{"id":"44a25d91-0288-4fff-80d3-d5869b768f5e","arxiv_id":"2608.03712","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"high","formal_verification":"none","parameter_count":7,"one_line_summary":"A nine-equation conductance-based model connects amygdala threat processing to cardiovascular responses, producing strong, moderate, and weak regulatory regimes and claiming validation on three datasets.","lead":"This paper builds a compact computer model linking amygdala activity, stress-related psychological traits, and heart rate and blood pressure changes. It aims to help interpret wearable stress data through a mechanistic, interpretable framework for anxiety and depression.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Strong-regime readout contradicts Eq. (9): quoted Vm≈−50 mV and θ≈−50.25 mV give A≈0.51, not the reported A<0.2 / mean 0.060, so the central regime-separation claim is internally inconsistent.","rationale":"I focused on the strong-regime readout because it attacks the central quantitative claim rather than only the interpretation of one variable. The h(t) sign issue raised by the reader is real: Eq. (3) makes h multiply inhibitory conductance and Eq. (7) makes h_inf increase with voltage, so h is an activity-dependent enhancer of inhibition, i.e., an adaptation process, not the sensitization substrate claimed in §II-B and §IV-B. However, §V-B's ablation shows h is not the principal driver of regime separation, so that concern, while undermining contribution 1, does not by itself destroy the regime-separation result. The numerical contradiction in §IV-C does: the reported voltages and thresholds, fed through Eq. (9) with u = 0.2, produce A ≈ 0.51 in the strong regime, directly contradicting the claims that A stays below 0.2 and that the threshold is never crossed. The downstream cardiovascular outputs are monotone transforms of A, so if A is wrong the HR/BP regime differences are unsupported. The external-validation concerns are also valid but secondary, since they affect the strength of the external-evaluation claim rather than the internal model behavior. My recommendation is REJECT, matching the reader's verdict: as written, the paper's central quantitative results are not reproducible from its own equations, and the absence of code and complete parameters prevents external resolution of the inconsistency.","tokens_in":17668,"tokens_out":8199,"duration_ms":86731,"concrete_test":"Compute Eq. (9) with u = 0.20 mV^−1 at the strong-regime values quoted in §IV-C (Vm ≈ −50 mV, θ ≈ −50.25 mV): σ(0.2·(+0.25)) ≈ 0.512, already above the claimed A < 0.2 threshold. Then run the Section III protocol with Table I/II parameters and extract the strong-regime mean and peak A. If the simulated mean A exceeds 0.2 (or even only the peak A exceeds 0.5), the reported A = 0.060 and never-crosses-threshold statement are unsupported; the authors would need to release code/parameter values or correct the quoted voltage and threshold numbers.","verdict_should_be":"REJECT","load_bearing_attack":"Section IV-C's strong-regime numbers are internally inconsistent with Eq. (9). The text states that in the strong regime peak Vm ≈ −50 mV and θ ≈ −50.25 mV, yet also that Vm−θ remains negative and A stays below approximately 0.2, with a reported mean A = 0.060. With u = 0.20 mV^−1, Eq. (9) gives A = σ(0.2·(+0.25 mV)) ≈ 0.512 at the quoted peak; to obtain A = 0.060 the decision variable would need to be Vm−θ ≈ −13.8 mV, not +0.25 mV. Even allowing for loose rounding, the sign is wrong: a +0.25 mV excursion places the readout at the sigmoid midpoint, not in the sub-0.2 no-threat range. This is not cosmetic: the reported strong/moderate/weak mean A values (0.060/0.241/0.475), the never-crosses-threshold claim, and the downstream hypothalamic and cardiovascular separation all inherit this readout. Since no code or complete parameter set is provided, the results cannot be checked externally. This internal contradiction directly undermines the central claim of robust regime separation.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a nine-equation conductance-based model of an amygdala–hypothalamus–cardiovascular loop. A single amygdala neuron receives sensory drive gated by prefrontal regulatory strength; an HH-style gating variable h is reinterpreted as a slow, history-dependent internal state; an adaptive threshold θ depends linearly on coping capacity, perceived stress, and prefrontal strength; a graded threat readout A feeds a hypothalamic integrator with baroreflex feedback; and the resulting state drives HR/SBP/DBP excursions. The authors report three regulatory regimes (strong, moderate, weak), robustness to noise and parameter perturbation, an ablation study identifying the adaptive threshold as the principal mechanism, and external validation against three datasets.","tokens_in":17947,"tokens_out":7617,"duration_ms":84464,"significance":"If the numerical and conceptual issues were resolved, this compact model could provide a useful, interpretable bridge between psychometric constructs and autonomic signals, with potential applications in digital phenotyping. The paper's strengths are that it states all equations explicitly, gives parameter tables, includes robustness and ablation analyses, and benchmarks against published stress-physiology ranges and external datasets. However, the strong-regime description is internally inconsistent with the stated readout equation, the slow-state h(t) has the wrong sign for the claimed sensitization mechanism, the psychometric-to-parameter mapping is not actually specified, and the external validation is in part circular. These are load-bearing problems for the paper's central claims.","major_comments":[{"comment":"The strong-regime numerical description is internally inconsistent. The text states peak Vm ≈ −50 mV and θ ≈ −50.25 mV, so Vm−θ ≈ +0.25 mV, while also stating that Vm−θ remains negative and A stays below approximately 0.2, with mean A = 0.060. With u = 0.20 mV⁻¹, Eq. (9) gives A = σ(0.20·0.25) ≈ 0.512, not 0.060; obtaining A = 0.060 would require Vm−θ ≈ −13.8 mV. This sign error propagates to y(t) in Eq. (10), the hypothalamic state in Eq. (11), and the reported cardiovascular excursions in Fig. 4. Since no code or complete parameter set is provided, the results cannot be independently checked.","section":"Section IV-C and Eq. (9)"},{"comment":"The slow state h(t) has the wrong sign/interpretation for the claimed sensitization mechanism. In Eq. (3), h multiplies the inhibitory synaptic conductance, so for Vm > Esyn = −70 mV, larger h increases a hyperpolarizing current. Equation (7) makes h_inf increase with Vm, so activation builds up inhibition. This is a negative-feedback adaptation variable, not a substrate for the stated 'sustained sensitization' or 'cross-pulse accumulation' that heightens threat responses (Sections II-B and IV-B). If h is instead intended to encode disinhibition, either Eq. (3) or the voltage dependence of h_inf must be reversed. The ablation in Section V-B shows h is not the principal regime separator, but Contribution 1 still assigns it a central mechanistic role.","section":"Section II-B, Eqs. (3) and (7)"},{"comment":"The claimed psychometric-to-parameter mapping is not specified. The paper states that C, P, and RPFC are operationalized through CD-RISC, PSS-10, and a neuroimaging-derived prefrontal index, but it never provides the transformation from raw scale scores to the bounded [0,1] modulator values. Table II simply assigns C = 0.75/0.50/0.25, P = 0.25/0.50/0.75, and RPFC = 0.75/0.50/0.25 by regime. Without this mapping, the model cannot be driven by clinical questionnaires, and the 'clinically grounded modulators' claim is untestable.","section":"Section II-B and Table II"},{"comment":"The external validation is not genuinely independent. For the first dataset, the model is given 'quartile statistics giving baseline inputs per regime' from the same dataset before simulated peaks are compared with observed peaks. This means resting cardiovascular values (HR_rest, SBP_rest, DBP_rest) and possibly gains are informed by the data being predicted. The reported 3.7% MAPE is therefore at least partly an in-sample calibration result, not an out-of-sample validation. The paper does not report which parameters were set from the data or how g_hr and g_sbp were chosen. Dataset 2 is a single-subject monotonic trend and dataset 3 provides directional LME results; these support qualitative ordering but not quantitative model predictions.","section":"Section V-D and Eq. (17)"}],"minor_comments":[{"comment":"Table I omits several parameters used in the equations: resting cardiovascular values (HR_rest, SBP_rest, DBP_rest), baroreflex parameters (η_b0, φ_0, δ), and the initial-condition constants hy0 and n(0) are only given in the text. A reader cannot reproduce the simulations from the paper alone.","section":"Section III-A / Table I"},{"comment":"Section III-B sets the simulation window to T = 5000 ms, but Section IV-D refers to an '8000 ms observation window'. This inconsistency should be reconciled.","section":"Section III-B vs Section IV-D"},{"comment":"The claim that the fraction of time with A > 0.5 increases by 'roughly two orders of magnitude' is not accompanied by the underlying fractions; numerical values would make the comparison more transparent.","section":"Section IV-C"},{"comment":"Reference [28] contains the author name 'LaR. Lindsey', which appears to be a typographical error for 'R. Lindsey'.","section":"References"}],"recommendation":"reject","confidential_remarks":"The manuscript has a potentially interesting modeling concept, but as submitted the central numerical claims are internally inconsistent, the slow-state mechanism does not implement the claimed sensitization, and the external validation is partly circular. These are not local presentation issues; they affect the paper's main conclusions. I would encourage the authors to revise the model and provide code and full parameter settings before resubmission."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a paper with a real kernel—a compact closed-loop amygdala–hypothalamus–baroreflex model with an adaptive threshold—but the strong-regime results in Section IV-C contradict the paper's own Eq. (9), and the external validation is partly calibration wearing validation clothes. I would send it to referees, but I would expect them to send it back for major revision.\n\nWhat's genuinely good: the model is small and interpretable, the idea of mapping CD-RISC/PSS-10-style modulators onto threshold dynamics is sensible, and the ablation in Section V-B is the strongest part. It cleanly shows that the adaptive threshold, not the slow state h(t), does most of the work separating regimes. The robustness sweeps over noise and ±20–25% parameter perturbations are a reasonable check, and the continuous-regulation-axis framing is consistent with the transdiagnostic literature.\n\nNow the problems, in order of severity.\n\nFirst, the arithmetic in Section IV-C does not close. The text says strong-regulated peaks sit near Vm ≈ −50 mV with threshold θ ≈ −50.25 mV. That puts Vm − θ at +0.25 mV. With u = 0.20 mV⁻¹, Eq. (9) gives A ≈ σ(0.05) ≈ 0.51, not the reported mean A = 0.060 or the claim that A stays below 0.2 and never crosses threshold. To get A = 0.060 you would need Vm − θ ≈ −13.8 mV. This is not rounding; the sign is backward. Since the mean A values (0.060 / 0.241 / 0.475) are the headline evidence for regime separation, the paper's central quantitative claim is currently unsupported.\n\nSecond, the slow state h(t) is described as the substrate for \"sustained sensitization,\" but it multiplies inhibitory synaptic conductance and its steady state h∞(Vm) increases with voltage. That is an activity-dependent inhibition enhancer, which should reduce responsiveness, not sensitize it. The ablation shows h is not the main driver, so the model may survive without this interpretation, but as written the claimed mechanism is wrong.\n\nThird, the external validation in Section V-D is not as independent as claimed. Dataset quartiles are fed in as baseline inputs, and the output gains g_hr and g_sbp are never reported. With hand-set gains and no code, the 2–5% errors are consistency checks, not predictions.\n\nWhat is here is publishable after serious revision: fix the strong-regime numbers and report the actual Vm−θ traces, either fix the h(t) interpretation or drop the sensitization language, report all gains and baselines, and re-frame the dataset comparisons as plausibility checks. The topic matters and the ablation is a legitimate contribution. Send it to peer review, but flag the Section IV-C contradiction for referees.","headline":"A promising compact model with a real kernel, but the reported strong-regime numbers contradict the paper's own Eq. (9), and the external validation is partly calibration; worth refereeing, not accepting as-is.","tokens_in":18543,"tokens_out":4117,"would_cite":false,"duration_ms":43728,"reading_group":"maybe","serious_thinker":"no","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"A nine-equation amygdala–hypothalamus loop reproduces three stress-regulatory regimes and links psychometric scores to cardiovascular output.","keywords":["amygdala","anxiety","depression","computational modeling","cardiovascular response","baroreflex","Hodgkin-Huxley","digital phenotyping"],"falsifier":"Record an amygdala neuron's membrane potential during repeated depolarizing pulses with inter-pulse intervals shorter than the 220 ms time constant of the weak regime: if successive pulses produce progressively smaller responses, the h(t) dynamics act as inhibition, contradicting the model's claimed cross-pulse sensitization. Alternatively, run the model with h(t) clamped to a constant and check whether regime separation persists; the paper's own ablation suggests it does, which any reader can verify from the nine equations.","tokens_in":17410,"feed_emoji":"🧠","tokens_out":7733,"duration_ms":81955,"temperature":0.7,"pith_summary":"This paper tries to show that a single, compact circuit model—nine equations describing an amygdala neuron, a hypothalamic integrator, and a baroreflex feedback loop—can account for how anxiety and depression shape the body's stress response. The authors extend the standard Hodgkin-Huxley neuron formalism with three clinically meaningful knobs: coping capacity, perceived stress load, and prefrontal regulatory strength. These knobs move an adaptive threshold and a slow internal state so that healthy, subclinical, and clinical individuals become three operating regimes of the same closed-loop system, not separate mechanisms. The model generates heart-rate and blood-pressure trajectories that stay inside published stress-physiology ranges and, the authors report, match three independent real-world datasets. The central insight is that an adaptive, state-dependent threshold, not the slow sensitization variable, is the load-bearing mechanism that separates the regimes.","feed_headline":"Anxiety and depression show up as three regimes in one circuit model","feed_subtitle":"Nine equations tie coping, stress, and prefrontal control to heart-rate and blood-pressure responses seen in real data.","key_machinery":"The adaptive threshold θ(t) = θ0 + Kc·C − Kp·P + Kr·RPFC, read out through a graded sigmoid A = σ(u(Vm−θ)), is the load-bearing device. It converts shifting psychometric state into a decision variable—membrane voltage minus threshold—whose sign and magnitude determine threat acknowledgment, hypothalamic drive, and ultimately heart-rate and blood-pressure output. The slow gating variable h(t), a reinterpreted Hodgkin-Huxley inactivation variable, multiplies inhibitory conductance and is described as accumulating sensitization across pulses, but the paper's ablation shows it is mainly an amplifier rather than the source of regime separation.","core_discovery":"On the paper's terms, the discovery is that a single amygdala–hypothalamus–cardiovascular loop, with no architecture switching, produces the full spectrum of threat responses by shifting two quantities: an adaptive threshold θ(t) that rises with coping and prefrontal control and falls with perceived stress, and a graded threat-acknowledgment signal A = σ(u(Vm−θ)). In the strong regime the membrane voltage never crosses threshold, in the moderate regime it crosses transiently, and in the weak regime it stays above threshold for extended periods; the resulting hypothalamic drive and baroreflex feedback yield heart-rate and blood-pressure excursions that increase and recover more slowly as regu","pith_inferences":["Editorial extension: because the threshold is linear in the three modulators, the model predicts a composite score (resilience minus perceived stress plus prefrontal index) should track cardiovascular reactivity; this is testable in the same datasets without collecting new physiology.","Editorial extension: the threshold-ablation result implies that interventions which raise the adaptive threshold—such as strengthening prefrontal regulation—should shift a weak-regime profile back toward moderate or strong, a prediction that can be checked in silico before any clinical trial.","Editorial extension: cohort-level validation supports population-level stratification and digital phenotyping more than person-specific prediction; moving to digital twins will require inverse estimation of the modulators from wearable data and adding the HPA axis and inflammatory signalling that the paper explicitly excludes.","Editorial extension: the slow internal state's sign is an empirical handle—if stressors spaced closer than its roughly 220 ms time constant do not produce cross-pulse accumulation under the stated equations, the sensitization interpretation should be revised."],"forward_implications":["A single closed-loop architecture, without switching, produces healthy, subclinical, and clinical cardiovascular stress profiles from the same sensory input.","The adaptive threshold is the principal mechanism for quantitative regime separation; the slow internal state mostly amplifies differences already set by the threshold.","Simulated heart-rate and blood-pressure peaks and recovery times stay inside published acute-stress ranges, so the model can serve as a mechanistic map from psychometric scores to wearable-observed autonomic trajectories.","The model reproduces real-world cardiovascular magnitudes (about 3.7% mean absolute percentage error on the truck-driver dataset), monotonic heart-rate increases with stress burden, and state-dependent hemodynamic shifts in the stroke-rehabilitation cohort.","Variability and parameter-perturbation analyses preserve regime ordering, supporting the claim that the three regimes are stable operating states rather than artifacts of a single parameter choice."],"supporting_citations":[{"why":"Supplies the Hodgkin-Huxley conductance formalism on which the amygdala neuron equation is built.","marker":"[20]"},{"why":"Provides the canonical statement of Hodgkin-Huxley theory that the gating-variable reinterpretation extends.","marker":"[21]"},{"why":"Operationalizes coping capacity C through the Connor-Davidson Resilience Scale.","marker":"[22]"},{"why":"Operationalizes perceived stress P through the Perceived Stress Scale-10.","marker":"[23]"},{"why":"Supplies the clinical premise linking stress, amygdala reactivity, and cardiovascular disease, plus the empirical ranges for stress physiology.","marker":"[2]"},{"why":"Motivates the inhibitory conductance with evidence on amygdala interneurons and intercalated cells.","marker":"[7]"},{"why":"Supplies the cortical-hypothalamic integrative and baroreflex pathway used for the autonomic output.","marker":"[10]"},{"why":"Occupational-stress dataset (nurse heart-rate recordings) used for external validation of continuous stress load.","marker":"[30]"},{"why":"Stroke-rehabilitation cohort with hemodynamic changes across functional states used for validation of stimulus-response and recovery.","marker":"[31]"},{"why":"Large truck-driver mental-health dataset used to validate cardiovascular magnitudes under inferred coping and stress levels.","marker":"[36]"}],"fun_headline_variants":["Three regulatory states explain anxiety and depression in one circuit model","Anxiety and depression: three regimes from one amygdala-cardiovascular loop","One compact model, three threat-response states for anxiety and depression","Stress, coping, and prefrontal control: one loop yields three regimes","Conductance model ties anxiety and depression to three autonomic states"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The load-bearing premise is the model's interpretation of its slow internal state h(t) (Section II-B) as a substrate for stress sensitization: h multiplies inhibitory conductance and rises with membrane voltage, so if cumulative activity actually inhibits rather than sensitizes the amygdala, the paper's sensitization claim collapses even though the adaptive threshold may still separate regimes.","fun_headline_variants_meta":{"raw":{"variants":["Three regulatory states explain anxiety and depression in one circuit model","Anxiety and depression: three regimes from one amygdala-cardiovascular loop","One compact model, three threat-response states for anxiety and depression","Stress, coping, and prefrontal control: one loop yields three regimes","Conductance model ties anxiety and depression to three autonomic states"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000858,"raw_usage":{"total_tokens":3583,"prompt_tokens":784,"completion_tokens":2799,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":528,"completion_tokens_details":{"reasoning_tokens":2711}},"tokens_in":528,"tokens_out":2799,"duration_ms":19757,"temperature":1.0,"reasoning_tokens":2711,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T14:16:17.594757+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Record an amygdala neuron's membrane potential during repeated depolarizing pulses with inter-pulse intervals shorter than the 220 ms time constant of the weak regime: if successive pulses produce progressively smaller responses, the h(t) dynamics act as inhibition, contradicting the model's claimed cross-pulse sensitization. Alternatively, run the model with h(t) clamped to a constant and check whether regime separation persists; the paper's own ablation suggests it does, which any reader can verify from the nine equations.","supporting_citations":[{"cited_title":"The effect of sodium ions on the electrical activity of the giant axon of the squid,","cited_arxiv_id":null,"evidence_quote":"Supplies the Hodgkin-Huxley conductance formalism on which the amygdala neuron equation is built."},{"cited_title":"The Hodgkin-Huxley theory of the action potential,","cited_arxiv_id":null,"evidence_quote":"Provides the canonical statement of Hodgkin-Huxley theory that the gating-variable reinterpretation extends."},{"cited_title":"Development of a New Resilience Scale: The Connor-Davidson Resilience Scale (CD-RISC),","cited_arxiv_id":null,"evidence_quote":"Operationalizes coping capacity C through the Connor-Davidson Resilience Scale."},{"cited_title":"A global measure of per- ceived stress,","cited_arxiv_id":null,"evidence_quote":"Operationalizes perceived stress P through the Perceived Stress Scale-10."},{"cited_title":"Stress and cardiovascular disease: an update,","cited_arxiv_id":null,"evidence_quote":"Supplies the clinical premise linking stress, amygdala reactivity, and cardiovascular disease, plus the empirical ranges for stress physiology."},{"cited_title":"Neuronal circuits for fear and anxiety,","cited_arxiv_id":null,"evidence_quote":"Motivates the inhibitory conductance with evidence on amygdala interneurons and intercalated cells."},{"cited_title":"Cortical–Hypothalamic Integration of Autonomic and Endocrine Stress Responses,","cited_arxiv_id":null,"evidence_quote":"Supplies the cortical-hypothalamic integrative and baroreflex pathway used for the autonomic output."},{"cited_title":"A multimodal sensor dataset for continuous stress detection of nurses in a hospital,","cited_arxiv_id":null,"evidence_quote":"Occupational-stress dataset (nurse heart-rate recordings) used for external validation of continuous stress load."},{"cited_title":"Cerebral Flow Velocities During Daily Ac- tivities Depend on Blood Pressure in Patients With Chronic Is- chemic Infarctions,","cited_arxiv_id":null,"evidence_quote":"Stroke-rehabilitation cohort with hemodynamic changes across functional states used for validation of stimulus-response and recovery."},{"cited_title":"Factors Associated with Common Mental Disorders in Truck Drivers - figshare - NLM Dataset Catalog,","cited_arxiv_id":null,"evidence_quote":"Large truck-driver mental-health dataset used to validate cardiovascular magnitudes under inferred coping and stress levels."}],"review_version":1}