{"id":"587dcdb9-1fe3-4c0d-8745-be6a2ac01fd9","arxiv_id":"2507.01684","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Chlamydomonas cells swimming away from two competing light beams follow an intensity-weighted average direction, and a broad distribution of eyespot offsets explains why some cells instead commit to one beam and switch stochastically.","lead":"Using two crossing light beams, researchers measured how swimming algae choose a direction when the two beams pull in different directions. The cells steer along an intensity-weighted average of the two light directions, and the paper explains why some cells instead commit to one beam at large angles.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The offset-distribution model used to explain the three-peaked distribution is never checked for consistency with the tangent law for the same (δ, η) range, so the population average may not equal Eq. (1).","rationale":"The reader's weakest assumption concerns additivity of the two light stimuli in the photoreceptor signal. While that assumption is indeed load-bearing for the derivation of Eq. (3), the empirical tangent law already provides strong indirect support for additivity across the tested η range. The more pressing, and less examined, risk is internal consistency between the two explanatory frameworks: the zero-offset model that yields the tangent law and the offset-distribution model introduced to explain the three-peaked distribution at large δ. Because the offset distribution is broad (σ_γ ~ 0.2 rad) and affects the stability of fixed points, the ensemble average over γ is not trivially equal to the γ=0 tangent law, especially for η ≠ 1. The paper's own simulation (Fig. 3) is only shown for η=1, 2δ=162°, so this consistency has not been demonstrated. A direct simulation check would settle the issue. This does not invalidate the empirical data, but it determines whether the model can simultaneously explain both the tangent law and the trimodality. The reader's verdict of CONDITIONAL remains appropriate, with this additional condition.","tokens_in":13743,"tokens_out":21965,"duration_ms":253525,"concrete_test":"Using the model described in the paper (Eqs. (6), the Langevin dynamics with P ≈ S, and γ drawn from the Gaussian q(h) with parameters in Fig. 3(c)), simulate N swimmers for each (δ, η) pair in the tangent-law dataset (Fig. 1(c)), including δ up to 67.6° and η from about 0.5 to 15. Compute the steady-state ensemble average ⟨φ⟩ and compare to Eq. (1). If ⟨φ⟩ deviates outside the experimental error bars for any data point, the offset model is inconsistent with the tangent law; if it matches, the concern is resolved.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The paper's central claim is the tangent law, Eq. (1), for the ensemble-averaged swimming direction. This law is derived for a cell with zero eyespot offset (γ=0) in the End Matter. To explain the three-peaked distribution at large δ, the paper introduces a population distribution of eyespot offsets with σ_h = 1.0 µm (σ_γ ≈ 0.2 rad) and simulates the Langevin dynamics (Eq. (6) plus noise). The model predicts that cells with negative γ have stable fixed points near ±δ, while positive γ cells follow the tangent law. The ensemble average over the measured γ distribution is therefore not automatically equal to the γ=0 fixed point; for η ≠ 1 the side peaks are asymmetric and will bias ⟨φ⟩ away from Eq. (1). The paper never compares the model's predicted ⟨φ⟩ with the tangent law for the same (δ, η) grid as Fig. 1(c). Without this check, the two main results — the tangent law and the offset-induced trimodality — could be mutually inconsistent, and the claim that the tangent law is the population-level decision rule is not supported by the model.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports experiments in which negatively phototactic Chlamydomonas reinhardtii cells are exposed to two mutually inclined collimated light beams with independently adjustable intensities. The authors find that the population-averaged swimming direction follows a 'tangent law', Eq. (1), which states that the cell swims along the intensity-weighted vector average of the two light propagation directions. They derive this law as the steady-state fixed point of an adaptive phototaxis model under an additivity assumption for the photoreceptor signal, and show that fast-turn experiments are consistent with the fast-response limit of that model. At large angular separations the population splits into three subpopulations: cells swimming away from one light, cells following the tangent law, and cells stochastically switching between these directions. The authors attribute this trimodality to a measured population-level distribution of photoreceptor (eyespot) offsets from the equatorial plane and support this with Langevin simulations using the measured offset distribution.","tokens_in":14019,"tokens_out":6889,"duration_ms":79308,"significance":"If the tangent law is correct, it provides a remarkably simple, parameter-free geometrical rule for phototactic decision-making in a unicellular organism, and it offers a link between line-of-sight photoreception and gradient-climbing descriptions of phototaxis. The paper combines careful single-cell tracking, an analytic derivation from a published adaptive model, an independent measurement of the eyespot-offset distribution, and publicly archived data on Zenodo, all of which are strengths. However, the consistency between the offset-dependent model used to explain the trimodal distribution and the tangent law is not demonstrated, and the two-light derivation is only sketched, so the significance as a self-contained theoretical statement is presently limited.","major_comments":[{"comment":"The paper never checks whether the population-averaged swimming direction predicted by the offset-dependent model, using the measured eyespot-offset distribution, reproduces the tangent law Eq. (1) over the same (δ, η) grid as Fig. 1(c). For η ≠ 1, the model's stable fixed points are asymmetric (cells with negative γ swim near ±δ, while cells with positive γ follow the tangent law), so the ensemble average ⟨φ⟩ will generally differ from the γ=0 fixed point. Please provide a supplementary comparison of the model's ⟨φ⟩ with Eq. (1) and with the experimental data for representative δ values (e.g., 12.5°, 38.4°, 67.6°) and a range of η. If deviations are present, quantify whether they are within the experimental error bars; if the model is intended only for δ near 90°, state the validity range of the tangent-law claim explicitly.","section":"Phototaxis at large angular light separation (Eq. (6), Fig. 3)"},{"comment":"The derivation of Eq. (3) from the two-light signal S = P*[η J+ H(J+) + J− H(J−)] is not actually presented. The End Matter derives the averaged dynamics only for a single light source and then states that the generalization to two lights is 'straightforward'. Please show the averaging explicitly, including the piecewise treatment of the Heaviside shading in the three angular regions (I, II, III) described in the main text, and state the conditions under which the averaging holds (additivity, no saturation, no cross-inhibition). This is needed to verify that the steady state of Eq. (3) is indeed the tangent law Eq. (1).","section":"End Matter, final paragraph"},{"comment":"No goodness-of-fit statistics are provided for the tangent law. Because Eq. (1) has no free parameters, the claim that the data are 'well-fit' should be substantiated with residuals, RMS angular deviation, or χ² per degree of freedom for each δ, and the increasing scatter at δ=67.6° near η→1 should be discussed quantitatively. Without this, the empirical support for the central claim is stated only visually.","section":"Fig. 1(c), Sec. 'Experimental determination of the angles' (Supplemental S2B)"},{"comment":"The parameters used in the Langevin simulations are not reported: the value of the scaled rotational diffusion constant D̃r (or Dr), the photoresponse amplitude P*, the number of swimmers N, the integration time, and the numerical scheme are all missing. This makes the comparison in Fig. 3(d) and the effective free energy in Fig. 3(e) irreproducible and prevents the reader from assessing parameter sensitivity. Please state the parameter values and, ideally, show how the peak positions and weights depend on Dr and P*.","section":"Fig. 3(d,e), numerical simulations"}],"minor_comments":[{"comment":"The word 'functios' should be 'function'.","section":"End Matter, after Eq. (17b)"},{"comment":"'Savitzy-Golay' is a misspelling of 'Savitzky-Golay'.","section":"Supplemental S1"},{"comment":"The caption says '(c) Enlargement of the first 10 Hz of the spectra shown in (c)', but it should refer to the spectra shown in (b).","section":"Supplemental Fig. S1 caption"},{"comment":"The Heaviside step function H is used without definition; please define H(x) explicitly (e.g., H(x)=1 for x>0, 0 otherwise) at first use.","section":"Main text, Eq. (3) and Eq. (6)"},{"comment":"The URL contains the placeholder 'xxx.xxxxxx'; the actual DOI or arXiv link should be provided.","section":"Reference [25]"},{"comment":"The independent variable is denoted inconsistently as both T and t in the same equation; make the notation uniform.","section":"End Matter, Eq. (19)"},{"comment":"The caption states that the experimental data are shown for δ increasing upwards, but it is unclear whether the plotted angle is φ* or φ*−δ; please clarify the ordinate.","section":"Fig. 1(c)"}],"recommendation":"major_revision","confidential_remarks":"The manuscript presents an appealing and potentially important result, and the experimental data appear to be of high quality. The main weaknesses are the missing consistency check between the offset-dependent model and the tangent law, the incomplete two-light derivation, and the absence of simulation parameters. These issues are fixable within the manuscript's scope, but they are load-bearing for the central claims. I recommend major revision rather than rejection."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know about this one for the experimental result, not for the theory. The two-beam tangent law, Eq. (1), is a genuinely new and simple empirical rule: negatively phototactic cells swim along the intensity-weighted vector average of the two light propagation directions. The data in Fig. 1(c) support it across five half-angles and a wide range of intensity ratios, with no free parameters in the law itself. The measurement protocol is careful—angle calibration from single-beam swimming, LED power calibration in the supplement, and open data on Zenodo. That alone makes the paper worth a serious look.\n\nThe eyespot-offset story is also interesting. Measuring the distribution of photoreceptor offsets and using it to explain the three-peaked swimming-direction distribution at large angular separation is a plausible mechanism, and the simulated Q(phi) in Fig. 3(d) does resemble the experimental one. Credit where due: the imaging and tracking work is solid, and the switch-reorientation data match the adaptive model with a single fitted P*.\n\nNow the soft spots, in order of importance. First, the stress-test note is right: the paper never checks whether the offset-distribution model, when ensemble-averaged over the measured gamma distribution, reproduces Eq. (1) on the same (delta, eta) grid as Fig. 1(c). The side peaks in the model are asymmetric for eta != 1, and a nonzero mean offset (0.15 microns) could bias <phi> away from the tangent law. This is a direct consistency check that should take one supplementary figure, and its absence is a real gap between the two main claims. Second, Eq. (3) is presented as the two-light generalization of the single-source derivation, but the End Matter only derives the single-light case and says the generalization is straightforward. It probably is, but the averaging over the spin phase with the Heaviside shading functions, especially near delta ~ 90 degrees, needs to be shown rather than asserted. Third, the stochastic model's rotational diffusion coefficient Dr is not specified; the reader has to hunt for it or take it as a hidden fit. That is minor compared to the first two.\n\nNone of this sinks the central experimental claim. The tangent law is an independent measurement, not an output of a tuned model, so the circularity burden is low. But the paper as written overclaims that the model explains the data it was designed to match, without checking the quantitative link between the two. The fixes are straightforward.\n\nWho is this for? People working on microbial phototaxis, active matter with multiple cues, and decision-making in aneural organisms. It deserves a serious referee, not a desk rejection. My recommendation: send it out, and require the authors to add the two-light derivation and the ensemble-averaged tangent-law consistency check.","headline":"The tangent law is a clean, parameter-free experimental result for Chlamydomonas; the paper's weak spot is the asserted two-light model and a missing consistency check between the offset-distribution model and the tangent-law data.","tokens_in":14520,"tokens_out":1956,"would_cite":true,"duration_ms":26046,"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":"Given two crossing light beams, negatively phototactic Chlamydomonas cells swim along the intensity-weighted average of the two beam directions, a geometric 'tangent law' that is a fixed point of an adaptive phototaxis model.","keywords":["phototaxis","Chlamydomonas reinhardtii","tangent law","adaptive response","eyespot offset","stochastic switching","conflicting stimuli","decision-making"],"falsifier":"Measure the swimming direction of single cells whose eyespot offset is known while presenting two beams with individual intensities high enough that the photoreceptor signal should saturate; if $\\phi^*$ departs from the tangent law as the sum of intensities increases, the additivity assumption fails. A second check is to image or record the photoreceptor current while both beams illuminate the eyespot simultaneously and compare the response with the sum of single-beam responses.","tokens_in":13607,"feed_emoji":"🦠","tokens_out":6933,"duration_ms":75386,"temperature":0.7,"pith_summary":"The paper asks how an aneural single cell chooses a direction when two light sources compete, and answers with a simple geometric rule. Negatively phototactic Chlamydomonas reinhardtii cells swim along the intensity-weighted average of the two light propagation vectors, the tangent law $\\tan \\phi^* = \\frac{\\eta-1}{\\eta+1} \\tan \\delta$. The paper derives this law as the fixed point of an adaptive model of phototaxis and shows it is equivalent to minimizing the average light intensity on the cell's anterior pole. At nearly antiparallel beam angles, the population splits into three behaviors—swimming away from one source, away from the other, or along the average direction—with some cells stochastically switching between choices. The paper attributes this apparent randomness to measured variation in eyespot position across the population, making the decision rule deterministic at the level of each cell's geometry.","feed_headline":"Algae pick a direction by averaging two light beams","feed_subtitle":"Negatively phototactic Chlamydomonas swims along the intensity-weighted sum of two competing beams, a new single-cell study shows.","key_machinery":"The machinery is the adaptive phototaxis model: rigid-body Euler-angle dynamics for the spinning cell coupled to a fast flagellar photoresponse $P$ and a slower adaptation variable $H$, reduced for a photoreceptor in the equatorial plane to $\\phi_T = -P \\sin T$ with $P$ satisfying a damped-oscillator equation. Under the additivity assumption $S = P^*[\\eta J_+ H(J_+) + J_- H(J_-)]$ with Heaviside shading and projections $J_\\pm = \\sin(\\phi \\mp \\delta) \\sin T$, averaging over the fast spin gives the reorientation dynamics $\\phi_T = -\\lambda[\\eta \\sin(\\phi - \\delta) + \\sin(\\phi + \\delta)]$, whose steady state is the tangent law. The generalization to off-equatorial eyespots replaces the projections with $J_\\pm = \\cos \\gamma \\sin(\\phi \\mp \\delta) \\sin T - \\sin \\gamma \\cos(\\phi \\mp \\delta)$ and adds rotational diffusion, turning the model into a Langevin equation whose simulated trajectories reproduce the three-peaked $Q(\\phi)$.","core_discovery":"The central discovery is that binary phototactic choice in Chlamydomonas obeys a geometric sum rule: for two beams with intensity ratio $\\eta$ and half-angle $\\delta$, the mean swimming direction $\\phi^*$ satisfies $\\tan \\phi^* = \\frac{\\eta-1}{\\eta+1} \\tan \\delta$, equivalently $\\hat{u}^* = (\\eta \\hat{v}_+ + \\hat{v}_-)/|\\eta \\hat{v}_+ + \\hat{v}_-|$. The paper shows this is not an empirical fit but the stable fixed point of the adaptive phototaxis dynamics, which also minimizes the Lyapunov function $V = -\\hat{e}_3 \\cdot (\\eta \\hat{v}_+ + \\hat{v}_-)$ representing the average light on the anterior pole. The same adaptive model quantitatively describes fast reorientation after a sudden switch of light direction with a single fitted parameter. The paper further shows that at large angular separation ($2\\delta = 162^\\circ$) the observed three-peaked distribution of swimming directions is reproduced by a Langevin extension of the model in which each cell's photoreceptor is displaced from the equator by an angle drawn from the measured Gaussian distribution of eyespot offsets $h = R \\sin \\gamma$. In this picture, cells with large negative offsets swim stably away from one source, cells with positive offsets follow the tangent law, and intermediate cells hop stochastically between choices, with an effective free energy that bifurcates from one well to two as $h$ increases.","pith_inferences":["If the linear additivity of photoreceptor signals holds beyond two beams, the same geometric sum rule should predict swimming in natural light fields composed of many discrete sources; this could be tested with three beams arranged in a triangle.","The model implies that apparent stochastic switching is largely deterministic given a cell's eyespot offset; selecting or sorting cells by eyespot position should shift the relative weights of the three peaks in $Q(\\phi)$.","The reliance on eyespot offset suggests a generic mechanism: any source of front-back asymmetry in photoreceptor placement will generate multi-stable taxis under conflicting stimuli, so similar three-peaked distributions may appear in other Volvocine algae.","A direct test of additivity at the photoreceptor level—measuring the transduction current while two beams illuminate the eyespot simultaneously—would confirm or refute the load-bearing signal formula."],"forward_implications":["For more than two lights, the intensity-weighted vector average predicts the swimming direction whenever the angular spread of sources is small enough ($2\\delta \\lesssim 140^\\circ$); as the number of sources grows, negatively phototactic cells should swim directly away from the brightest spot.","The adaptive model, with the same parameter values, quantitatively captures the fast reorientation after a light-direction switch, so the tangent law and the turn dynamics share one mechanism.","The three-peaked direction distribution is a superposition of single-well and double-well effective free energies across the population, not a sharp bifurcation; therefore no true critical angle exists.","Because the tangent law minimizes average anterior-pole illumination, the choice rule can be read as an optimization principle for avoiding light damage to the anterior pole."],"supporting_citations":[{"why":"Supplies the adaptive phototaxis model and the parameter values ($\\alpha = 7$, $\\beta = 0.14$) from which the tangent law is derived as a fixed point.","marker":"[22]"},{"why":"Supplemental material containing the experimental protocol, calibration, trajectory analysis, and eyespot imaging used to measure the quantities compared with theory.","marker":"[25]"},{"why":"Provides the reflection-microscopy method for localizing the eyespot, used to measure the distribution of eyespot offsets.","marker":"[27]"},{"why":"Early demonstration of conflicting-stimulus decision-making in bacteria, the baseline summation-rule result this paper extends to phototaxis.","marker":"[3]"},{"why":"Motivates the bifurcation analysis of decision-making under two choices and supplies the effective-free-energy viewpoint used for the three-peak distribution.","marker":"[24]"},{"why":"Provides the Crocker-Grier particle-tracking algorithm used to link and label cell trajectories.","marker":"[26]"}],"fun_headline_variants":["Algae pick direction by intensity-weighted average","Two competing lights: algae take the weighted middle path","Micro-algae merge two beams into one swim direction","Algae average light vectors to avoid both sources"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument assumes that the two light stimuli add linearly inside the photoreceptor signal $S = P^*[\\eta J_+ H(J_+) + J_- H(J_-)]$ and that the fast-spin average used to derive the reorientation dynamics remains valid near $\\delta \\approx 90^\\circ$; if two-beam photoreception saturates, cross-inhibits, or the shading average breaks down, the tangent-law fixed point and the anterior-pole illumination minimum would not follow.","fun_headline_variants_meta":{"raw":{"variants":["Algae pick direction by intensity-weighted average","Two competing lights: algae take the weighted middle path","Micro-algae merge two beams into one swim direction","Algae average light vectors to avoid both sources"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000414,"raw_usage":{"total_tokens":2172,"prompt_tokens":1010,"completion_tokens":1162,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":626,"completion_tokens_details":{"reasoning_tokens":1102}},"tokens_in":626,"tokens_out":1162,"duration_ms":14946,"temperature":1.0,"reasoning_tokens":1102,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T20:46:42.709308+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure the swimming direction of single cells whose eyespot offset is known while presenting two beams with individual intensities high enough that the photoreceptor signal should saturate; if $\\phi^*$ departs from the tangent law as the sum of intensities increases, the additivity assumption fails. A second check is to image or record the photoreceptor current while both beams illuminate the eyespot simultaneously and compare the response with the sum of single-beam responses.","supporting_citations":[{"cited_title":"Leptos, M","cited_arxiv_id":null,"evidence_quote":"Supplies the adaptive phototaxis model and the parameter values ($\\alpha = 7$, $\\beta = 0.14$) from which the tangent law is derived as a fixed point."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplemental material containing the experimental protocol, calibration, trajectory analysis, and eyespot imaging used to measure the quantities compared with theory."},{"cited_title":"Isogai, R","cited_arxiv_id":null,"evidence_quote":"Provides the reflection-microscopy method for localizing the eyespot, used to measure the distribution of eyespot offsets."},{"cited_title":"Decision","cited_arxiv_id":null,"evidence_quote":"Early demonstration of conflicting-stimulus decision-making in bacteria, the baseline summation-rule result this paper extends to phototaxis."},{"cited_title":"Sridhar, L","cited_arxiv_id":null,"evidence_quote":"Motivates the bifurcation analysis of decision-making under two choices and supplies the effective-free-energy viewpoint used for the three-peak distribution."},{"cited_title":"Crocker and D.G","cited_arxiv_id":null,"evidence_quote":"Provides the Crocker-Grier particle-tracking algorithm used to link and label cell trajectories."}],"review_version":1}