{"id":"4d139590-f710-48ca-82ed-01408982abee","arxiv_id":"2608.12116","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":7,"one_line_summary":"First tomographic CIBER×galaxy cross-correlations reveal near-infrared background fluctuations at z<0.6 several times stronger than integrated-galaxy-light model predictions.","lead":"This paper cross-correlates near-infrared background maps from the CIBER rocket experiment with galaxy catalogs from DESI and HSC, and finds excess clustering signal at low redshift. The result suggests that faint galaxies and diffuse light in low-redshift structures produce more near-infrared background fluctuation than standard models predict, which matters for interpreting cosmic background light measurements and future missions like SPHEREx.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Headline claim depends on unpublished co-authored Ares IGL baseline; independent reproduction is required before the 'bright NIR background' interpretation can be accepted.","rationale":"The paper is an impressive observational advance: the first tomographic CIBER×galaxy cross-correlation, with extensive mock validation, foreground checks (Gaia stars), field consistency, and robustness to scale cuts. The measurement of cross-power itself appears credible. However, the paper's central claim is framed as a discrepancy relative to a 'standard IGL' model, and that model is Ares, an unpublished co-authored code. All model-prediction curves in Figs. 4, 5, 6, 8, 9, and 11 depend on this model. The reader's conditional verdict is appropriate: the result should be accepted only after the IGL baseline is independently verified or released. My concrete test would settle whether the excess is robust to the model choice. If an independent model gives the same excess, the claim is strong; if not, the claim reduces to a statement about Ares rather than about the universe. This concern does not change the verdict from CONDITIONAL, because the measurement itself and the internal consistency checks are strong. However, it is the single most load-bearing assumption for the headline interpretation.","tokens_in":39630,"tokens_out":9507,"duration_ms":85488,"concrete_test":"Reproduce the standard IGL cross-power predictions using an independent, publicly available model (e.g., Helgason et al. 2012 galaxy-count-based IGL, or Driver et al. 2016) with the same tracer redshift distributions, bias assumptions, and masking. If the measured C_Ig at ℓ<2000 still exceeds these independent predictions by the same factor, the headline claim is robust to the Ares choice; if the excess shrinks or disappears, the 'bright, low-redshift NIR background' is an artifact of a low Ares baseline. The Ares configuration and outputs should also be released as a supplementary data product so that exact reproduction is possible.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim of 'significantly higher cross-power than expectations from an IGL model' is a comparison against the Ares semi-empirical model (ref [58], in prep). This model supplies the two-halo prediction (via luminosity-weighted bias b_I and dI/dz), the one-halo templates (satellite fractions and luminosity assignment), and the Poisson level. The quoted significances (7.4σ/6.2σ for DESI-LS and 4.3σ/5.3σ for HSC at ℓ<2000) are relative to this single unpublished model, with no propagated model uncertainty. If Ares underpredicts the low-redshift luminosity function, satellite fractions, or intra-halo light, the measured 'excess' and the inferred b_I×dI/dz would be inflated. The DESI-LS/HSC cross-check reduces catalog-specific systematics but does not validate the model, since both are compared to the same Ares baseline. Bracketing b_I up to tSZ-like values only tests the bias, not the assumed dI/dz. Thus the headline result is structurally dependent on an unvalidated proprietary baseline.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents the first tomographic cross-correlation analysis of near-infrared extragalactic background light (EBL) anisotropies, combining CIBER 1.1 and 1.8 μm imaging with photometric galaxy samples from DESI-LS and HSC over z<1. The authors report significant cross-power detections at ℓ<2000, exceeding predictions of the 'Ares' semi-empirical integrated galaxy light (IGL) model by factors of several, with the excess concentrated at z≲0.6. Halo-model fits separate one- and two-halo components, and the inferred bias-weighted intensity kernel b_I×dI/dz is claimed to exceed Ares-based predictions even for an intensity bias as high as that of large SZ clusters. The paper also reconstructs a lower bound on the CIBER auto-power contributed by z<1 large-scale structure, arguing that it explains a substantial fraction of the excess auto-power reported in earlier CIBER work. The pseudo-Cℓ pipeline is validated on 500 mock realizations, with additional Gaia cross-correlation checks and field-consistency tests.","tokens_in":39856,"tokens_out":9242,"duration_ms":83054,"significance":"If the model-dependence of the comparison is resolved, these are important measurements: they constitute the first tomographic EBL-galaxy cross-spectra in the near-infrared, with a validated pipeline, stellar-foreground checks, and internal consistency tests. The claim that low-redshift clustered galaxies and intra-halo light contribute significantly to NIR EBL fluctuations is falsifiable and has direct implications for CIBER-2, SPHEREx, and EoR foreground studies. However, the central quantitative claims are currently conditional on an unpublished co-authored model, so the headline 'bright, low-redshift NIR background' interpretation is not yet independently supported. The underlying cross-power measurements are likely robust and valuable regardless of the model comparison.","major_comments":[{"comment":"The central claim that the cross-power is 'significantly higher than expectations from an IGL model' is made relative to the Ares semi-empirical model [58], which is unpublished and developed by a co-author. The quoted significances (7.4σ/6.2σ for DESI-LS and 4.3σ/5.3σ for HSC at ℓ<2000) are conditional on the Ares two-halo, one-halo, and Poisson predictions, but no model uncertainty is propagated. Since the one-halo templates are generated from the same model, the comparison of fitted A_1h amplitudes to 'predictions' is partly a self-consistency check. If Ares underpredicts the low-redshift luminosity function, satellite fraction, or intra-halo light, the measured 'excess' and the inferred deficiency in b_I×dI/dz would be inflated. The paper should quantify model uncertainties and/or show that the excess persists against an independent IGL baseline (e.g., Helgason et al. 2012 or Driver et al. 2016) before claiming a bright low-redshift NIR background.","section":"§5.1 and §7.1"},{"comment":"The fiducial model gives statistically unacceptable fits in several bins that drive the main conclusions. For CIBER×DESI-LS in z∈[0.2,0.4), Table 3 reports χ²=30.1 (1.1 μm) and 34.6 (1.8 μm) for 13 degrees of freedom, corresponding to χ²_red=2.31 and 2.66; for z∈[0.6,0.8) at 1.8 μm, χ²=37.6 (χ²_red=2.89). These are PTE<0.001 fits, yet Appendix C states that 'our halo model provides acceptable fits to the data' and these same fits provide the A_2h amplitudes that feed the b_I×dI/dz constraints. Model misspecification at this level can bias the inferred amplitudes. The paper should either improve the model for these bins, report the poor fit quality as a caveat on the interpretation, or restrict quantitative conclusions to bins with acceptable χ².","section":"§8.2, Table 3"},{"comment":"The argument that 'standard IGL predictions underestimate our measurements, even when assuming an intensity bias as high as 3' only varies b_I while holding the Ares dI/dz fixed. The four curves in Fig. 11 correspond to different b_I assumptions, not to different dI/dz; therefore the conclusion that 'a higher dI/dz is required' is conditional on the Ares emissivity kernel. To make this claim robust, the paper should compare the derived b_I×dI/dz values with independent dI/dz estimates from galaxy counts (e.g., Driver et al. 2016; Helgason et al. 2012) and propagate their uncertainties. Without such a comparison, the discrepancy is a statement about the Ares model rather than about the NIR background.","section":"§8.2.4, Fig. 11"}],"minor_comments":[{"comment":"The text refers to 'Appendix 4.2' for the random catalog construction, but the relevant appendix is A.3 (and A.4 for HSC).","section":"§4.2"},{"comment":"Equation (8.1) uses P_ℓ for the Gaussian damping factor, but P_ℓ was already used for the Poisson component in §2.3; please use a distinct symbol (e.g., D_ℓ) to avoid confusion.","section":"§8.2.1"},{"comment":"The label 'DESI-LS CMGs with CMGs removed' in the left panel is confusing; it presumably means 'DESI-LS with CMGs removed'.","section":"Fig. 7"},{"comment":"The symbol n_g appears as a comoving density in Eq. (5.2) and as an angular density in Eq. (5.8); please define these separately to avoid ambiguity.","section":"§5.2.3 and Eq. (5.8)"},{"comment":"The claim of 'first tomographic analysis' should be qualified in light of earlier EBL-tomography work cited as [15] and [19]; please clarify the novel element, e.g., the first measurement with NIR intensity maps rather than the first method development.","section":"Abstract and §1"}],"recommendation":"major_revision","confidential_remarks":"The main scientific claim is conditional on an unpublished co-authored model (Ares, ref. [58]). I would ask the editor to consider whether Ares will be released or a public, independent baseline substituted before acceptance, since independent verification of the headline excess is currently impossible. The measured cross-spectra and pipeline validation are strong; the interpretation needs to be de-risked against model uncertainty."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"First, what's new: this is the first tomographic cross-correlation of the CIBER NIR background with photometric galaxies. The detection of excess cross-power at ℓ<2000, z<0.6 is credible: the pseudo-Cℓ pipeline is validated on 500 mocks, the Gaia cross-correlation checks for stellar foregrounds, field-consistency tests are reasonable, and the parameter consistency checks across ℓ_max are reassuring. The two-halo detection, the one-halo amplitude evolution, and the cluster contribution of 15–20% are all genuinely new measurements. The paper is also honest about the lower-bound nature of the auto-power reconstruction from the coherence estimator.\n\nThe soft spot is structural. The 'standard IGL' baseline is the Ares model, reference [58], listed as 'Mirocha, J. et al. 2026, in prep.' — a co-author's unpublished model. That model provides the two-halo prediction (via b_I and dI/dz), the one-halo templates, and the Poisson level. The quoted significances (7.4σ/6.2σ and 4.3σ/5.3σ) are relative to this single model, with no propagated model uncertainty. The DESI-LS/HSC cross-check does not cure this, since both are compared to the same Ares baseline. So the excess relative to Ares is a measurement, but the interpretation that the low-z NIR background is 'bright' rests on Ares's IGL predictions being correct. If Ares under predicts the low-z luminosity function or satellite fractions, the excess is inflated. The paper does bracket b_I with tSZ-like values, which tests bias, but not the assumed dI/dz.\n\nI'd like to see the Ares predictions released or independently reproduced before the 'bright background' claim is taken as definitive. That said, the measurement itself is important and well-executed; this isn't a case where the pipeline is the problem. The paper deserves peer review. The authors should be asked to either make Ares public and benchmark it against other IGL models, or soften the headline accordingly.\n\nWho is this for? People working on EBL fluctuations, low-z galaxy clustering, and intensity mapping. It's a valuable reference for CIBER-2/SPHEREx planning. I'd cite it, but with the caveat about the baseline. Serious thinker: yes, the thinking is clear and the limitations are acknowledged in the text; the issue is a missing independent baseline, not incoherence.\n\nFinally, I'd send this to a good referee rather than desk reject. The measurement should be published; the interpretation needs to be fenced in.","headline":"Solid new measurement; the interpretation leans on an unpublished co-authored model, so the 'bright NIR background' headline is conditional until that baseline is independently benchmarked.","tokens_in":40454,"tokens_out":2796,"would_cite":true,"duration_ms":25045,"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":"Cross-correlating CIBER maps with galaxy catalogs reveals that the near-infrared background fluctuates far more at low redshift than standard models predict, pointing to clustered galaxies and diffuse intra-halo light as the dominant…","keywords":["cosmic infrared background","extragalactic background light","cross-correlation","tomography","large-scale structure","near-infrared","halo model","intra-halo light"],"falsifier":"Cross-correlate an independent near-infrared intensity map with a different instrument and systematics against DESI-LS galaxies at $300<\\ell<2000$ and $z<0.6$; if the recovered $b_I \\times dI/dz$ matches the Ares IGL prediction rather than the CIBER excess, the central claim collapses. A direct measurement of the $z<0.6$ galaxy luminosity function, satellite luminosity fractions, and intra-halo light fractions from deep complete catalogs could also falsify the inferred deficiency.","tokens_in":39396,"feed_emoji":"🔭","tokens_out":5198,"duration_ms":50791,"temperature":0.7,"pith_summary":"The paper reports the first tomographic cross-correlation of near-infrared extragalactic background light (EBL) fluctuations with photometric galaxy catalogs, using CIBER 1.1 and 1.8 micron maps and galaxies from DESI Legacy Survey DR8 and Hyper-Suprime-Cam. It claims that on angular scales corresponding to multipoles $\\ell < 2000$, the measured cross-power exceeds predictions from a standard integrated galaxy light (IGL) model, with the excess concentrated at $z\\lesssim 0.6$. If correct, this identifies low-redshift clustered galaxies and intra-halo light as a major, previously underappreciated contributor to near-infrared background fluctuations that earlier CIBER auto-power measurements had left unexplained. The paper further argues that standard model predictions for the bias-weighted intensity kernel $b_I \\times dI/dz$ are too low at $z<1$, requiring either brighter low-redshift galaxy light or a higher NIR background intensity than galaxy-count models allow.","feed_headline":"Cross-correlations reveal a bright, low-redshift NIR background","feed_subtitle":"CIBER maps paired with galaxy catalogs show excess clustering power at z<1 that standard models miss.","key_machinery":"The load-bearing identity is the two-halo ratio $C^{Ig,2h}_\\ell / C^{gg,2h}_\\ell \\approx (b_I\\,dI/dz)/(b_g\\,dN/dz)$, which lets the authors convert measured galaxy-intensity cross-spectra into constraints on the bias-weighted intensity redshift kernel. The analysis decomposes the cross-power into one-halo, two-halo, and Poisson terms within a halo model, using Navarro-Frenk-White profiles and separate templates for star-forming and quiescent centrals, with a satellite luminosity fraction parameter $f^L_{\\mathrm{sat}}$ to encode satellite and diffuse intra-halo light contributions. The CIBER intensity maps and galaxy overdensity fields are processed through a pseudo-$C_\\ell$ pipeline validated on mocks, and the baseline comparison is the Ares semi-empirical IGL model, which generates galaxy populations calibrated to luminosity functions and star-forming main-sequence measurements.","core_discovery":"The paper establishes that CIBER 1.1 and 1.8 $\\mu$m maps cross-correlated with $z<1$ galaxy samples from DESI-LS and HSC show significantly more power on scales $304<\\ell<2000$ than the standard IGL model predicts, at $7-13\\sigma$ significance depending on band and tracer. The excess is strongest in redshift bins $0.1<z<0.5$, where the measured cross-power exceeds predictions by factors of 5-10, and it appears in both one-halo and two-halo terms of a parametric halo model decomposition. Cluster member galaxies contribute 15-20% of the large-angle cross-power despite being fewer than 4% of the sample, but most of the signal comes from group- and galaxy-scale halos. Converting the two-halo fits to $b_I \\times dI/dz$ shows that even an intensity bias as high as that of large SZ clusters cannot reconcile the standard IGL model with the measurements, implying that the low-redshift NIR intensity kernel is underpredicted. Finally, a coherence-based reconstruction shows that correlated large-scale structure at $z<1$ accounts for a substantial fraction of the CIBER auto-power previously reported, meaning the long-standing auto-power excess is largely a low-redshift clustering signal rather than an exotic high-redshift component.","pith_inferences":["If the excess is real, future full-sky near-infrared intensity mappers cross-correlated with dense spectroscopic samples should localize the signal to specific halo masses and redshifts, potentially separating satellite galaxies from diffuse intra-halo light through the shape of the one-halo term.","A testable corollary of the similar one-halo amplitudes between shallow and deep catalogs is that flux-weighted cross-correlations split by stellar mass should show stronger signal in lower-mass bins than standard halo-occupation models predict.","The inferred higher $dI/dz$ at $z<0.6$ predicts a specific spectral energy distribution; multi-band cross-correlations could check whether the excess has a stellar-continuum shape, distinguishing extra galaxy light from more exotic contributions."],"forward_implications":["Correlated large-scale structure at $z<1$ accounts for a substantial portion of the CIBER auto-power excess on scales $300<\\ell<2000$ when combined with stellar and diffuse Galactic light estimates.","Both one-halo and two-halo clustering are detected in the cross-spectra at high significance, with the one-halo amplitude similar between DESI-LS and the deeper HSC catalog, pointing to satellites and/or diffuse intra-halo light in lower-mass halos.","Cluster member galaxies contribute 15-20% of the large-angle cross-power, so groups and galaxy-scale halos rather than massive clusters dominate the signal.","Standard IGL predictions for $b_I \\times dI/dz$ at $z<0.6$ fall below the measurements even with a tSZ-like intensity bias, implying a higher $dI/dz$ that tensions with galaxy-count and gamma-ray EBL constraints but aligns with absolute photometric measurements."],"supporting_citations":[{"why":"Supplies the Ares semi-empirical IGL model that defines the baseline predictions the measurements are compared against.","marker":"[58]"},{"why":"Provides the CIBER fourth-flight auto-power measurements and fiducial science masks used for the cross-correlation analysis.","marker":"[32]"},{"why":"Details the CIBER fourth-flight data reduction and fluctuation analysis methodology that the pipeline builds on.","marker":"[31]"},{"why":"Earlier CIBER fluctuation measurements that found excess auto-power and motivate the cross-correlation search.","marker":"[98]"},{"why":"Supplies the DESI Legacy Survey DR8 photometric redshift catalog used as one of the galaxy tracers.","marker":"[29]"},{"why":"Supplies the HSC Public Data Release i-band catalog with photometric redshifts used as the deeper tracer.","marker":"[2]"},{"why":"Provides the cluster member galaxy catalog used to quantify the cluster contribution to the cross-power.","marker":"[92]"},{"why":"Earlier stacking analysis that detected a one-halo amplitude using the same CIBER maps, extended here to full cross-spectra.","marker":"[16]"}],"fun_headline_variants":["Tomographic CIBER-galaxy cross-correlations reveal low-z NIR excess","Low-redshift structure dominates near-IR background fluctuations","CIBER cross-correlations find bright low-z NIR glow unseen by models","Galaxy clustering explains excess NIR background power at z<1","New cross-correlation analysis exposes hidden low-z NIR background"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The central result depends on the Ares semi-empirical model being a correct standard-baseline description of integrated galaxy light, including its low-redshift galaxy luminosity function, satellite fractions, and intra-halo light; if that baseline is too faint, the reported excess and the inferred deficiency in $b_I \\times dI/dz$ are inflated.","fun_headline_variants_meta":{"raw":{"variants":["Tomographic CIBER-galaxy cross-correlations reveal low-z NIR excess","Low-redshift structure dominates near-IR background fluctuations","CIBER cross-correlations find bright low-z NIR glow unseen by models","Galaxy clustering explains excess NIR background power at z<1","New cross-correlation analysis exposes hidden low-z NIR background"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000628,"raw_usage":{"total_tokens":3033,"prompt_tokens":1203,"completion_tokens":1830,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":819,"completion_tokens_details":{"reasoning_tokens":1739}},"tokens_in":819,"tokens_out":1830,"duration_ms":12974,"temperature":1.0,"reasoning_tokens":1739,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-16T00:15:51.666405+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Cross-correlate an independent near-infrared intensity map with a different instrument and systematics against DESI-LS galaxies at $300<\\ell<2000$ and $z<0.6$; if the recovered $b_I \\times dI/dz$ matches the Ares IGL prediction rather than the CIBER excess, the central claim collapses. A direct measurement of the $z<0.6$ galaxy luminosity function, satellite luminosity fractions, and intra-halo light fractions from deep complete catalogs could also falsify the inferred deficiency.","supporting_citations":[{"cited_title":"CIBER 4th flight fluctuation analysis: Measurements of near-IR auto- and cross-power spectra on arcminute to sub-degree scales","cited_arxiv_id":"2501.17933","evidence_quote":"Provides the CIBER fourth-flight auto-power measurements and fiducial science masks used for the cross-correlation analysis."},{"cited_title":"CIBER 4th flight fluctuation analysis: Pseudo-power spectrum formalism, improved source masking and validation on mocks","cited_arxiv_id":"2501.17932","evidence_quote":"Details the CIBER fourth-flight data reduction and fluctuation analysis methodology that the pipeline builds on."}],"review_version":1}