{"id":"61b1a400-2e55-428e-83bc-ac11a5200a64","arxiv_id":"1908.01243","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"A refocus formed by wavefront shaping keeps working over a much wider wavelength span when the detection plane is shifted axially or when a broadband transmission matrix is used, and confocal geometry minimizes the required shift.","lead":"Wavefront shaping through scattering media usually works only over a narrow wavelength range. This paper shows the range can be widened by moving the detection plane axially and by using broadband laser pulses, with the best results in a confocal geometry.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Broadband bandwidth claim lacks a spectral isolation check: a TM measured with an 8 nm pulse is asserted to focus over 260 nm, which needs direct verification at a single detuned wavelength.","rationale":"The reader correctly identifies the thin/forward-scattering assumption as a limitation, and the thick TiO2 control supports that this effect is not universal. However, I see a more specific and testable weakness in the broadband section. The paper's strongest quantitative claim, a 260 nm bandwidth, depends on a mechanism that is not directly evidenced by the data. The 8 nm source bandwidth versus 260 nm claimed bandwidth is a large ratio, and the cited coherent time-gating mechanism does not by itself establish spectral validity over that range. A spectral-filtering experiment would settle this cleanly. The first part of the paper, based on Eq. (1) and the axial shift recovery, is plausible and qualitatively supported by the data, though it also suffers from undefined bandwidth thresholds. Since the reader's conditional verdict already captures the need for further validation, I do not propose changing the verdict.","tokens_in":10549,"tokens_out":7889,"duration_ms":91951,"concrete_test":"After measuring the broadband TM at the confocal geometry, detune the OPO center by a fixed amount, e.g. +100 nm, and spectrally filter the light at the refocus with a bandpass filter centered on the detuned wavelength. Measure the enhancement ratio of the refocused spot at that wavelength. Also repeat with a narrowband (picosecond) probe at the same detuned wavelength while keeping the broadband TM. If the enhancement at the isolated detuned wavelength is absent or much smaller than the reported 3D bandwidth implies, the 260 nm bandwidth claim is an artifact of spectral integration rather than true spectral memory.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The least secure support for the central claim is the broadband result in Sec. 5. The source is a 150 fs pulse with a spectral width of about 8 nm, yet the reported 3D bandwidth is 260 nm. If the OPO center wavelength is tuned over this range, the transmission matrix used for refocusing was only measured with an 8 nm bandwidth at one center wavelength. The paper attributes the enlargement to the coherent time-gating effect of broadband TM [38,41], but those references demonstrate temporal recompression and time-gated correlations, not that a single TM remains valid over a spectral span roughly 30 times the source bandwidth. No spectral filtering at the output is used to show that a detuned wavelength, e.g. 130 nm away, is actually enhanced by the same phase pattern. Without such isolation, the 260 nm value could be inflated by wavelength-integrated background from the 8 nm pulse or by the undefined correlation threshold used to define the bandwidth. Because the second part of the central claim, that broadband excitation further enlarges the accessible bandwidth, rests entirely on this measurement, it is load-bearing and needs an explicit spectral test.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports experiments and a Fresnel-propagation model addressing how the spectral bandwidth of a wavefront-shaped focus through scattering media depends on axial geometry and on the excitation bandwidth. In the quasi-monochromatic regime, the authors show that when the detection plane is translated axially as the wavelength is detuned, speckle correlation and focus quality are recovered over a spectral range several times larger than the conventional 2D speckle correlation bandwidth (e.g., 87 nm versus 18 nm for a diffuser). They interpret this through an axial spatio-spectral coupling and show that the required axial translation is minimized in the confocal geometry. They extend the measurement to a mouse brain slice and to TiO2 samples, and report that the effect disappears in a thick multiply scattering control. In a second set of experiments with a 150 fs, 8 nm bandwidth source, they report a further enlarged 3D bandwidth of about 260 nm and attribute this to coherent time-gating in broadband transmission-matrix measurements.","tokens_in":10766,"tokens_out":3866,"duration_ms":45148,"significance":"If the central claims hold, the paper is significant for multispectral and nonlinear imaging through scattering media: it identifies a practically useful degree of freedom (axial refocusing) that is not captured by the usual 2D spectral correlation bandwidth, and it provides a simple parameter-free Fresnel model for the observed invariance. The experimental work includes valuable controls: the chromatic-aberration control (less than 30% of the observed axial shift) and the thick-TiO2 control showing loss of spatio-spectral memory. The model is not fitted to produce the main effect, and the 3D bandwidth is an empirical observable. The main weakness is that the broadband-enlargement claim, which is the second pillar of the abstract, rests on measurements that lack a direct spectral-isolation test. The reported 260 nm bandwidth therefore needs additional support before the claim can be considered established.","major_comments":[{"comment":"The central broadband claim, that a single broadband transmission matrix yields a refocus over a 260 nm range, is not supported by a spectral isolation measurement. The TM is measured with a source whose spectral width is about 8 nm at λ0 = 785 nm, yet the reported Δλ3D is 260 nm, roughly thirty times the source bandwidth. The manuscript does not show that a phase pattern measured at λ0 produces an actual enhancement at a detuned wavelength, e.g., 130 nm away, when the output is detected through a narrowband spectral filter. Without such a check, the 260 nm value could be inflated by wavelength-integrated intensity from the 8 nm pulse or by the particular correlation threshold used. I request a direct experiment in which the source is tuned to a detuned wavelength and the refocus enhancement is measured with spectral filtering at the output, or an explicit comparison with a multispectral TM measurement over the same range.","section":"Sec. 5, Fig. 5"},{"comment":"The quantitative definition of the 3D bandwidth is under-specified. The text states that Δλ3D is the speckle spectral width obtained when correlations are measured at the planes of highest correlation for each wavelength, but no threshold criterion (e.g., 1/e of the maximum correlation), fit function, or uncertainty is given. The central numerical comparisons (18 nm vs 87 nm vs 260 nm, and the factor-of-3 and factor-of-5 claims) depend directly on this definition. Please state the exact criterion used to extract each bandwidth value, report error bars or confidence intervals, and specify how many independent measurements underlie each reported number.","section":"Sec. 2 and Sec. 5"},{"comment":"The claim that the confocal geometry minimizes the axial extent of the 3D bandwidth is supported only by qualitative agreement with the model and by data shown without error bars. Figure 3c compares experimental points to the law λ z z1/(z−z1) = const but no quantitative measure of agreement is given. Since this is a central conclusion of the paper, please provide a statistical or quantitative assessment of the agreement, or at minimum report the measurement uncertainty on the δz values used to extract the Δz extents.","section":"Sec. 3, Fig. 3"}],"minor_comments":[{"comment":"The acknowledgment heading is numbered 7.1, duplicating the funding heading; it should be 7.2.","section":"Sec. 7"},{"comment":"The affiliation line '3Currently with the Department of Electronic Journals' appears to be erroneous and should be removed or corrected.","section":"Author affiliations"},{"comment":"The 2D speckle correlation curve is fit by a Gaussian, but the fit function and the criterion for extracting the 18 nm and 87 nm widths are not given; please state them explicitly.","section":"Sec. 2, Fig. 2"},{"comment":"The text says the confocal geometry has z2 = z1 = 0, but Fig. 5a shows z1 being varied; please clarify the exact geometry used for the broadband experiments and for each panel of Fig. 5.","section":"Sec. 5, Fig. 5"},{"comment":"The biological tissue results are presented as a single point in Fig. 4 with no indication of the number of samples or repeat measurements; a sentence on sample size and reproducibility would strengthen the claim.","section":"Sec. 4, Fig. 4"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of the journal and the quasi-monochromatic 3D bandwidth effect is well supported. The decision hinges on whether the broadband enlargement claim (260 nm from an 8 nm source) can be backed by a spectral isolation measurement; this is experimentally feasible and should be requested rather than rejecting the manuscript."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The real news here is the axial spatio-spectral coupling: when you shift the wavelength, moving the observation plane along z recovers a good focus, and the resulting 3D spectral bandwidth runs 5–15 times larger than the 2D speckle bandwidth. That is an actual experimental observation, not a fitting artifact, and it is the paper's strongest contribution. The Fresnel model in Eq. (1) is standard, parameter-free, and reproduces the measured invariance law qualitatively; the authors are honest that it is only qualitative. The controls are also solid: the chromatic aberration contribution is measured and kept under 30%, and the thick TiO2 sample shows no effect, which correctly bounds the regime of validity to thin or forward-scattering media.\n\nWhere the paper gets softer is in the quantitative reporting. Bandwidth values come with no error bars, and the correlation threshold used to define each bandwidth is never stated. That is fixable but matters because the 3D bandwidth numbers are the headline. The theory section is also more of a plausible explanation than a derivation; it does not predict the factor-of-5-to-15 enhancement, so the experimental claim stands on the data alone.\n\nThe bigger soft spot is the broadband result in Sec. 5. The TM is measured with a 150 fs pulse whose spectral width is about 8 nm, yet the paper claims a 260 nm 3D bandwidth. The mechanism cited, coherent time-gating from a broadband TM, is real [38,41], but those papers show temporal recompression and time-gated correlations, not that a single TM remains valid over a span 30 times the source bandwidth. No spectral filter is used at the output to prove that a wavelength 130 nm away from the center is actually enhanced by the same phase mask. Without that check, the 260 nm value could be inflated by wavelength-integrated background from the 8 nm pulse or by the undefined threshold. This is load-bearing: the second central claim, that broadband excitation enlarges the accessible bandwidth, rests entirely on this measurement.\n\nI disagree with the reader only on severity: the 2D-to-3D enhancement is well supported and should survive revision. But the broadband claim needs a direct spectral isolation experiment, or the paper should be reframed so that claim is optional.\n\nWho gets value: anyone working on wavefront shaping for nonlinear bio-imaging, particularly CARS or SFG, where polychromatic refocusing matters. The paper deserves a serious referee, but the referee should push for error bars, a defined threshold, and a spectral check on the broadband result. I would not desk-reject it; I would send it out with a request for revision.","headline":"Solid demonstration of axial spatio-spectral memory in thin scattering media; the broadband 260 nm claim needs a spectral isolation check before it carries the paper's second conclusion.","tokens_in":11293,"tokens_out":2123,"would_cite":true,"duration_ms":23058,"reading_group":"yes","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 claims that a wavefront-shaped refocus through thin or forward-scattering media has a three-dimensional spectral memory, widening the usable range from 18 nm to 87 nm, and to 260 nm with broadband excitation.","keywords":["wavefront shaping","spectral memory","speckle correlation","spatio-spectral coupling","scattering media","confocal geometry","broadband transmission matrix","nonlinear bio-imaging"],"falsifier":"Take the same thin diffuser and vary the distance $z_1$ between the diffuser and the geometric focus from 1 mm to 5 mm while recording the axial shift needed to recover correlation at a fixed detuning; the shift should grow roughly in the way predicted by $\\lambda z z_1/(z-z_1)=\\text{const.}$ A curve independent of $z_1$, or a sample with thickness exceeding its transport mean free path that still shows 3D enhancement, would break the claim.","tokens_in":10385,"feed_emoji":"🔬","tokens_out":7469,"duration_ms":76196,"temperature":0.7,"pith_summary":"Wavefront shaping can refocus light that has passed through a scattering medium, but the refocus was thought to survive only over the medium's speckle correlation bandwidth—about 18 nm in the thin diffuser studied here. This paper shows that the spectral memory is actually three-dimensional: when the wavelength is detuned, the best-focus plane moves axially along a predictable curve, so scanning the detection plane recovers the focus out to about 87 nm, almost five times farther. The axial spread is smallest in a confocal geometry, where the refocus plane sits at the geometric focus; there the usable bandwidth is limited only by the medium itself. Using a broadband femtosecond pulse and measuring the broadband transmission matrix extends the range further, to 260 nm, because the measurement acts as a coherent time gate. These results matter because multicolour nonlinear microscopies—CARS, sum-frequency, and four-wave-mixing—need to focus several wavelengths at once through scattering tissue.","feed_headline":"Axial shifts widen refocus memory from 18 nm to 87 nm","feed_subtitle":"Scanning the detection plane and using broadband pulses extends the usable range to 260 nm for multicolour imaging.","key_machinery":"The machinery is the axial spatio-spectral coupling of speckle in convergent illumination, described by Eq. (1): a parabolic phase inside the Fresnel integral makes wavelength and axial distance interchangeable. This identity creates the 3D spectral memory; it is what allows axial translation $\\delta z$ to compensate wavelength detuning $\\delta\\lambda$. The second mechanism is the broadband transmission matrix, whose self-referenced interferometric measurement acts as a coherent time gate, preserving short trajectories and thereby enlarging the refocus bandwidth beyond the intensity-speckle correlation bandwidth.","core_discovery":"On the authors' terms, the discovery is that a wavefront-shaped refocus through a thin or forward-scattering medium obeys a spatio-spectral memory: the complex field behind the medium retains the parabolic phase of the excitation objective, so changing the wavelength is equivalent, to first order, to moving the observation plane along the optical axis. The Fresnel model (Eq. 1) predicts correlation is conserved along surfaces $\\lambda z z_1/(z-z_1)=\\text{const.}$, and the experiments confirm this curve: with $z_1=5$ mm, the 3D speckle bandwidth $\\Delta\\lambda_{3D}=87$ nm versus $\\Delta\\lambda=18$ nm in 2D, with an axial extent of about 30 $\\mu$m. The confocal arrangement, $z_2=0$, minimizes the axial expansion and makes the measured bandwidth an intrinsic property of the medium. In the broadband regime, a 150 fs, 8 nm pulse measured through the broadband transmission matrix gives $\\Delta\\lambda_{3D}=260$ nm, about three times the monochromatic 3D value and larger than the 80 nm intensity-speckle correlation, an effect the authors attribute to coherent time gating that selects short propagation paths.","pith_inferences":["Because the slope $\\Delta\\lambda_{3D}/\\Delta z$ in the model is governed by $z_1$, the predicted scaling with sample-to-focus distance is directly testable; verifying it across several $z_1$ values would separate the geometrical-phase effect from residual chromatic aberration.","The same spatio-spectral invariance may apply to spatial memory effects: if the medium preserves the parabolic phase, a lateral shift of the input beam might also be convertible into a wavelength shift, extending the result to multispectral wide-field imaging rather than point focusing.","The broadband transmission matrix's coherent gating suggests that the refocus bandwidth may be limited by the incident laser bandwidth rather than by the medium in sufficiently forward-scattering samples; if so, octave-spanning sources could push the usable range further.","The thick-TiO2 control implies a practical selection rule: the 3D bandwidth enhancement is available only when the transport mean free path exceeds the sample thickness; this could be used as a quick diagnostic to decide whether axial spectral scanning will help in an unknown tissue."],"forward_implications":["A single transmission matrix can refocus a range of wavelengths if the detection plane is translated axially along the predicted spatio-spectral curve, removing the need to re-measure the matrix at each colour.","In tissues that scatter mainly forward (anisotropy factor close to 1), millimetre-scale samples should support the same 5- to 10-fold bandwidth enhancement seen in diffusers, so multicolour nonlinear imaging in depth becomes feasible.","Broadband femtosecond excitation should yield refocus bandwidths several times the medium's speckle correlation bandwidth, up to 260 nm in the demonstrated geometry, because coherent gating favours short paths.","Spectral characterization of a scattering medium should report both 2D and 3D bandwidths; quoting only the 2D value underestimates usable bandwidth in thin media.","In non-confocal configurations, the axial shift can be compensated by adding a correction wavefront, potentially making sequential wavelength-by-wavelength optimization unnecessary."],"supporting_citations":[{"why":"Supplies the transmission-matrix measurement method that creates the refocus in all experiments.","marker":"[7]"},{"why":"Establishes the baseline 2D spectral bandwidth of a wavefront-shaped refocus, which this paper shows is not the whole story.","marker":"[14]"},{"why":"Supplies the axial spectral correlations of speckle that motivate the 3D spatio-spectral coupling.","marker":"[20]"},{"why":"Provides the Fresnel convergent-illumination model and the $\\lambda z z_1/(z-z_1)$ invariance used to explain the axial shift.","marker":"[21]"},{"why":"Accounts for the lateral spatio-spectral dilatation that explains small discrepancies between speckle and refocus correlation curves.","marker":"[22]"},{"why":"Supplies the anisotropic factor and transport mean free path values used to argue that brain tissue is forward-scattering at 1 mm thickness.","marker":"[28]"},{"why":"Links the large geometrical and angular memory in anisotropic tissues to the large 3D spectral bandwidths observed.","marker":"[29]"},{"why":"Generalizes shift and tilt memory, which the paper invokes to explain spectral-memory capacity in anisotropic media.","marker":"[30]"},{"why":"Provides the broadband transmission matrix measurement and coherent time-gating mechanism claimed to enlarge the 3D bandwidth to 260 nm.","marker":"[38]"},{"why":"Supports the claim that time gating enlarges spectral capabilities, used to explain the broadband enhancement.","marker":"[41]"}],"fun_headline_variants":["Spectral refocus memory expanded 5-fold via axial shifts","Broadband pulses triple refocus bandwidth through scattering","Spatio-spectral memory extends refocus to 260 nm bandwidth","Axial scanning widens spectral refocus from 18 to 87 nm"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The effect assumes the sample is thin enough, or scatters mostly forward, that the lens's focusing curvature is not completely erased; in a thick sample that fully scrambles the wavefront, the effect is absent.","fun_headline_variants_meta":{"raw":{"variants":["Spectral refocus memory expanded 5-fold via axial shifts","Broadband pulses triple refocus bandwidth through scattering","Spatio-spectral memory extends refocus to 260 nm bandwidth","Axial scanning widens spectral refocus from 18 to 87 nm"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000474,"raw_usage":{"total_tokens":2337,"prompt_tokens":912,"completion_tokens":1425,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":528,"completion_tokens_details":{"reasoning_tokens":1352}},"tokens_in":528,"tokens_out":1425,"duration_ms":11532,"temperature":1.0,"reasoning_tokens":1352,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T15:19:06.468451+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the same thin diffuser and vary the distance $z_1$ between the diffuser and the geometric focus from 1 mm to 5 mm while recording the axial shift needed to recover correlation at a fixed detuning; the shift should grow roughly in the way predicted by $\\lambda z z_1/(z-z_1)=\\text{const.}$ A curve independent of $z_1$, or a sample with thickness exceeding its transport mean free path that still shows 3D enhancement, would break the claim.","supporting_citations":[{"cited_title":"Measuring the Transmission Matrix in Optics: An Approach to the Study and Control of Light Propagation in Disordered Media,","cited_arxiv_id":null,"evidence_quote":"Supplies the transmission-matrix measurement method that creates the refocus in all experiments."},{"cited_title":"Frequency bandwidth of light focused through turbid media,","cited_arxiv_id":null,"evidence_quote":"Establishes the baseline 2D spectral bandwidth of a wavefront-shaped refocus, which this paper shows is not the whole story."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the axial spectral correlations of speckle that motivate the 3D spatio-spectral coupling."},{"cited_title":"Information inferred from the observation of speckles,","cited_arxiv_id":null,"evidence_quote":"Provides the Fresnel convergent-illumination model and the $\\lambda z z_1/(z-z_1)$ invariance used to explain the axial shift."},{"cited_title":"Spatio-temporal X-wave,","cited_arxiv_id":null,"evidence_quote":"Accounts for the lateral spatio-spectral dilatation that explains small discrepancies between speckle and refocus correlation curves."},{"cited_title":"Optical properties of biological tissues: a review,","cited_arxiv_id":null,"evidence_quote":"Supplies the anisotropic factor and transport mean free path values used to argue that brain tissue is forward-scattering at 1 mm thickness."},{"cited_title":"Characterization of the angular memory effect of scattered light in biological tissues,","cited_arxiv_id":null,"evidence_quote":"Links the large geometrical and angular memory in anisotropic tissues to the large 3D spectral bandwidths observed."},{"cited_title":"Generalized optical memory effect,","cited_arxiv_id":null,"evidence_quote":"Generalizes shift and tilt memory, which the paper invokes to explain spectral-memory capacity in anisotropic media."},{"cited_title":"Temporal recompression through a scattering medium via a broadband transmission matrix,","cited_arxiv_id":null,"evidence_quote":"Provides the broadband transmission matrix measurement and coherent time-gating mechanism claimed to enlarge the 3D bandwidth to 260 nm."},{"cited_title":"Scattering correlations of time-gated light,","cited_arxiv_id":null,"evidence_quote":"Supports the claim that time gating enlarges spectral capabilities, used to explain the broadband enhancement."}],"review_version":1}