{"id":"55ef6b3b-383d-4899-a980-79822dc12da5","arxiv_id":"1908.03552","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":3,"one_line_summary":"Deep VISIR imaging detects extended polar mid-infrared emission in 8 of 9 obscured AGN predicted to show it, strengthening the case that polar dust is ubiquitous and dominates AGN mid-infrared emission.","lead":"This paper uses new, deep mid-infrared images to test whether active galaxies commonly show extended dusty emission along their polar axis. The prediction is confirmed in eight of nine objects, supporting the idea that a polar dusty wind, rather than a torus, produces most of the mid-infrared light of these nuclei.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 8/9 detection claim is robust, but the 'polar dust dominates the MIR' headline depends on transferring the median MIDI polar fraction (67±13%) to the unresolved VISIR cores of a different sample; that transfer is unmeasured and per-object thresholds are not met for several objects.","rationale":"The paper is a careful detection experiment and the central detection claim is credible: the prediction from A16 is tested out-of-sample, PSF subtraction is checked with standard stars, and the 350-image simulation quantifies the by-eye bias. The weakest link is the extrapolation from interferometric polar fractions to the unresolved VISIR cores. I agree with the reader's CONDITIONAL verdict: the detection claim should be accepted, but the dominance claim should be presented as conditional on the unmeasured compact polar fraction. The per-object arithmetic shows the median is robust to the self-shielding correction, so the more precise issue is not the median but the lowest-Rext objects, where dominance requires a substantial compact polar fraction that has not been observed. A MATISSE follow-up on those objects would settle it. The minor NGC 2110 selection issue slightly weakens the '8/9 predicted' framing but does not change the verdict.","tokens_in":21548,"tokens_out":11765,"duration_ms":125051,"concrete_test":"Measure the compact (sub-0.3 arcsec) MIR polar fraction with VLTI/MATISSE for the three new sources with the lowest Rext_AGN (NGC 5135, NGC 5506, NGC 7582), and compare the VISIR unresolved-core flux in a 0.5 arcsec aperture to the MATISSE zero-baseline flux for these and for A16 objects with existing MIDI measurements. If for any of the low-Rext sources the measured compact polar fraction falls below the required per-object threshold (36%, 26%, 11%), then the claim that polar dust dominates the total MIR emission in that object—and hence 'on average'—is not supported. At minimum, recompute the sample median Rpol_AGN using per-object MIDI fractions instead of the single transferred median; if the median drops below 50%, the dominance conclusion fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The detection experiment (Sections 3–4.2) is well supported: PSF-subtraction checks, standard-star comparisons, and 350 simulations validate the presence of extended emission. The load-bearing weak point is the dominance claim in Section 4.3.2. Equation Rpol_AGN = Rext_AGN + (100 − Rext_AGN)(Rpol_MIDI/100) assumes that (i) the VISIR-unresolved core flux equals the MIDI zero-baseline flux for each source and (ii) the Lopez-Gonzaga et al. (2016) median polar fraction Rpol_MIDI = 67 ± 13%, measured in a small, different sample, applies to the unresolved cores of these nine new objects plus the A16 extended objects. The §4.3.1 simulations validate Rext_AGN but not this transfer. The median is not fragile: even with the self-shielding correction reducing Rpol_MIDI to 51%, the median Rpol_AGN stays ~75%. The fragile part is per-source: for NGC 5135 (Rext=22%) dominance requires the compact polar fraction to be >36%; for NGC 5506 (Rext=32%) it requires >26%; for NGC 7582 (Rext=44%) >11%. These compact polar fractions have not been measured for the new objects, so the statement that polar dust dominates the total MIR emission on average is an extrapolation rather than a direct measurement. In addition, one of the nine objects (NGC 2110) was added specifically because previous imaging suggested extension, so the prediction-based detection rate is 7/8, not 8/9; this is a minor but real selection blemish.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper presents a detection experiment for polar dust emission in AGN. Using VLT/VISIR imaging of nine [O IV]-bright, obscured Seyfert galaxies that were predicted, on the basis of a threshold established in Asmus, Hönig & Gandhi (2016, A16), to show extended mid-infrared (MIR) emission, the author reports extended emission in eight of nine objects at more than 3σ and in the remaining object at roughly 2–3σ. The extended emission aligns with the system axis as inferred from [O III] cones, masers, and radio morphology, supporting a polar origin. Combining the VISIR-resolved fractions with the median polar fraction from VLTI/MIDI observations (Rpol_MIDI = 67 ± 13%, L´opez-Gonzaga et al. 2016), the paper estimates that polar dust contributes on average 83% of the total AGN MIR emission, concludes that polar dust dominates the MIR energy budget, derives lower limits on the physical extent of the polar dust (median ~200 pc), and reports a tentative correlation between MIR size and Eddington ratio.","tokens_in":21857,"tokens_out":4326,"duration_ms":41782,"significance":"If the dominance claim holds, this is a field-level result: it would strengthen the case that the classical dusty torus is not the dominant MIR-emitting structure in AGN, that polar dusty winds are ubiquitous at least in the Seyfert regime, and that MIR-derived covering factors from torus models must be revisited. The detection experiment itself is well executed and unusually well tested: the paper includes PSF-subtraction checks against observatory standard stars, a 350-image simulation quantifying a modest −6% bias and 11% scatter in the resolved-fraction estimator, and a discussion of the flat-profile assumption, including a worst-case zero-emission-center test that changes Rext_AGN by only ~7%. The prediction being tested is the author's own from A16, and the same paper defined the [O IV] threshold; however, the nine new objects were not used to set that threshold, so the 8/9 (or 7/8, counting only the selection-defined targets) detection rate is a genuine out-of-sample test rather than a circular fit.","major_comments":[{"comment":"The claim that `the polar dust on average dominates the total MIR emission of AGN' depends on two unverified identifications: (i) that the VISIR-unresolved core equals the MIDI zero-baseline flux for each source, and (ii) that the Lopez-Gonzaga et al. (2016) median polar fraction Rpol_MIDI = 67 ± 13%, measured in a small and partially different sample, applies to the unresolved cores of the nine new objects and the A16 extended objects. None of the nine new objects has published MIDI data, and for individual sources the required compact polar fraction is not measured: for NGC 5135 (Rext = 22%) dominance requires a compact polar fraction >36%, for NGC 5506 (Rext = 32%) >26%, and for NGC 7582 (Rext = 44%) >11%. Because these thresholds are not established, the 83% median is an extrapolation rather than a direct measurement. I recommend rewording the abstract and Section 5 to present the dominance statement as an inference conditional on the MIDI fraction applying to these cores, and ideally adding per-object MIDI constraints or a sensitivity analysis showing how the conclusion changes if Rpol_MIDI varies per source.","section":"§4.3.2, Eq. Rpol_AGN = Rext_AGN + (100 − Rext_AGN)(Rpol_MIDI/100)"},{"comment":"NGC 2110 was added to the sample explicitly because previous MIR imaging suggested possible extension (`we further added another source, NGC 2110, because previous MIR imaging indicated that its nucleus is possibly extended'), rather than because it passed the [O IV]-based prediction. The paper states that `the prediction is that all of these 8 should exhibit detectable polar MIR emission' and then treats NGC 2110 as part of the test in the abstract and Section 5 (`Extended emission was detected in 8 out of 9 cases'). The prediction-based detection rate is therefore 7/8, not 8/9. This is a selection blemish that should be reported transparently; the paper should state both the rate among the eight prediction-selected objects and the rate including the added target, and show that 7/8 still constitutes a statistically significant confirmation of the prediction.","section":"§2, sample selection (NGC 2110)"}],"minor_comments":[{"comment":"Section 5 states that extended emission was detected in 8 out of 9 cases at more than 3σ and in the remaining case at ~2σ, but Section 4.1 describes NGC 5135 as showing possible extended structures at ~3σ (`as indicated by the low significance structures in the image (~3σ)'). These statements are inconsistent; the significance assigned to the marginal detection should be reconciled.","section":"§5 vs §4.1 (NGC 5135 significance)"},{"comment":"The assertion that `the VISIR unresolved component corresponds to the total flux as seen MIDI, i.e., the value at baseline length 0' is load-bearing for the dominance calculation but is presented without a specific justification for these sources; a citation to Burtscher et al. (2013) is given for a related point, but the equivalence should be argued explicitly or flagged as an assumption.","section":"§4.3.2 (MIDI zero-baseline identification)"},{"comment":"The Figure 1 caption reads `All images where slightly smoothed' and should read `were'; elsewhere the text has minor typos such as `conrmed' and inconsistent use of `Lopez-Gonzaga' with and without the accent.","section":"Figure captions and typography"},{"comment":"The in-text citation `Almeida & Ricci 2017' does not match the reference-list entry `Almeida C. R., Ricci C.'; the correct surname is Ramos Almeida and should be cited consistently throughout.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The detection experiment is solid and well validated; the 350-image simulation and standard-star checks are a genuine strength. The paper's main problem is that the abstract's headline (`polar dust is dominating the total MIR emission') goes beyond what the data directly measure, because the MIDI polar fraction is transferred from a small, different sample without per-object verification. This is fixable through careful rewording and, if available, additional MIDI data or sensitivity tests. I do not see grounds for rejection, as the central detection claim is well supported and the dominance conclusion can be reframed as a conditional inference. The paper would also benefit from a more careful treatment of the NGC 2110 selection disclosure."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nRead Asmus's new VISIR paper. Short version: the detection experiment is solid and worth taking seriously; the broader dominance claim is not as well grounded as the abstract suggests.\n\nWhat is actually new: nine obscured AGN selected from a prediction in the author's own A16 work—objects with bright [O IV] should show polar MIR emission detectable from the ground. Eight of nine show extended emission at >3σ (the ninth at ~2σ). Crucially, these nine were not used to set the [O IV] threshold, so this is a real out-of-sample test. The robustness section is the best part: subtraction with their own calibrators, consistency checks with standard stars, and 350 simulated images give a small −6% bias and ~11% scatter. The PA alignment between the MIR extensions and the system axes (median difference 23°) supports the polar interpretation. That is a genuine, useful step toward establishing polar dust as a common AGN component.\n\nThe soft spot is Section 4.3.2. The claim that polar dust dominates the total MIR emission comes from combining the VISIR resolved fractions with the MIDI polar fraction from López-Gonzaga et al. (2016), measured in a small, different sample. The equation Rpol = Rext + (100−Rext) × (Rpol_MIDI/100) assumes the VISIR-unresolved core equals the MIDI zero-baseline flux for each source and that the old median transfers. The authors do test the self-shielding correction and show the median stays ~75%, but per-source it is fragile: several new objects have low resolved fractions, and dominance would require compact polar fractions of 26–36% that have not been measured. This is an extrapolation, not a measurement. The paper should say so more clearly.\n\nTwo minor blemishes: NGC 2110 was added because previous imaging suggested extension, so the prediction-based detection rate is effectively 7/8. And the individual Rext values are presented without per-object uncertainties; the simulations give a global scatter, not object-specific error bars. The Eddington-ratio size trend is explicitly tentative, and the authors caution appropriately.\n\nThe citation pattern is fine. The self-citation to A16 is warranted because that is where the prediction and threshold come from. There is no data or code release, but the observations are in the ESO archive and the analysis is described in enough detail to reproduce.\n\nWho gets value: AGN observers and anyone working on torus/unification or MIR–X-ray relations. It deserves serious peer review; a good referee should push for a tempered dominance claim and per-object uncertainties. I'd take it.","headline":"A genuinely out-of-sample detection experiment with careful PSF-subtraction work; the 8/9 detection rate is the real result, while the 'dominance' headline is an extrapolation that should be softened.","tokens_in":22453,"tokens_out":2422,"would_cite":true,"duration_ms":24822,"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":"Polar dust, not the torus, dominates the mid-infrared light of active galaxies.","keywords":["active galactic nuclei","polar dust","mid-infrared emission","AGN unification","dusty torus","Seyfert galaxies","mid-infrared imaging","Eddington ratio"],"falsifier":"Observe the same nine nuclei at about 0.1-arcsecond resolution, measure the extended fraction without assuming a flat profile, and combine it with per-object interferometric polar fractions; the central claim is refuted if the median total polar fraction drops below 50%, or if any individual source shows its extended emission aligned with the host disk rather than the ionization axis.","tokens_in":21267,"feed_emoji":"🔭","tokens_out":10668,"duration_ms":106601,"temperature":0.7,"pith_summary":"The paper tests whether the dusty polar outflows seen around the best-studied active galactic nuclei (AGN—galaxies whose centers host an accreting supermassive black hole) are a universal feature rather than a curiosity. It images nine [O IV]-bright, obscured Seyfert-type galaxies with deep subarcsecond-resolution mid-infrared observations and finds extended emission in all nine, eight at high significance; that emission aligns with each galaxy's ionized-gas cone, not with host-galaxy structure. Combining the new resolved fractions with earlier interferometric measurements of the compact core, the paper argues that the polar component carries, on average, about 83% of the total AGN mid-infrared emission. If correct, the classical obscuring torus is not the main mid-infrared emitter, and torus-derived quantities—covering factors and spectral-energy-distribution classifications—would need to be rethought.","feed_headline":"Polar dust, not the torus, rules AGN mid-infrared light","feed_subtitle":"In nine obscured Seyferts, extended axis-aligned mid-infrared emission carries about 83% of the light.","key_machinery":"The load-bearing selection rule is an empirical threshold: earlier imaging showed that every AGN with a [O IV] $25.89\\,\\mu$m line flux above about $6\\times10^{-13}\\,\\mathrm{erg\\,s^{-1}\\,cm^{-2}}$ had resolvable polar mid-infrared emission, because that line traces the ionization cone and is isotropic, so brighter [O IV] selects larger or closer polar structures. The measurement that converts images into an energy budget is scaled point-source subtraction: the unresolved core is represented by a nearby calibrator star's point-spread function, subtracted at increasing amplitude, and the remaining flux inside a 2-arcsecond aperture defines the extended fraction $R_{\\rm ext}$. Under the assumption of a flat surface-brightness profile for the extended component, the paper obtains $R_{\\rm ext}$ values that, according to its own simulations, are if anything underestimates. To pass from $R_{\\rm ext}$ to the polar fraction of the total AGN mid-infrared emission it uses the identity $R_{\\rm pol} = R_{\\rm ext} + (100 - R_{\\rm ext}) R_{\\rm pol,MIDI}/100$, inserting the interferometric polar fraction $R_{\\rm pol,MIDI} = 67\\% \\pm 13\\%$ and the median $R_{\\rm ext} = 49\\%$ to arrive at a median $R_{\\rm pol} = 83\\%$.","core_discovery":"On the paper's own terms, the mid-infrared light of an active galactic nucleus is dominated by dust that is not in an equatorial torus but in a polar, roughly conical region extending from a few parsecs out to hundreds of parsecs, and this polar dust is present in essentially every AGN that is luminous and inclined enough to reveal it. The evidence is a deliberately selected sample: all nine obscured Seyferts with [O IV] fluxes above the empirical threshold at which extended mid-infrared emission had previously become resolvable show such emission after subtracting the compact core, and the measured position angles agree with each system's known polar axis to a median of 23 degrees. The paper then combines the resolved extended fraction with the sub-parsec-scale polar fraction determined by interferometry on a smaller set of objects, computing a median total polar fraction of 83% for the combined sample. It also finds that the size of the mid-infrared emission grows, tentatively, with Eddington ratio, which it reads as a possible widening of the polar outflow at high accretion rates.","pith_inferences":["If polar dust is as ubiquitous as this sample suggests, then unresolved mid-infrared photometry of distant AGN—where only a point source is seen—should be interpreted as a mixture of torus and polar emission, and single-component torus fits will systematically bias black-hole-growth and obscuration statistics; a testable extension is to stack resolved sizes at fixed luminosity to see whether the 1","The 67% interferometric polar fraction comes from a handful of well-studied objects; transferring it to all nine sources is the paper's main extrapolation. A direct check would be to run the same PSF-subtraction analysis on the full archival imaging sample and see whether the [O IV]-selected objects above threshold all remain resolved once sensitivity is uniform—if some do not, ubiquity becomes a ","Because the polar dust is inferred to be optically thin on average, its emission should be polarized in a predictable orientation relative to the polar axis; imaging polarimetry at 10–20 μm could verify the dust geometry independently of surface-brightness assumptions.","The Eddington-ratio trend connects naturally to the observed deficit of X-ray-obscured AGN at high Eddington ratios: if outflows widen with accretion rate, the covering factor of obscuring gas falls, which would predict a negative correlation between mid-infrared size and X-ray column density at fixed luminosity."],"forward_implications":["Covering factors of circumnuclear obscuration derived from mid-infrared-to-bolometric luminosity ratios would overestimate the true covering factor, because the mid-infrared includes optically thin polar dust that does not obscure the line of sight.","The mid-infrared–X-ray luminosity relation should be tight and largely independent of viewing angle, since the dominant mid-infrared component is optically thin and roughly isotropic; this matches the small observed scatter.","Spectral energy distribution fitting with clumpy-torus models alone would misattribute the mid-infrared bump to a torus, so spatial information must be included to break degeneracies between torus and polar-wind geometries.","The detection rate in a sample selected purely by [O IV] flux and obscuration supports a picture in which polar dust is an integral part of AGN structure, not a rare accident, at least across the Seyfert luminosity regime.","Mid-infrared emission sizes that increase with Eddington ratio, if confirmed, imply that the opening angle of the dusty outflow widens as accretion rate rises, linking the polar dust to radiation-pressure-driven feedback."],"supporting_citations":[{"why":"Supplies the empirical [O IV] flux threshold and the earlier sample of resolved AGN on which the selection prediction is based.","marker":"A16"},{"why":"Provides the median interferometric polar fraction Rpol_MIDI = 67 ± 13% used to convert resolved VISIR fractions into total polar fractions.","marker":"López-Gonzaga et al. 2016"},{"why":"Establishes that the VISIR-unresolved core corresponds to the zero-baseline total interferometric flux, the link that lets the two instruments be combined.","marker":"Burtscher et al. 2013"},{"why":"Justifies [O IV] 25.89 μm as an isotropic indicator of intrinsic AGN luminosity, which anchors the sample-selection logic.","marker":"Meléndez et al. 2008"},{"why":"Shows that the resolved morphology and interferometric visibilities of the Circinus AGN cannot be reproduced by a clumpy torus, motivating the polar-wind interpretation.","marker":"Stalevski, Tristram & Asmus 2019"},{"why":"First interprets the resolved polar structure as a dusty radiation-driven wind, the physical model the new detections are testing.","marker":"Hönig et al. 2012"},{"why":"Provides the absolute flux calibration of the nearby calibrator stars used for photometry and point-spread-function subtraction.","marker":"Cohen et al. 1999"}],"fun_headline_variants":["Torus dethroned: AGN mid-IR dominated by polar dust","Polar dust wins in AGN: 8 of 9 show extended MIR","Mid-IR from AGN? It's polar dust, not the torus","Polar dust dominates AGN mid-IR, torus takes a backseat"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The dominance result holds only if the compact unresolved core seen in the images is the same emission that interferometry measures at zero telescope separation, and if the 67% polar fraction measured in a small set of well-studied objects applies to all nine sources.","fun_headline_variants_meta":{"raw":{"variants":["Torus dethroned: AGN mid-IR dominated by polar dust","Polar dust wins in AGN: 8 of 9 show extended MIR","Mid-IR from AGN? It's polar dust, not the torus","Polar dust dominates AGN mid-IR, torus takes a backseat"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000592,"raw_usage":{"total_tokens":2817,"prompt_tokens":1032,"completion_tokens":1785,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":648,"completion_tokens_details":{"reasoning_tokens":1700}},"tokens_in":648,"tokens_out":1785,"duration_ms":12368,"temperature":1.0,"reasoning_tokens":1700,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:09:16.505072+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Observe the same nine nuclei at about 0.1-arcsecond resolution, measure the extended fraction without assuming a flat profile, and combine it with per-object interferometric polar fractions; the central claim is refuted if the median total polar fraction drops below 50%, or if any individual source shows its extended emission aligned with the host disk rather than the ionization axis.","supporting_citations":[],"review_version":1}