{"id":"e10bec70-72e5-43da-a2f1-95c35bcc6bcc","arxiv_id":"2509.03418","paper_version":1,"verdict":"UNVERDICTED","confidence":"HIGH","novelty_score":3.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"The paper argues that feedforward, feedback, and adaptive control, together with information theory, provide a unifying quantitative framework for biological mechanical function across scales.","lead":"This paper is a review and perspective arguing that control theory and information theory can unify how we understand mechanical regulation in biology, from molecular motors and cells to whole animals and evolution. It organizes the case for this framework across many fields and proposes it as a guide for future research and technology.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The quantitative cross-scale claim rests on a Landauer-based bound that conflates bit erasure with information extraction; without a corrected derivation, the paper's central quantitative assertion is unsupported.","rationale":"The reader's verdict of UNVERDICTED is appropriate for a perspective article: the qualitative argument is plausible, honestly hedged, and supported by a broad literature, while the article itself provides no new derivations or testable predictions. The weakest point is the quantitative bridge. The paper's only explicit quantitative connection between thermodynamics and information—the 10^9 bit/s estimate—assumes that entropy production rate limits mutual information extraction via Landauer's principle. Landauer's principle concerns the minimum dissipation of logically irreversible bit erasure, not a universal cap on information acquisition or transduction. This is not a calibration issue but a category error, and it directly undermines the 'quantitatively connect' phrasing of the central claim. The reader flagged the Landauer estimate as fragile and also emphasized the unproven control abstraction; I agree with the latter but see the former as the more decisive, testable vulnerability. Because the paper remains a perspective rather than a research claim, and because the authors themselves acknowledge the estimate is likely too generous, this concern does not change the overall UNVERDICTED assessment—it sharpens the reason why the quantitative part of the claim should not be taken as established.","tokens_in":18740,"tokens_out":6793,"duration_ms":76993,"concrete_test":"Recompute the cellular information-rate ceiling using a correct stochastic-thermodynamic relation for autonomous information flow (e.g., Barato-Hartich-Seifert or Horowitz-Esposito bipartite bound) applied to a minimal molecular sensor model with the same bulk dissipation, 10 fW/K. If the result is orders of magnitude smaller than 10^9 bit/s, or if the bound depends on unmeasured microscopic parameters (sensor copy number, coupling strength, relaxation times), the paper's quantitative claim fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim has two components: a qualitative unifying claim (control theory as a framework) and a quantitative cross-scale claim (control/information theory 'can quantitatively connect' molecular activity to macroscopic mechanical function). The qualitative claim is a reasonable perspective, supported by a broad literature. The quantitative claim is the load-bearing part, and the paper's only explicit quantitative bridge is the 10^9 bit/s estimate in the 'Control over microscopic activity' section. The text assumes I_dot <= S_dot/(k_B ln 2), citing Landauer's principle. That inequality is not a consequence of Landauer's principle: Landauer's bound applies to the heat dissipated by logically irreversible erasure of a bit, S_dot >= k_B ln 2 * R_erase, so it bounds bit-erasure rate, not the rate at which a system can extract or transmit mutual information. Measurement and transduction can in principle occur without erasure, and information rates in autonomous molecular systems are governed by bipartite fluctuation theorems with additional model-dependent terms. Thus the estimate is not merely 'likely too generous' (as the authors note) but is derived from the wrong physical quantity. No other quantitative derivation in the paper connects molecular-scale parameters to a macroscopic controller/plant description. The block-diagram mapping (Fig. 1) is introduced by analogy rather than by system identification. Consequently, the strongest version of the central claim—that the connection is quantitative—is not established; at most the paper supports a qualitative, programmatic unification.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper is an expository perspective arguing that control theory and information theory provide a unifying framework for biological mechanical function across scales. It introduces feedforward, feedback, and adaptive control architectures, reviews stochastic thermodynamics and a Landauer-based estimate of cellular information processing, discusses robophysical models of locomotion, and extends the framework to population dynamics, evolution, synthetic cells, prosthetics, and space medicine. The abstract and introduction make both a qualitative claim (control theory is a powerful unifying framework) and a quantitative claim (control and information theory can quantitatively connect molecular nonequilibrium activity to macroscopic biological mechanical function). The paper contains no new derivations; its contribution is synthetic and conceptual.","tokens_in":19066,"tokens_out":2780,"duration_ms":32133,"significance":"The qualitative thesis is valuable and timely: it draws together a broad literature and offers a plausible common language for molecular, cellular, organismal, and evolutionary regulation. The authors are commendably candid about limitations: the gigabit-per-second estimate is flagged as likely too generous, robophysical models are acknowledged to be non-biological, and the human-engineering proposal is explicitly described as facing substantial technical and ethical hurdles. These caveats strengthen the paper's credibility. However, the central quantitative claim rests on a single estimate that is derived from an incorrect application of Landauer's principle. Because no other quantitative bridge between molecular parameters and macroscopic control descriptions is supplied, the strongest version of the paper's thesis is currently unsupported. The manuscript is best viewed as a perspective that would be significantly improved by either correcting or carefully softening its quantitative framing.","major_comments":[{"comment":"The derivation of the 10^9 bit/s estimate is not a valid consequence of Landauer's principle. The text states, 'assuming an upper bound S/k = I/k_s' and obtains I_dot = 10^9 bit/s. Landauer's bound applies to the heat dissipated when a bit is erased, giving S_dot >= k_B ln 2 * R_erase; it does not bound the rate at which a system can extract or transmit mutual information. Measurement and transduction can in principle proceed without erasure, and in autonomous molecular systems the relationship between information flow and entropy production is governed by bipartite fluctuation theorems with model-dependent terms, as the paper itself notes via refs. [80,81]. Thus the estimate is not merely 'likely too generous'; it is derived from the wrong physical quantity. Since this is the only explicit quantitative bridge between molecular entropy production and information-processing rate, the pape","section":null},{"comment":"The block-diagram mapping of biological components onto controller K, plant G, sensor output y(t), and control signal u(t) is introduced by analogy, and no system identification or transfer-function-based demonstration is provided. For the quantitative cross-scale claim, it is not enough to assert that intracellular signaling, muscle actuation, and organismal movement 'can be viewed as' control loops; one needs at least one worked example in which molecular-scale parameters are quantitatively connected to a macroscopic controller/plant description. The paper offers several qualitative examples (e.g., RhoA/Rac signaling, stretch-activated channels, spindle reflexes) but no quantitative bridge. This is a load-bearing gap for the quantitative thesis, and it should be addressed by either supplying a worked quantitative case study or explicitly limiting the claim to a qualitative organization","section":null}],"minor_comments":[{"comment":"The notation 'S/k = I/k_s' is confusing: k is Boltzmann's constant and k_s = log 2 bit, but the equation as written mixes units and does not clearly define the information rate. Please rewrite with explicit definitions and use '≈' or an order-of-magnitude symbol for the 10^9 bit/s result.","section":null},{"comment":"The text refers to 'Fig. 1a', but the figure has top/center/bottom rows rather than labeled panels a/b. Please either add panel labels or correct the cross-reference.","section":null},{"comment":"The statement that LOCKR protein cages implement 'PID-like logic' is an overstatement; the cited work demonstrates de novo designed protein switches, not proportional-integral-derivative control. Please qualify this claim.","section":null},{"comment":"Several typographical and formatting issues need correction: missing spaces before references (e.g., 'Recently, attention has turned toward' and 'e.g. [77]'), 'consumes∼' without a space, and inconsistent use of italics for variables. A careful proofread is needed.","section":null}],"recommendation":"major_revision","confidential_remarks":"The paper is a perspective, not a technical contribution, so the standard of proof for its central quantitative claim should be calibrated accordingly. Still, the specific error in the Landauer application is central to the paper's quantitative thesis, and it should be corrected or the thesis softened. I would not reject the paper because the qualitative synthesis is sound and the authors have already shown willingness to acknowledge limitations. A major revision that fixes the physical derivation and reframes the quantitative claim would make the paper publishable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nQuick take: this is a readable perspective arguing that control theory—feedforward, feedback, adaptive—plus information theory can unify biological mechanics from molecular motors to organismal locomotion. The qualitative argument is plausible and well organized, but the one quantitative bridge it builds (10^9 bit/s per cell) comes from the wrong physical quantity. That undercuts the strong claim of a \"quantitative connection,\" though not the overall value of the review.\n\nWhat is genuinely useful: the paper organizes a large literature into a clean hierarchy—molecular signaling, muscle actuation, robophysics, population dynamics, evolution—under the controller/plant/sensor vocabulary. The robophysics section is the strongest: the authors know the field firsthand and are explicit that these models are not biological. They also flag limitations throughout, including the gigabit estimate itself (\"likely no\"), which is honest. The references are broad and current; the self-citations (refs 106–107) point to real experimental work, so no problem there.\n\nThe soft spots, in order. One: the quantitative claim. The paper takes the cell's entropy production rate, divides by k ln2, and calls it a bound on \"mutual information extraction rate.\" Landauer's principle bounds erasure, not measurement or transduction; the inequality as written is not a consequence of any published result, and the authors' hedge does not repair the derivation. Two: the block-diagram mapping (Fig. 1) is by analogy, not system identification. That is acceptable for a perspective, but it means the \"control across scales\" thesis is a framing, not a derivation. Three: the space-omics and human-engineering section leans toward grant-proposal prose and a trade-book endorsement, with the speculation flagged but still overlong.\n\nOverall, as a review and roadmap, it deserves a serious referee. The weak unifying thesis—that control language is useful across these scales—is likely right. The strong quantitative version is not established. Send it to peer review as a perspective that should be tightened on the Landauer step and trimmed in the applications section. The authors are thinking clearly and engaging honestly with the literature; the flaw is a specific technical misapplication, not incoherence.","headline":"A clearly written control-theory perspective whose qualitative synthesis is useful, but whose only quantitative bridge rests on a misapplied Landauer bound.","tokens_in":19520,"tokens_out":2212,"would_cite":false,"duration_ms":24783,"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":"This paper argues that control and information theory can quantitatively connect molecular-scale nonequilibrium activity to macroscopic biological mechanical function, spanning timescales from reflexes to evolution.","keywords":["control theory","information theory","biological mechanics","feedback control","stochastic thermodynamics","active matter","robophysics","adaptation"],"falsifier":"Measure the mutual information between an applied mechanical or chemical stimulus and a cell's contractile output under repeated identical trials; if the sustained information rate exceeds the cell's entropy-production budget, or if no finite-order controller can reproduce the measured input-output map, the claimed quantitative bridge fails.","tokens_in":18645,"feed_emoji":"🔄","tokens_out":6640,"duration_ms":72998,"temperature":0.7,"pith_summary":"The paper argues that control theory provides a unifying framework for biological mechanics: the same mathematics used to regulate engineered systems—feedforward commands, feedback from sensors, and adaptive updating of controller parameters—describes how cells, muscles, organisms, populations, and evolving species keep functioning far from equilibrium. Its central claim is that control and information theory can quantitatively connect microscopic nonequilibrium molecular activity to macroscopic phenomenological mechanical function, and connect short task timescales to evolutionary ones. The authors survey molecular motors, intracellular signaling, muscle activation, animal locomotion, robophysics, population dynamics, and medical and space-engineering applications to show the same loop structure recurring at each scale. If the claim is right, biological mechanical behavior becomes not an unstructured consequence of physical law but a set of regulation problems solvable by common control-theoretic tools.","feed_headline":"Control theory can tie molecular motors to whole-organism behavior","feed_subtitle":"If right, cell mechanics, locomotion, and evolution become one hierarchy of regulation problems.","key_machinery":"The central object is the closed control loop: controller K, plant G, control signal u(t), and sensory measurement y(t). Around this loop the paper organizes all scales; the quantitative bridge is supplied by the Landauer bound (minimum heat kT ln2 per erased bit) and bipartite stochastic-thermodynamics descriptions of entropy production and mutual information, used to compute information rates from metabolic dissipation.","core_discovery":"The paper proposes that a controller K issuing control signals u(t) to a physical system G, with sensors returning outputs y(t) to close the loop, is the common architecture underlying biological mechanical function from molecular to organismal scales. It claims that this architecture, together with information-theoretic quantities, allows quantitative connections: entropy production constrains information processing, and metabolic free-energy dissipation can be converted into an upper bound on the rate of mutual-information extraction—roughly 10^9 bits per second for an average human cell. The paper then extends the same control picture upward in timescale: adaptive control models robustnes","pith_inferences":["The paper's control-loop vocabulary could be tested by asking whether observed sensory histories and actuation signals in single cells admit a consistent input-output controller; if they do not, the framework would remain a metaphor rather than a quantitative description.","The 10^9 bit/s estimate suggests a concrete empirical program: measure mutual information between a defined stimulus and a mechanical response in single cells and compare it with the entropy budget.","The framework implies a robustness principle—sensing must be fast relative to disturbances—so one could look for cells or organisms that tune their sensing bandwidth to the disturbance spectrum.","Treating 'mechanical intelligence' as embodied feedforward control could reconcile self-organization descriptions with control descriptions, making the two complementary rather than competing."],"forward_implications":["If the claim is right, experiments on molecular motors, cells, and animals can share the same control-theoretic vocabulary, so insights from one scale transfer to another.","Metabolic entropy production becomes a measurable budget for biological computation: the paper's estimate gives an upper information-processing rate near 10^9 bits per second for an average cell.","Adaptive control offers a quantitative model of population dynamics and evolution as robust adjustment to slow environmental drift, potentially improving predictions where classical models fail.","Robophysical systems can test how thermodynamic efficiency, control effort, and information transmission trade off in mechanical tasks, because their internal states and energy use are directly measurable.","Framing medical interventions and prosthetics as feedback controllers turns therapy into a closed-loop regulation problem with design rules from control theory."],"supporting_citations":[{"why":"Supplies the control-theoretic formalism of plants, controllers, sensors, feedback, and adaptive control that the paper applies across scales.","marker":"[21]"},{"why":"Provides the stochastic-thermodynamics relationships between work, entropy, free energy, mutual information, and memory that anchor the molecular-scale bridge.","marker":"[20]"},{"why":"Gives the Landauer bound of k log 2 per erased bit, the basis for the paper's estimate of cellular information-processing rates.","marker":"[76]"},{"why":"Precedent for applying Landauer's bound to biological information processing, extending it beyond computer memory.","marker":"[78]"},{"why":"Formalizes bipartite stochastic systems, allowing energy, work, and information exchanges between coupled subsystems to be described.","marker":"[80]"},{"why":"Provides a continuous-information-flow formalism for coupled thermodynamic systems, used to connect molecular activity to control.","marker":"[81]"},{"why":"Applies feedback-control thinking to biological locomotion and behavior, supporting the organism-scale extension of the framework.","marker":"[17]"},{"why":"Reviews feedback control of active matter, supporting the paper's claim that control methods apply to soft and living materials.","marker":"[23]"},{"why":"Treats locomotion over complex terrain as a problem of information transmission, a load-bearing example for the robophysics and information-theory connection.","marker":"[96]"}],"fun_headline_variants":["One control loop rules biology's scales","Control theory bridges molecule to migration","Biological feedback: from motors to evolution","How cells and organisms share one control design","Universal feedback architecture in living systems"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"The argument depends on the premise that biological systems genuinely instantiate a control loop—a controller, a plant, and information-bearing sensory feedback—rather than merely being describable as if they did; the cell-scale information-rate estimate additionally assumes Landauer's bound applies to whole-cell information processing.","fun_headline_variants_meta":{"raw":{"variants":["One control loop rules biology's scales","Control theory bridges molecule to migration","Biological feedback: from motors to evolution","How cells and organisms share one control design","Universal feedback architecture in living systems"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000192,"raw_usage":{"total_tokens":1133,"prompt_tokens":643,"completion_tokens":490,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":387,"completion_tokens_details":{"reasoning_tokens":429}},"tokens_in":387,"tokens_out":490,"duration_ms":5590,"temperature":1.0,"reasoning_tokens":429,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T10:54:14.688781+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the mutual information between an applied mechanical or chemical stimulus and a cell's contractile output under repeated identical trials; if the sustained information rate exceeds the cell's entropy-production budget, or if no finite-order controller can reproduce the measured input-output map, the claimed quantitative bridge fails.","supporting_citations":[{"cited_title":"Irreversibility and Heat Generation in the Computing Process","cited_arxiv_id":null,"evidence_quote":"Gives the Landauer bound of k log 2 per erased bit, the basis for the paper's estimate of cellular information-processing rates."},{"cited_title":"Ouldridge, Christopher C","cited_arxiv_id":null,"evidence_quote":"Precedent for applying Landauer's bound to biological information processing, extending it beyond computer memory."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Formalizes bipartite stochastic systems, allowing energy, work, and information exchanges between coupled subsystems to be described."},{"cited_title":"Horowitz and Massimiliano Esposito","cited_arxiv_id":null,"evidence_quote":"Provides a continuous-information-flow formalism for coupled thermodynamic systems, used to connect molecular activity to control."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Treats locomotion over complex terrain as a problem of information transmission, a load-bearing example for the robophysics and information-theory connection."}],"review_version":1}