{"id":"0b1a2938-dcd8-44fb-b288-763da9882b22","arxiv_id":"2608.11761","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Heavy cosmic ray nuclei (iron, carbon, oxygen) produce concentrated Cherenkov light inside the eye that matches the reported astronaut light flash phenomenon, while hydrogen, helium, and ground-level muons do not.","lead":"Astronauts have seen mysterious light flashes in space since the Apollo missions. This paper uses particle simulations to argue the flashes come from Cherenkov light emitted when heavy cosmic ray nuclei, mainly iron, carbon, and oxygen, pass through the fluid of the human eye.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Predicted LF rate in Table 5 (5.5/min for Ap≥5, 1.9/min for Ap≥14) is 10–50× above observed 0.05–0.34/min; the paper's own numbers contradict the claimed consistency.","rationale":"The paper has real strengths: the Geant4/Cherenkov-model cross-check in Fig. 8 is convincing for photon yields, the code is public, and the qualitative argument that heavy nuclei produce sufficient concentrated Cherenkov light to activate retinal rod patches is physically plausible. The fatal issue is quantitative. Eq. (1) combined with Table 5 gives 5.5 flashes/min (Ap≥5) or 1.9 flashes/min (Ap≥14), whereas the rates the authors themselves list in Table 1 are almost all 0.048–0.343 flashes/min. The text even concedes an order-of-magnitude excess. A factor-of-2 error in G and the axial-vs-isotropic Pi inconsistency both inflate the prediction, so correcting them cannot make the current tabulated result consistent; at best they move the model in the right direction. The abstract's claim of consistency is therefore not supported by the manuscript's own output. I agree with the reader's rejection, though I would put the primary weight on the direct numeric mismatch rather than on the geometry assumption alone. A focused isotropic-incidence rerun would settle whether a corrected version can recover the claimed agreement.","tokens_in":25599,"tokens_out":9946,"duration_ms":111337,"concrete_test":"Re-run the SG-Vac eye model with isotropic primary directions for C, N, O, and Fe at 5–50 GeV; tally A500 activations to form an angle-averaged Pi(Tkin,Ap) for Ap=5 and 14, then recompute Eq. (1) with the correct projected-area geometry constant G=(2/3)·4π²R² and the 1.667-cm mean-chord yield. If the summed predicted rate remains above the observed 0.05–0.34 LF/min band (and especially above 1 LF/min), the claimed consistency is refuted; if it falls inside, the central claim would need the corrected model to replace the current Table 5.","verdict_should_be":"REJECT","load_bearing_attack":"The central claim—Cherenkov emission reproduces observed astronaut light-flash rates—is contradicted by the paper's own Table 5. Summing the predicted N. of flashes per minute over H–O and Fe gives ≈5.5 LF/min for Ap≥5 and ≈1.9 LF/min for Ap≥14. The observed rates in Table 1 (Apollo 14/15/16/17, Sileye, ALTEA) are 0.048–0.343 LF/min, with the single Skylab-2 value 2.618 LF/min. Even the highest typical point is ≈8× below the Ap≥14 prediction and ≈16× below Ap≥5; most points are 20–100× lower. The abstract's 'consistent with observed frequency' is not supported without an error budget or a selection of the 2.618 LF/min outlier. The text itself concedes 'one order of magnitude more particles creating light flashes ... than Apollo numbers'. The numerical overestimate is compounded by the modeling assumptions in Eq. (1): the geometry constant G uses the full eye surface (19.6 cm²) times 2π sr, whereas the isotropic-flux crossing rate is 4π²R²=61.7 cm²·sr before the 2/3 retinal factor—an overcount of 2—and the activation probability Pi is taken from axial 2.5-cm Geant4 tracks while the yield weighting uses the 1.667-cm mean chord. Both corrections push predicted rates further down, so they cannot rescue the claim as written; they only widen the discrepancy between the claimed consistency and what the model actually computes. The qualitative mechanism (heavy nuclei concentrated enough to activate rod patches) remains plausible and testable, but the headline quantitative consistency does not hold.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proposes that Cherenkov radiation produced by cosmic-ray nuclei in the ocular fluid is the mechanism behind the light flashes reported by astronauts. It combines Geant4 simulations of particle passage through simplified water-eye models with a Frank–Tamm Cherenkov yield model, physiological inputs (rod quantum efficiency 29%, Hecht-style thresholds of 5 or 14 absorbed photons in ~500-rod patches), and AMS-02/PAMELA/BESS spectra to predict light-flash rates in interplanetary space, low Earth orbit, and at ground level. The central quantitative claim is that predicted rates are consistent with Apollo/Skylab/ISS observations and that the dominant contributors are heavy nuclei, primarily iron, with carbon and oxygen also contributing. The paper also argues that muon-induced Cherenkov light is insufficient to produce flashes at Earth's surface and that South Atlantic Anomaly flashes require a different mechanism.","tokens_in":25988,"tokens_out":8724,"duration_ms":93343,"significance":"If the quantitative claim held, the paper would resolve a long-standing puzzle in space radiation biology and would identify iron, carbon, and oxygen as the flash-producing primaries, with testable predictions for geomagnetic-cutoff dependence. The work has notable strengths: the simulation code and Cherenkov model are publicly archived, the Geant4 and Frank–Tamm yields are cross-checked against each other, no parameter is fitted to the observed flash rates, and the model makes a falsifiable prediction about the decrease of flash rate with geomagnetic cutoff rigidity. However, the headline claim of consistency with observed flash rates is contradicted by the paper's own Table 5, and several modeling choices further inflate the predicted rates. The qualitative idea is plausible and worth pursuing, but the central quantitative result is not supported as written.","major_comments":[{"comment":"The predicted rates do not match the observed rates quoted in the paper. Table 5 gives 5.51 LF/min for Ap≥5 and 1.86 LF/min for Ap≥14 for elements hydrogen through oxygen, while Table 1 lists observed rates of 0.048–0.343 LF/min for Apollo, Skylab-1, Sileye, and ALTEA (the single Skylab-2 value 2.618 LF/min is a clear outlier). Even against the highest typical point, 0.343 LF/min, the Ap≥14 prediction is about 5.4 times too high and the Ap≥5 prediction is about 16 times too high; against the ALTEA value of 0.048 LF/min the overprediction is roughly 40–115 times. The abstract's statement that results are \"consistent with the observed frequency\" is therefore not supported by the paper's own numbers, and the text in §3.2.1 explicitly concedes \"one order of magnitude more particles creating light flashes ... than Apollo numbers.\" Since the model has no free parameters fitted to the flash rates, this discrepancy is a direct failure of the central quantitative claim; an error budget or a revised, honest statement of the predicted-to-observed ratio is required.","section":"Table 5 vs. Table 1; §3.2.1"},{"comment":"The geometry constant G in Eq. (1) overcounts the cosmic-ray crossing rate. G is defined as eye surface 19.6 cm² × 2/3 × solid angle 2π = 82.1 cm²·sr, but for isotropic intensity I [cm⁻² s⁻¹ sr⁻¹] the correct crossing-rate coefficient for a sphere of radius R=1.25 cm is 4π²R² = 61.7 cm²·sr if particles arrive from all 4π sr, or 2π²R² = 30.9 cm²·sr if they arrive from one exposed hemisphere, as is the case for an astronaut's eye in space. The manuscript's G is therefore larger than the full-sky coefficient by a factor of 1.33 and larger than the one-hemisphere coefficient by a factor of 2.66. Moreover, the 2/3 retinal factor belongs in the probability of perception, not in the particle-crossing rate. Correcting G would lower all predicted LF rates, which only widens the discrepancy with Table 1; it cannot rescue the consistency claim.","section":"§3.2.1, Eq. (1)"},{"comment":"The activation probability P_i(Tkin,Ap) is derived from Geant4 tracks that cross the eye along the central axis with a full 2.5-cm path in water, but in Eq. (1) it is applied to isotropically incident particles whose mean chord length is only 4r/3 = 1.667 cm. The paper itself uses the 1.667-cm mean chord for the yield model in §3.2 and for the N. of photons per particle in Table 5, yet the P_i curves used in the flash-rate integral come from the axial 2.5-cm simulations described in §3.2.1 and Figures 12–13. Because Cherenkov photon number scales with track length, the activation probability for a typical off-axis or grazing crossing is overestimated, and no convolution over impact parameter or chord-length distribution is provided. This is a load-bearing inconsistency in the rate calculation and should be corrected before the predicted rates can be accepted.","section":"§3.2 and §3.2.1"},{"comment":"The stated conclusion that iron is the dominant flash-producing primary is inconsistent with the paper's own Table 5. For Ap≥5, the predicted rates are carbon 1.84 LF/min, oxygen 1.81 LF/min, helium 0.67 LF/min, and iron 0.13 LF/min; for Ap≥14 the rates are oxygen 1.19, carbon 0.39, and iron 0.13. Thus carbon and oxygen dominate the predicted rate by more than an order of magnitude over iron, and even helium out-predicts iron at Ap≥5. The abstract and conclusion should be revised to state that carbon and oxygen are the dominant contributors, with iron contributing only a small fraction of the predicted rate; alternatively, the model that produces the iron-dominated result should be specified and reconciled with Table 5.","section":"Abstract, §3.2.3, Table 5"},{"comment":"The manuscript contains an explicit, unresolved admission that undermines the central claim: \"There are not enough particles creating higher numbers of photons by Cherenkovov radiations to reproduce Apollo results. If light flash experience need few hundred thousand or million or more photons, than light flashes are not Cherenkovov light effect.\" This passage should be treated as part of the paper's evidence, and it directly contradicts the abstract's claim of consistency. Either the model must be shown to reproduce the observed rates with the stated thresholds, or the conclusion must be reframed as an upper limit or as a qualitative mechanism rather than a quantitative explanation.","section":"§3.2.1, paragraph beginning \"This could also mean\""}],"minor_comments":[{"comment":"The column header \"N. of flashes per minute Ap≥5-Ap≥14\" is ambiguous; separate columns for Ap≥5 and Ap≥14 should be used, and the iron row should explicitly state both values.","section":"Table 5"},{"comment":"There are numerous typos and inconsistent spellings, e.g., \"Cherenkovov,\" \"simmilar,\" \"parrallel,\" \"vacuume,\" and \"CUBE GMA\"; the manuscript needs careful proofreading.","section":"Throughout"},{"comment":"The placeholder phrase \"setup x from table y\" should be replaced with the actual simulation setup designation, and the callouts to Figures 12 and 13 should list the exact configurations used.","section":"§3.2.1"},{"comment":"The in-text citation \"Aguilar et al. (2021)\" for iron is listed with a January 2021 date while \"Aguilar et al. (2021a)\" is used for the Physics Reports spectra; the reference style should be made consistent so that the reader can identify which AMS-02 data were used for each element.","section":"References"},{"comment":"The units of the solid angle should be written explicitly as sr, and the decimal comma in \"19,6cm2\" should be replaced with a decimal point for consistency with the rest of the paper.","section":"§3.2.1, Eq. (1)"}],"recommendation":"reject","confidential_remarks":"The paper has a plausible qualitative mechanism and a commendable level of transparency (public code, no fitted parameters, cross-checked simulations), but the central quantitative claim is contradicted by its own Table 5 and by an explicit in-text concession. The geometry and track-length issues further inflate the predicted rates, so the discrepancy cannot be fixed by the straightforward corrections I identify. I would view a substantially revised manuscript favorably if it reframed the result as an upper limit or a qualitative heavy-nucleus mechanism, reported the predicted-to-observed ratio honestly, and corrected the geometry and chord-length treatment; as written, the abstract and conclusions do not match the paper's own numbers."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a serious attempt—public Geant4 code, element-resolved predictions, a real falsifiable experiment proposed—but the central quantitative claim is not supported by their own numbers. Table 5 predicts 5.5 LF/min for Ap≥5 and 1.9 for Ap≥14, against roughly 0.05–0.35 LF/min observed on Apollo, Sileye, and ALTEA. That's a 10–50× overprediction. The abstract says 'consistent with the observed frequency'—it isn't.\n\nWhat's genuinely good: the Cherenkov idea itself is old (Fuglesang's estimate), but the full chain here—Geant4 eye model, Frank–Tamm yields for H through O plus Fe, AMS-02 spectra, rod quantum efficiency, A500 activation maps—is new and shipped as code with Zenodo archives. The element-resolved flash-rate predictions are a real, falsifiable output. The muon dilution argument for ground level is coherent: even though muons produce comparable Cherenkov yields, their retinal photon density doesn't reach the 5-photon activation patch. The SAA result (negligible Cherenkov rate, so another mechanism is needed) is honest, and the proposed rigidity-cutoff experiment (Indian Ocean vs North America) plus the zenith/nadir asymmetry give the field a concrete way to settle the mechanism.\n\nSoft spots, in order of importance. First, the overprediction above is not a small error—it's the paper's own table. The text notes 'one order of magnitude more particles creating light flashes... than Apollo numbers', but then the perception-threshold model still ends up at 5.5/min; no error budget or selection of the 2.6/min Skylab outlier rescues that. Second, the geometry factor G uses 2π sr times eye surface, which overcounts the isotropic-flux projection by a factor of 2 (should be π times the surface). Third, the activation probability Pi is taken from central-axis Geant4 tracks with a full 2.5 cm path, while the yield model uses the 1.667 cm mean chord for isotropic incidence; shorter off-axis tracks will activate rods less often. Both corrections push predicted rates down, i.e. they widen the discrepancy, not close it. Fourth, the iron 'cloud' interpretation is plausible but speculative, and their own caveat—'if light flash experience needs a few hundred thousand or million photons, then light flashes are not Cherenkov'—is not fully resolved.\n\nWho this is for: space radiation biologists and anyone studying astronaut perception; also a useful example of the gap between a plausible mechanism and a validated quantitative model. It deserves a serious referee—the code and elemental predictions warrant engagement—but in current form it should not be accepted. I'd recommend major revision, requiring the authors to fix the geometry and chord-length issues and revisit the threshold, or explicitly report the overprediction as a limitation rather than claiming consistency.","headline":"Plausible mechanism, reproducible code, but the paper's own predicted flash rates are an order of magnitude above the observed rates, so the central consistency claim fails as written.","tokens_in":26539,"tokens_out":5945,"would_cite":false,"duration_ms":61574,"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":"Cherenkov light generated inside the eye by heavy cosmic-ray nuclei explains the light flashes astronauts see in space, the paper argues.","keywords":["Cherenkov radiation","astronaut light flashes","cosmic ray nuclei","retinal activation threshold","iron cosmic rays","low Earth orbit","South Atlantic Anomaly","Monte Carlo eye model"],"falsifier":"Measure astronaut light-flash rates over a region of low vertical geomagnetic cutoff (below 1 GV, such as high-latitude North America) and over a high-cutoff region (above 14 GV, such as the Indian Ocean at low latitude); the model predicts roughly a tenfold reduction in flash rate, and the paper itself states that if no such difference is observed the Cherenkov model is challenged.","tokens_in":25410,"feed_emoji":"👁️","tokens_out":15425,"duration_ms":155575,"temperature":0.7,"pith_summary":"Astronauts have seen unexplained flashes of light in space since the very first crewed Moon landing, and the cause has never been settled. This paper argues that the flashes are Cherenkov light: the faint blue glow produced when a charged particle travels through the eye's fluid faster than light travels through that fluid. Using a Monte Carlo simulation of the eye and published limits on human visual perception, the paper shows that heavy cosmic-ray nuclei—chiefly iron, with carbon and oxygen contributing—deposit enough Cherenkov photons into small retinal patches to cross the eye's activation threshold, while the much more numerous protons and helium nuclei do not. The same mechanism explains why no such flashes are reported at ground level: muons emit plenty of photons but spread them too thinly over the retina. If the paper is right, astronaut flash reports become a working probe of the heavy-nucleus component of the cosmic-ray spectrum.","feed_headline":"Cherenkov glow in the eye explains astronaut light flashes","feed_subtitle":"A simulation model matches reported flash rates and explains why the same flashes do not occur on the ground.","key_machinery":"The load-bearing machinery is the Cherenkov light production model: a Frank–Tamm calculation of visible photon yield (380–700 nm) for the first thirty elements, evaluated over the average straight-line path of 1.667 cm in a 2.5 cm water sphere and weighted by energy-differential cosmic-ray intensities. For a single nucleus the yield climbs from about 300 photons for a proton to about 19,500 for oxygen and about 56,000 for iron at high energy, but raw photon number is not the deciding factor. The model then projects photon trajectories onto the retinal surface, treats each rod-photon encounter as absorption with 29% probability, and requires that some patch of roughly 500 rods (0.044 mm × 0.044 mm) absorb at least $A_p = 5$ or 14 photons before a flash is registered. This defines the activation probability $P_i(T_{\\mathrm{kin}}, A_p)$, and the predicted flash rate for element $i$ is $N_{\\mathrm{LF},i} = G \\int_0^{500\\,\\mathrm{GeV}} I_{\\mathrm{CR},i}(T) \\, P_i(T, A_p) \\, dT$, with $G$ the eye surface area times the retinal fraction times the $2\\pi$ inward solid angle. The same machinery applied to the measured ground-level muon spectrum and its $\\cos^2\\theta$ angular distribution produces the predicted null result at Earth's surface.","core_discovery":"The paper's central claim is that one mechanism—Cherenkov emission by cosmic-ray nuclei moving through the water-like interior of the eye—accounts for the light flashes reported by astronauts in interplanetary space and in most of low Earth orbit. Iron nuclei dominate the perceived events because a single iron nucleus produces tens of thousands of visible photons along its roughly 1.7-centimeter average path, and among the lighter nuclei only carbon and oxygen concentrate enough photons on a roughly 500-rod retinal patch to meet the activation condition of at least five absorbed photons. Hydrogen and helium, although vastly more numerous, are effectively invisible because their few hundred photons arrive spread over too wide a retinal area. The paper explicitly exempts the South Atlantic Anomaly, where trapped protons would give less than one flash per week and another mechanism must operate, and it shows that ground-level muons cannot activate the retina despite producing bright Cherenkov tracks.","pith_inferences":["If the mechanism is Cherenkov emission, the astronaut retina is effectively a composition-sensitive cosmic-ray detector; pairing flash reports with an on-board particle telescope could yield an independent iron-to-carbon abundance measurement.","The activation logic predicts a sharp elemental threshold: single-particle flashes should occur only for nuclei at least as heavy as carbon, regardless of how high the proton flux is; an instrument that records both the charge of each crossing nucleus and the astronaut's response could map this threshold directly.","The model's spatial-concentration claim is testable in a laboratory: a dark-adapted eye placed in a beam of heavy ions at tens of GeV per nucleon should perceive spot and cloud flashes matching the predicted retinal activation patterns."],"forward_implications":["Outside the South Atlantic Anomaly, the light-flash rate becomes a predictable function of geomagnetic cutoff: the model gives about a tenfold reduction between low-cutoff regions (below 1 GV, such as high-latitude North America) and high-cutoff regions (above 14 GV, such as the equatorial Indian Ocean).","Iron nuclei above roughly 40 GeV should be perceived as diffuse 'cloud' flashes, because a single nucleus activates tens to hundreds of separate 500-rod patches at once; this explains why cloud flashes are a minority of astronaut reports.","Looking toward zenith versus nadir in high-latitude low Earth orbit should change the flash rate, since Earth blocks roughly half the inward cosmic-ray hemisphere and the retinal area exposed to inward tracks differs.","At aircraft altitudes, flashes should be heavy-nucleus events rather than muon events, so the mechanism observed in space also accounts for the rare high-altitude reports.","Muon-induced Cherenkov flashes at ground level should remain unobservable, with an upper limit near one flash per 24 hours of continuous viewing, because individual muons cannot concentrate enough photons on one retinal patch."],"supporting_citations":[{"why":"Supplies the visual threshold of 5–14 absorbed photons in an area of about 500 rods, which defines the activation condition Ap.","marker":"Hecht et al. (1942)"},{"why":"Provides the measured cosmic-ray spectra for hydrogen through oxygen used to weight the Cherenkov yield.","marker":"Aguilar et al. (2021a)"},{"why":"Provides the iron cosmic-ray spectrum used to estimate the iron flash rate.","marker":"Aguilar et al. (2021)"},{"why":"Supplies the observed light-flash rates and the dual-mechanism classification, including the Sileye-2 cutoff-region comparison.","marker":"Casolino et al. (2003b)"},{"why":"Reports the early space-mission light-flash rates and morphologies that the model must reproduce.","marker":"Johnston et al. (1975)"},{"why":"Provides the ground-level positive-muon spectrum used to conclude that muon-induced flashes are unobservable.","marker":"Motoki et al. (2003)"},{"why":"Supplies the 29% rod quantum efficiency used in the retinal absorption simulation.","marker":"Phan et al. (2014)"},{"why":"Supplies the human eyeball dimensions at the basis of the 2.5 cm spherical eye model.","marker":"Bekerman et al. (2014)"},{"why":"Documents the simulation toolkit used for particle transport, Cherenkov emission, and photon trajectory tracking in the eye model.","marker":"Geant4 Collaboration (2020a)"}],"fun_headline_variants":["Cherenkov in the eye: cause of astronaut light flashes","Iron-rich cosmic rays trigger Cherenkov flashes in eyes","Astronaut flash mystery solved by Cherenkov in eye","Cherenkov glow from cosmic rays explains astronaut flashes"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a cosmic-ray nucleus crossing the eye off-center or at an angle deposits Cherenkov photons onto the retina just as efficiently as the straight-through, central-axis tracks used to calibrate the activation probability; a second factor of two sits in the geometry constant that counts the incoming flux, and either overestimate would lower the predicted flash rate toward the observed values.","fun_headline_variants_meta":{"raw":{"variants":["Cherenkov in the eye: cause of astronaut light flashes","Iron-rich cosmic rays trigger Cherenkov flashes in eyes","Astronaut flash mystery solved by Cherenkov in eye","Cherenkov glow from cosmic rays explains astronaut flashes"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000781,"raw_usage":{"total_tokens":3431,"prompt_tokens":904,"completion_tokens":2527,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":520,"completion_tokens_details":{"reasoning_tokens":2459}},"tokens_in":520,"tokens_out":2527,"duration_ms":17624,"temperature":1.0,"reasoning_tokens":2459,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T00:30:19.865291+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Measure astronaut light-flash rates over a region of low vertical geomagnetic cutoff (below 1 GV, such as high-latitude North America) and over a high-cutoff region (above 14 GV, such as the Indian Ocean at low latitude); the model predicts roughly a tenfold reduction in flash rate, and the paper itself states that if no such difference is observed the Cherenkov model is challenged.","supporting_citations":[{"cited_title":"Astroparticle Physics , publisher=","cited_arxiv_id":null,"evidence_quote":"Provides the ground-level positive-muon spectrum used to conclude that muon-induced flashes are unobservable."}],"review_version":1}