{"id":"4408a050-6819-49da-a112-172639c0a34f","arxiv_id":"2412.01360","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"Photons can be detected after traveling diametrically through an adult human head, at roughly one photon per second, using high-power illumination and time-correlated single-photon counting.","lead":"This study reports detecting light that passes all the way through an adult human head, from one side to the other, despite light being attenuated by a factor of about a billion billion. The result suggests that optical brain imaging might one day reach deep brain structures that current near-infrared devices cannot access.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No control experiment is reported: the ~1 photon/s diametric signal could be scalp/air bypass or stray light, so detection through the entire adult head is not yet established.","rationale":"The paper's central claim has two parts: (1) detection of photons across a diametric source-detector separation, and (2) the implication that deep-brain regions can be probed. The experimental detection is the foundation, and it is supported only by a raw TCSPC histogram with no control measurements. The reader identified exactly this as the weakest assumption, and I agree. The ToF comparison to simulation is helpful but not sufficient: CSF-guided paths around the head are present in the authors' own Fig. 3, and the text concedes that most detected photons likely propagated around the top of the head. Those paths could have similar time delays to genuine through-brain transmission. A source-blocked/no-subject/opaque-barrier control would directly test the origin of the counts. Because the missing control is a standard experimental safeguard and the claim is extraordinary, the appropriate verdict is conditional pending that test. This does not change the reader's CONDITIONAL verdict, so no adjustment is needed.","tokens_in":7063,"tokens_out":5441,"duration_ms":53902,"concrete_test":"Run the identical TCSPC acquisition with the subject replaced by an opaque, non-scattering barrier (e.g., a metal sheet) of comparable thickness at the same source-detector separation, and also with the source beam blocked at the laser output. If either configuration yields a coincidence peak above the PMT dark-count rate, the Fig. 2b signal includes extracranial or stray paths and the through-head claim fails; if both yield only dark counts, the signal is consistent with true transmission.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is the experimental detection of photons transmitted diametrically through an adult head at ~1e18 attenuation. The only supporting evidence is the TCSPC histogram in Fig. 2b from a single subject, with no source-blocked, no-subject, opaque-barrier, or background baseline, and no calibration of the absolute detected count rate against a known attenuation. Without such controls, the measured ~1 count/s cannot be attributed to through-head photons rather than to light bypassing the head via scalp or surface paths, reflections around the curved head, ambient light, or optical/electronic crosstalk. The ToF agreement with simulation is suggestive but not decisive, because surface and CSF-guided paths are longer in physical length and can produce comparable delays; indeed, the paper itself notes that most experimentally detected photons likely propagated 'around the top of the head' rather than through the deep interior. Thus the load-bearing assumption—that the detected counts originate from photons traversing the head—is currently untested.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports Monte Carlo simulations of photon transport in a five-layer adult head model, showing that despite an extreme attenuation of ~10^18, diametric transmission is possible and that different source positions can probe deep brain regions (sulci, midbrain, cerebellum). The authors also present a TCSPC experiment on a single adult male subject (head diameter 15.5 cm) that measures roughly one photon per second with a 1.2 W source, and they compare the measured time-of-flight histogram to simulation, claiming agreement in the first two moments. They further analyze photon migration pathways and Jacobian sensitivity maps to propose source-detector configurations for deep-brain optical access.","tokens_in":7257,"tokens_out":4014,"duration_ms":36985,"significance":"If the experimental claim is correct, the paper overturns a widely assumed impossibility and would open a route to non-invasive optical sensing of deep brain regions. The numerical work is substantial: a large-scale MCX simulation (more than 850 GPU hours) on an open-source five-layer mesh with literature optical coefficients, no parameter fitting to the measured data, and explicit discussion of uncertainties in optical properties and anatomy. The experimental setup is described in enough detail to reproduce, and the authors are transparent about the limitations of their comparisons. However, the central experimental claim is currently supported by a single-subject TCSPC measurement without control experiments, so the significance, while potentially high, is not yet established.","major_comments":[{"comment":"The central claim that photons are detected diametrically through the adult head is not supported by any control experiment. The paper reports no source-blocked measurement, no opaque-barrier test between source and detector, no no-subject baseline, and no subtraction of ambient light or electronic crosstalk. The stated PMT dark count rate of 15 cps is more than an order of magnitude larger than the claimed ~1 photon/s signal, so dark counts alone could constitute a substantial fraction of the detected events. Without such controls, the measured ToF histogram cannot be uniquely attributed to photons that traversed the head.","section":"Experimental evidence (Fig. 2)"},{"comment":"The paper states that 'it is likely that most of the experimentally detected photons propagated around the top of the head' (the sentence immediately following the description of Fig. 3b). This directly conflicts with the abstract's claim of light 'transmitted diametrically through the entire adult human head.' If most detected photons travel around the top of the head via extracerebral, CSF-guided paths, then the experiment demonstrates transmission across the head's surface, not deep-brain transmission. The authors need to clarify the exact claim and provide evidence that the detected signal is sensitive to deep-brain regions rather than scalp and CSF paths.","section":"Photon migration pathways (paragraph after Fig. 3)"},{"comment":"The reported attenuation of ~10^18 is inferred from a detected count rate of roughly 1 photon/s and a 1.2 W source, but no calibration chain is given. The end-to-end detection efficiency, including the tapered fiber bundle transmission, the PMT quantum efficiency (15%), coupling losses, and reflection losses at interfaces, is not measured or estimated. A known-attenuation calibration or a direct measurement of the system's detection efficiency is required to convert the observed count rate into an attenuation value. Without this, the 10^18 figure is an order-of-magnitude estimate with unknown systematic error.","section":"Experimental evidence (absolute attenuation)"},{"comment":"The agreement between simulated and experimental ToF distributions is limited to 'the first two moments,' i.e., peak delay and width. The authors acknowledge that the optical properties and anatomy in the simulation differ from the subject, and they note that surface-guided paths can produce similar delays to through-head paths. Therefore, the reported agreement is not a stringent test of the claim that the detected photons followed the simulated deep-brain pathways. This comparison does not substitute for the missing control experiments.","section":"Experimental evidence (Fig. 2b, simulation comparison)"}],"minor_comments":[{"comment":"The first author's contribution list contains a duplicate 'conceptualisation' entry; one should be removed.","section":"Author Contributions"},{"comment":"'Savitsky-Golay' is a misspelling; the correct name is 'Savitzky-Golay'.","section":"Experimental evidence"},{"comment":"'single-point precision' should likely read 'single precision' when describing the MCX simulation.","section":"Numerical modelling"},{"comment":"The exact source and detector positions are not specified with respect to anatomical landmarks (e.g., ear canal, vertex), and the subject's hair and scalp condition are not described; these details are needed for reproducibility.","section":"Experimental evidence"},{"comment":"The 5 cm diameter source launch area in the simulation is introduced to 'account for the uncertainty in position throughout the experiments,' but this is a modeling assumption rather than a measured parameter; it should be explicitly identified as such.","section":"Photon migration pathways"},{"comment":"The statement of 'good agreement' would be more quantitative if the authors reported the values of the first two moments (mean and variance) with their uncertainties, rather than only showing the curves.","section":"Experimental evidence (Fig. 2b)"}],"recommendation":"major_revision","confidential_remarks":"The paper's provocative claim is likely to attract attention, but the current experimental evidence is insufficient to support it. I recommend major revision with the explicit requirement of control experiments (source blocked, opaque barrier, no-subject baseline, and dark-count characterization) and a calibration chain for the absolute attenuation. If these cannot be provided, the paper should be reframed as a simulation study with a preliminary experimental illustration, with the abstract moderated accordingly."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Dear [Colleague],\n\nThe headline claim — time-correlated detection of light transmitted across the full 15.5 cm adult head — is worth taking seriously, but the paper doesn't yet establish it. What is new: this is the first time-resolved measurement attempting that geometry in an adult, and the simulations that go with it are properly done. The Monte Carlo work uses an open-source head mesh and literature optical coefficients, with no parameters fitted to the measured ToF data. The two-day reproducibility of the histogram's first two moments against the simulated distribution is genuinely encouraging, and the sensitivity Jacobians in Fig. 4 are a nice extension of the usual fNIRS banana profiles to very large source-detector separations.\n\nThe soft spot is exactly the one the stress-test flags: there is no control measurement. No source-blocked run, no no-subject baseline, no opaque barrier, no crosstalk check. At roughly one count per second with a 1.2 W source, stray light or a scalp/air bypass could plausibly produce that rate. The paper even concedes that most experimentally detected photons likely propagated around the top of the head rather than through the deep interior — which sits awkwardly with the abstract's claim of diametric transmission through the entire adult head. That internal tension needs to be resolved. The absolute attenuation of ~10^18 is also asserted without a detailed calibration chain; the ToF shape is robust to that uncertainty, but the absolute count claim is not.\n\nThe single subject and two-day repetition are acceptable for a proof-of-principle, though replication would help. The authors are transparent about the uncertainties in the optical coefficients and the anatomy mismatch between the mesh and the participant, which I credit. The citation pattern is fine; the self-references are to their own prior diffuse optics work and are relevant.\n\nBottom line: the simulation study is valuable on its own, and the experimental claim is plausible but unproven until the obvious controls are run. I would send this to peer review, because the potential significance is high and the methodology is clear enough that reviewers can demand the missing baselines. But I would not cite the detection claim in my own work yet.\n\nRecommendation: engage with it, but treat the central detection as conditional.","headline":"A plausible, potentially important result that is under-supported: the single-subject TCSPC detection lacks the controls needed to rule out light bypassing the head, so the 'diametric transmission' claim is not yet established.","tokens_in":7768,"tokens_out":3636,"would_cite":false,"duration_ms":30208,"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":"Near-infrared photons can be detected after crossing the full 15.5 cm width of an adult human head, at roughly one photon per second, despite an attenuation of about 10^18—challenging the long-held view that such detection is impossible.","keywords":["time-correlated single photon counting","diffuse optical tomography","near-infrared spectroscopy","cerebrospinal fluid light guiding","photon transport in human head","deep brain optical sensing","Monte Carlo photon simulation"],"falsifier":"Run the identical time-tagged acquisition with an opaque sheet shadowing the entire scalp and air path between the source and detector, or with the subject replaced by an absorbing phantom that leaves the geometry unchanged; if a coincidence peak at the simulated transit delay persists, through-head transmission is not the only source of the detected counts.","tokens_in":6903,"feed_emoji":"🧠","tokens_out":6418,"duration_ms":53695,"temperature":0.7,"pith_summary":"The paper tries to establish that near-infrared light can be transmitted diametrically through the entire adult human head—across the widest 15.5 cm of the skull—even though attenuation is on the order of $10^{18}$. If true, this overturns the common assumption that optical sensing cannot reach deep brain structures in adults and suggests that a well-chosen source and detector, together with time-correlated single-photon counting, can recover roughly one photon per second from a 1.2 W laser. The authors support the claim with large-scale numerical simulations that predict where photons migrate inside the head, and with time-of-flight measurements on an adult subject whose detected distribution matches simulation in its peak delay and width. They further argue that cerebrospinal fluid acts as a low-loss guide, steering photons along preferred routes over and under the brain, so that source placement can select which deep regions are interrogated.","feed_headline":"Photons detected after crossing an entire adult head","feed_subtitle":"Near-infrared light survives 15.5 cm of tissue and ~10^18 attenuation, opening a route to deep-brain sensing.","key_machinery":"The mechanism that carries the claim is light guiding by low-scattering cerebrospinal fluid layers, which sit between the higher-scattering skull and grey matter and act as channels of least extinction through the head. On the experimental side, the load-bearing instrument is a time-correlated single-photon counting chain: a high-power pulsed laser expanded over a large skin-safe area, a demagnifying fiber taper that increases collection etendue, and a photomultiplier with low dark count and sub-nanosecond timing jitter. This chain lets the roughly one-photon-per-second signal be time-tagged against the laser repetition rate, separating genuine transmitted photons from background and enabling comparison with simulated time-of-flight distributions.","core_discovery":"The paper's central claim is that diametric photon transmission through an adult head is experimentally measurable: an 800 nm pulsed laser expanded to a 25 mm spot, a large-area fiber taper feeding a low-dark-count photomultiplier, and time-correlated single-photon counting yield a detected flux of about one photon per second, corresponding to an attenuation of about $10^{18}$. The measured time-of-flight distribution overlaps the simulated distribution for the first two moments, which the authors take as evidence that the detected photons followed head-crossing migration pathways rather than an external route. Monte Carlo analysis of individual photon trajectories shows that low-scattering cerebrospinal fluid channels guide light around and beneath the cerebrum, creating reproducible sensitivity volumes that, at extreme source–detector separation, extend into the midbrain, sulci, and deep cerebellum. The authors conclude that with careful source–detector geometry and time-of-flight gating, these guided pathways could be used to probe deep brain regions currently outside the reach of functional near-infrared spectroscopy.","pith_inferences":["If the guidance mechanism is as robust as the paper suggests, multi-source and multi-detector time-of-flight data could be inverted into tomographic reconstructions of deep brain activity without relying on fMRI-based correlation.","Because the detected flux is roughly one photon per second, detectors with higher quantum efficiency or arrays of detectors could increase the count rate enough to make deep-brain functional monitoring practical in a clinical setting.","A decisive next test, not reported here, would be to demonstrate that the through-head coincidence peak disappears when an opaque barrier blocks the scalp path; this would rule out the main alternative explanation for the counts.","The strong dependence of guided paths on cerebrospinal fluid geometry implies that posture, head size, and even head position relative to gravity could be used to tune which deep structures are optically sampled."],"forward_implications":["Time-domain diffuse optical tomography could be extended to source–detector separations far beyond the usual few centimetres, reaching the midbrain, sulci, and deep cerebellum.","Source placement becomes a control knob: lowering the source by 40 mm shifts sensitivity almost exclusively to regions beneath the cerebrum, which could be used to target specific deep structures.","Cerebrospinal-fluid-guided light, normally treated as a nuisance in fNIRS, can instead be exploited to localise absorption changes in deep brain regions.","The hardware recipe—large-area illumination and collection plus single-photon time tagging—provides a template for future deep-brain optical sensing devices.","The first two moments of the simulated and measured time-of-flight distributions agree, suggesting that transit time itself can be used as a filter to select photons that interacted with deep tissue."],"supporting_citations":[{"why":"Supplies the five-layer optical coefficients (absorption and scattering at 810 nm) used in the Monte Carlo head model.","marker":"[18]"},{"why":"Provides the open-source, averaged-MRI five-layer head volume mesh used for all photon-transport simulations.","marker":"[21]"},{"why":"Gives the theoretical result that photons at scattering length scales comparable to the head diameter can carry imaging information, motivating the experimental search.","marker":"[6]"},{"why":"Describes the fundamental mechanism of guided light in diffusive materials, which underpins the cerebrospinal-fluid pathways identified here.","marker":"[23]"},{"why":"Supplies the perturbation Monte Carlo method used to compute the Jacobian sensitivity maps for deep brain regions.","marker":"[27]"},{"why":"Represents the prior claim that diametric detection through the adult head is impossible, the baseline the paper's experiment challenges.","marker":"[17]"},{"why":"Demonstrates imaging with highly scattered transmitted photons in thick diffusive media, the technique the authors extend to the head.","marker":"[4]"}],"fun_headline_variants":["Photons cross a full adult head, enabling deep-brain access","1 photon per second: light survives entire adult head","Through-head photon paths reach midbrain and sulci","Diametric head transmission: new route to deep brain"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that the counted photons really traversed the head, rather than reaching the detector by travelling around the scalp and hair or through an air gap, and that the coincidence peak is not an electronic or optical crosstalk artefact; the paper reports no control run with the source blocked, an opaque barrier inserted, or the subject removed.","fun_headline_variants_meta":{"raw":{"variants":["Photons cross a full adult head, enabling deep-brain access","1 photon per second: light survives entire adult head","Through-head photon paths reach midbrain and sulci","Diametric head transmission: new route to deep brain"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000237,"raw_usage":{"total_tokens":1464,"prompt_tokens":859,"completion_tokens":605,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":475,"completion_tokens_details":{"reasoning_tokens":539}},"tokens_in":475,"tokens_out":605,"duration_ms":6391,"temperature":1.0,"reasoning_tokens":539,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T04:25:32.312144+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the identical time-tagged acquisition with an opaque sheet shadowing the entire scalp and air path between the source and detector, or with the subject replaced by an absorbing phantom that leaves the geometry unchanged; if a coincidence peak at the simulated transit delay persists, through-head transmission is not the only source of the detected counts.","supporting_citations":[{"cited_title":"Selective photobiomodulation for emotion regulation: model- based dosimetry study,","cited_arxiv_id":null,"evidence_quote":"Supplies the five-layer optical coefficients (absorption and scattering at 810 nm) used in the Monte Carlo head model."},{"cited_title":"Improving model- based functional near-infrared spectroscopy analysis us- ing mesh-based anatomical and light-transport models,","cited_arxiv_id":null,"evidence_quote":"Provides the open-source, averaged-MRI five-layer head volume mesh used for all photon-transport simulations."},{"cited_title":"Information transport and limits of optical imaging in the highly diffusive regime,","cited_arxiv_id":null,"evidence_quote":"Gives the theoretical result that photons at scattering length scales comparable to the head diameter can carry imaging information, motivating the experimental search."},{"cited_title":"Energy transport in diffusive waveguides,","cited_arxiv_id":null,"evidence_quote":"Describes the fundamental mechanism of guided light in diffusive materials, which underpins the cerebrospinal-fluid pathways identified here."},{"cited_title":"Direct approach to com- pute Jacobians for diffuse optical tomography using per- turbation Monte Carlo-based photon “replay","cited_arxiv_id":null,"evidence_quote":"Supplies the perturbation Monte Carlo method used to compute the Jacobian sensitivity maps for deep brain regions."},{"cited_title":"Becker, Advanced time-correlated single photon count- ing techniques","cited_arxiv_id":null,"evidence_quote":"Represents the prior claim that diametric detection through the adult head is impossible, the baseline the paper's experiment challenges."},{"cited_title":"Computational time- of-flight diffuse optical tomography,","cited_arxiv_id":null,"evidence_quote":"Demonstrates imaging with highly scattered transmitted photons in thick diffusive media, the technique the authors extend to the head."}],"review_version":1}