion-channel kinetics, or noise model—as must occur with real experimental data—the regularization may be insufficient and the imputed potentials could be artifacts of the simulator rather than the observations.","agreement_with_reader":"agree"},"referee_report":{"model":"grok-4.3","summary":"The paper develops a stochastic whole-brain and body simulator of C. elegans grounded in the connectome and uses sequential Monte Carlo (SMC) to impute latent membrane potentials from partial calcium fluorescence observations. It further proposes variational optimization of simulator parameters via the SMC evidence estimate. All reported experiments and validation are performed exclusively on synthetic trajectories generated from the same model (with parameters either known or optimized), using data dimensions representative of current laboratory measurements.","tokens_in":1854,"tokens_out":450,"duration_ms":11142,"significance":"If the connectome-derived simulator supplies biologically faithful regularization that enables accurate recovery of membrane potentials under realistic noise and partial observations, the method would provide a novel route to infer unobservable neural states at cellular resolution from standard calcium imaging. The SMC-based parameter learning is a secondary contribution. However, the exclusive use of in-model synthetic data leaves open whether the regularization holds when the true dynamics deviate from the assumed rules, limiting immediate impact on experimental neuroscience.","major_comments":[{"comment":"Abstract and experiments section: The central claim that the simulator is 'sufficiently regularizing' to allow imputation rests on recovery performance for trajectories drawn from the identical generative model. This design cannot distinguish biological fidelity from successful inversion of the forward process; no experiments under model misspecification (altered ion-channel kinetics, noise statistics, or connectome rules) or on real calcium traces are reported, leaving the regularization claim untested for the intended use case.","section":"Abstract / Experiments"},{"comment":"Methods / Results: The SMC imputation and variational parameter estimation are demonstrated only when the data-generating parameters are either known or recovered from the same simulator; no cross-validation against held-out real or perturbed data is shown to establish that the inferred potentials reflect observations rather than simulator priors.","section":"Methods / Results"}],"minor_comments":[{"comment":"Abstract: 'data of dimension and type representative of that which are measured' contains a subject-verb agreement error ('which are' should be 'which is').","section":"Abstract"}],"recommendation":"major_revision","confidential_remarks":null},"author_rebuttal":{"model":"grok-4.3","summary":"We thank the referee for their constructive feedback. We address the major comments point by point below, clarifying the intended scope of the work as a synthetic proof-of-principle.","responses":[{"response":"The manuscript explicitly states that all experiments use synthetic trajectories generated from the simulator itself, with data dimensions representative of current laboratory measurements. The central claim is scoped to this controlled setting: that the connectome-derived dynamics provide regularization sufficient for SMC-based imputation when observations are consistent with the model. This is a necessary first validation step before real-data application. We do not claim to have tested biological fidelity or robustness to misspecification, as those would require either real traces or deliberate model perturbations outside the current scope. The abstract and methods already qualify the synthetic nature of the results.","revision_made":"no","referee_comment":"[Abstract / Experiments] Abstract and experiments section: The central claim that the simulator is 'sufficiently regularizing' to allow imputation rests on recovery performance for trajectories drawn from the identical generative model. This design cannot distinguish biological fidelity from successful inversion of the forward process; no experiments under model misspecification (altered ion-channel kinetics, noise statistics, or connectome rules) or on real calcium traces are reported, leaving the regularization claim untested for the intended use case."},{"response":"When parameters are known, imputation performance is evaluated on held-out synthetic trajectories. When parameters are variationally optimized, the evidence estimate is maximized on training trajectories and imputation is assessed on separate held-out trajectories from the same model. This demonstrates that the procedure recovers both parameters and latents when the generative assumptions hold. The design isolates the contribution of the observations within the model; pure prior sampling would not match the specific observed calcium dynamics. Cross-validation on real or perturbed data is not included because the work is positioned as synthetic validation of the method.","revision_made":"no","referee_comment":"[Methods / Results] Methods / Results: The SMC imputation and variational parameter estimation are demonstrated only when the data-generating parameters are either known or recovered from the same simulator; no cross-validation against held-out real or perturbed data is shown to establish that the inferred potentials reflect observations rather than simulator priors."}],"tokens_in":1313,"tokens_out":477,"duration_ms":23368,"standing_objections":[]},"desk_editor":{"model":"grok-4.3","letter":"The main takeaway is that the authors created a stochastic whole-brain simulator grounded in the C. elegans connectome and showed that sequential Monte Carlo can recover latent membrane potentials from partial calcium observations when the data comes from the simulator itself. They also use SMC to variationally optimize some simulator parameters. This integration of an anatomically detailed forward model with particle-based inference is new in the literature they cite. The technical execution looks competent: they handle the high-dimensional state space on distributed hardware and report that the synthetic data matches the scale of current lab recordings. The approach is a reasonable way to close the loop between simulation and measurement in a fully mapped organism. The central limitation is that every reported experiment draws trajectories from the identical generative model, either with known parameters or after fitting. This setup cannot separate whether success comes from the simulator capturing real biology or from the inference simply inverting its own assumptions. No results appear on actual calcium traces, and there are no tests under deliberate model misspecification. The math itself follows standard SMC practice, and the citations to connectome and inference work are appropriate. The paper is aimed at computational neuroscientists who work on C. elegans or similar small nervous systems and want to combine detailed simulation with data assimilation. It is coherent on its own terms and shows clear thinking about the inference problem. I would send it to peer review so that referees can press on the validation gap and ask for either real-data experiments or explicit robustness checks.","headline":"The paper builds a connectome-derived stochastic simulator for C. elegans and uses SMC to impute membrane potentials from calcium data, but only demonstrates this on synthetic trajectories generated from the same model.","tokens_in":2371,"tokens_out":373,"would_cite":false,"duration_ms":15123,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":{"model":"grok-4.3","evidence":[],"headline":"C. elegans connectome SMC simulator uses standard HMM inference; no RS cost, φ-ladder or 8-tick structures","alignment":"orthogonal","rationale":"Paper constructs a biological HMM (membrane potentials vt, calcium ct, body bt, fluorescence yt) and applies SMC/PMVO for imputation and parameter learning. Central machinery is connectome-driven ODE discretization + proprioceptive feedback loop, not J-cost, ratio-symmetric cost, golden-ratio identities or parameter-free constant derivations. Domain (q-bio.NC whole-worm simulation) lies outside RS forcing theorems.","tokens_in":58204,"confidence":"high","tokens_out":138,"duration_ms":5038,"cache_read_input_tokens":38528,"cache_creation_input_tokens":0},"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"grok-4.3","headline":"A whole-connectome stochastic simulator of C. elegans imputes single-cell membrane potentials from partial calcium imaging data.","keywords":["C. elegans","calcium imaging","membrane potential","sequential Monte Carlo","whole-brain simulation","connectome","state imputation"],"falsifier":"Direct comparison of SMC-imputed membrane potentials against simultaneous electrophysiological recordings in the same animal, checking whether the imputed traces lie within the measurement noise of the true voltages.","tokens_in":2603,"feed_emoji":"🪱","tokens_out":639,"duration_ms":13239,"temperature":0.7,"pith_summary":"The paper constructs a stochastic simulator of the entire C. elegans nervous system and body that incorporates the known anatomical connectome. It shows that this simulator supplies enough regularization for sequential Monte Carlo methods to recover hidden membrane-potential time series from incomplete calcium-fluorescence observations. The same SMC machinery is also used to optimize simulator parameters by maximizing an approximation to the marginal likelihood. Experiments are performed on synthetic data whose statistics match current laboratory recordings. The work thereby links measurable fluorescence traces to latent voltage states at cellular resolution inside an anatomically grounded model.","feed_headline":"Connectome simulator recovers C. elegans voltages from calcium data","feed_subtitle":"A stochastic whole-body model regularizes SMC to impute single-cell membrane potentials from partial fluorescence observations on synthetic,","key_machinery":"Stochastic whole-brain and body simulator built from the C. elegans connectome, combined with sequential Monte Carlo (SMC) for state imputation and evidence approximation.","core_discovery":"The anatomically grounded whole-connectome simulator is sufficiently regularizing to allow imputation of latent membrane potentials from partial calcium fluorescence imaging observations via sequential Monte Carlo, and the same procedure yields a variational route to parameter estimation.","pith_inferences":["If the simulator remains accurate on real rather than synthetic data, the method could supply voltage estimates in any preparation where only calcium imaging is feasible.","Parameter learning inside the simulator could identify which synaptic or cellular properties are most constrained by population calcium recordings.","The framework might be tested by withholding subsets of cells from the imputation step and checking whether held-out cells are still recovered at usable accuracy.","Extending the same SMC machinery to multi-animal or longitudinal datasets could reveal how circuit parameters change across individuals or over development."],"forward_implications":["Imputation yields time-varying brain-state estimates at single-cell fidelity from covariates that are already measurable in the lab.","Simulator parameters can be learned by variational optimization of the noisy model-evidence approximation supplied by SMC.","The approach operates on synthetic data whose dimension and noise statistics match current calcium-imaging experiments.","The loop from connectome to simulator to imputed voltages constitutes the first reported use of a full anatomical model for this inference task."],"fun_headline_variants":["Simulator imputes C. elegans membrane potentials from calcium imaging","Connectome simulator imputes voltages from partial calcium observations","SMC enables imputation of membrane potentials in C. elegans","Whole-body simulator imputes latent states from calcium fluorescence"],"cache_read_input_tokens":2112,"weakest_assumption_plain":"The custom stochastic simulator supplies enough biological regularization that SMC can accurately recover membrane potentials from partial calcium observations on data representative of real experiments.","fun_headline_variants_meta":{"raw":{"variants":["Simulator imputes C. elegans membrane potentials from calcium imaging","Connectome simulator imputes voltages from partial calcium observations","SMC enables imputation of membrane potentials in C. elegans","Whole-body simulator imputes latent states from calcium fluorescence"]},"model":"grok-4.3","cost_usd":0.006145,"raw_usage":{"total_tokens":2847,"prompt_tokens":563,"num_sources_used":0,"completion_tokens":63,"cost_in_usd_ticks":61449500,"prompt_tokens_details":{"text_tokens":563,"audio_tokens":0,"image_tokens":0,"cached_tokens":256},"completion_tokens_details":{"audio_tokens":0,"reasoning_tokens":2221,"accepted_prediction_tokens":0,"rejected_prediction_tokens":0}},"tokens_in":563,"tokens_out":63,"duration_ms":13950,"temperature":1.0,"reasoning_tokens":2221,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-05-24T16:33:06.796419+00:00","model_set":{"reader":"grok-4.3"},"falsifier":"Direct comparison of SMC-imputed membrane potentials against simultaneous electrophysiological recordings in the same animal, checking whether the imputed traces lie within the measurement noise of the true voltages.","supporting_citations":[],"review_version":1}