{"id":"fa0fae64-5e5d-4f0a-9e34-8d1ef4fe19ef","arxiv_id":"2608.07221","paper_version":1,"verdict":"UNVERDICTED","confidence":"HIGH","novelty_score":1.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A comprehensive observational review of galaxy quenching, synthesizing two decades of survey results and laying out the empirical case that star formation cessation is a mass-dependent, globally regulated process.","lead":"This paper is a review of observations on why galaxies stop forming stars, covering surveys from the local Universe to the earliest galaxies seen by JWST. A generalist might read it to see how astronomers define 'quenching', which galaxy properties correlate with it, and why it matters for testing the standard cosmological model.","discovery_kind":"review","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The claim that central velocity dispersion is uniquely fundamental to central galaxy quenching may be an artifact of random forest feature-importance under collinearity; halo mass receives zero importance only because importance was split arbitrarily.","rationale":"Reader's weakest assumption identifies the random forest feature-importance analyses as the key vulnerability; I agree that this is load-bearing but locate the problem more precisely in predictor collinearity rather than unmeasured confounders. Halo mass, stellar mass, bulge mass, and central velocity dispersion are strongly inter-correlated in the SDSS samples used. Impurity-based feature importance in random forests is known to be unreliable under this condition: the algorithm can arbitrarily assign importance to one correlated variable and near-zero to others (Strobl et al., 2008). The zero importance of halo mass in Fig. 14 could reflect that the forest happened to split on sigma first, not that sigma is causal. The paper acknowledges the conditional nature of RF results and cites mock tests, which is good practice, but the summary conclusion ('the only parameter to survive rigorous control of nuisance variables is central velocity dispersion') overstates the evidence. If a re-analysis with conditional permutation importance or with sigma removed shows halo mass recovers high importance, the review's central synthesis—that quenching of centrals is governed by central velocity dispersion, linking to AGN feedback and the ΛCDM problems—would be undermined. This does not invalidate the review's value as a synthesis, but it means the strongest claim is not yet established. Hence no verdict change, but the caveat should be more prominent.","tokens_in":51074,"tokens_out":6191,"duration_ms":60590,"concrete_test":"Re-run the Bluck et al. (2022) random forest classification (Fig. 14 bottom-right) on the identical SDSS/Yang group catalog data with central velocity dispersion removed from the feature set. If the measured importance of halo mass rises to the level previously held by sigma, the two predictors are empirically indistinguishable, and the claim that sigma is uniquely fundamental is unsupported. Additionally compute conditional permutation importance (Strobl et al. 2008) on the original features; if halo mass importance becomes comparable to sigma after conditioning on the correlation structure, the conclusion in §4 fails.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The review's central synthesis (§4.4, Fig. 14 bottom-right; §4.6, Fig. 18; §4 summary) is that central velocity dispersion is the only intrinsic correlator of central galaxy quenching, with halo mass, stellar mass, and morphology having no importance. This underpins the narrative that quenching resolves the three ΛCDM problems via processes tied to the central regions (e.g., AGN feedback through the M_BH–sigma relation). The load-bearing assumption is that the random forest feature-importances correctly identify the causal driver among strongly correlated predictors. Halo mass, stellar mass, bulge mass, and sigma are all tightly correlated in samples like SDSS. Impurity-based random forest importance is known to split importance arbitrarily among collinear predictors; the algorithm may select one variable and assign near-zero importance to the others regardless of causality (e.g., Strobl et al. 2008). The paper itself cautions that RF results are conditional on included observables, but then draws a categorical conclusion. If halo mass is the true driver and sigma is merely a correlated proxy, the review's interpretation of the bimodality problem and the inefficiency problem as being solved by central velocity dispersion collapses. This is the weakest point because the entire chain from observations to the three problems passes through this synthesis.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"This manuscript is Part I of a two-part review series on galaxy quenching, aimed at answering why galaxies cease forming stars. It defines quenching as a significant, long-duration reduction in star formation relative to typical star-forming systems at the same mass and epoch, and frames the problem through three purported fundamental issues with ΛCDM: inefficient star formation, the stability of hot gas haloes in clusters, and galaxy bimodality. The review then synthesizes observational results on identifying quenched galaxies, the star-forming main sequence, intrinsic and environmental correlators of quiescence, spatially resolved quenching, gas physics, and high-redshift JWST results. The central narrative is that central velocity dispersion and bulge mass are the fundamental intrinsic drivers of central galaxy quenching, with environment dominating for satellites, based largely on random forest feature-importance analyses from Bluck et al. (2022) and Brownson et al. (2022).","tokens_in":51348,"tokens_out":2754,"duration_ms":29026,"significance":"If its central synthesis holds, this review would provide a valuable unified observational framework for galaxy quenching, connecting central structure and kinematics to the three cosmological problems and offering a well-organized reference for the field. The manuscript is particularly strong in its breadth of literature coverage, clear definitions, and explicit caveats in several sections, and it reproduces many key figures from the primary literature. However, the load-bearing claim—that central velocity dispersion is uniquely fundamental among strongly correlated galaxy properties—rests on random forest feature-importance results whose interpretation under collinearity is fragile. The review's usefulness as a definitive synthesis therefore depends on how convincingly this methodological concern is addressed.","major_comments":[{"comment":"The central claim that central velocity dispersion, and not halo mass or stellar mass, is the fundamental intrinsic driver of central galaxy quenching rests on random forest feature-importance analyses (Fig. 14 bottom-right; Fig. 18). The manuscript itself correctly states that random forest results are 'conditional on the included observables, sample selection, and central-galaxy focus here, rather than as a universal causal proof,' yet the §4 summary asserts categorically that 'the only parameter to survive rigorous control of nuisance variables is that of central velocity dispersion.' Impurity-based feature importance is known to split importance arbitrarily among strongly collinear predictors (e.g., Strobl et al. 2008), and halo mass, stellar mass, bulge mass, and sigma are all tightly correlated in SDSS-like samples. Because the paper's entire observational synthesis passes through this result, the categorical wording overstates the evidence. Please soften the summary, or better, add a discussion of the collinearity limitation and cite corroborating evidence from methods not based on impurity importance (e.g., partial correlations, direct tests, or permutation-based approaches).","section":"§4.4, §4.6, and §4 summary"},{"comment":"The statement 'If one cannot rectify the shapes of the mass functions in Fig. 2, then (despite all of its successes) ΛCDM cannot be an accurate description of the Universe' is too strong as written. The discrepancy in Fig. 2 is between the observed stellar mass function and the scaled halo mass function in the absence of baryonic feedback; within ΛCDM, baryonic feedback is part of the theory. The paper itself immediately explains that SN and AGN feedback can rectify the discrepancy, so the quoted sentence appears to conflate 'ΛCDM with no feedback' with 'ΛCDM.' Please rephrase to avoid a straw-man framing, e.g., 'in the absence of baryonic feedback processes, ΛCDM cannot account for the observed mass functions.'","section":"§1.2.1, p.7"},{"comment":"The review dismisses the halo-mass quenching narrative of Woo et al. (2013) (Fig. 21) by appealing to the same random forest analysis that assigns zero importance to halo mass once central velocity dispersion is included (Fig. 14 bottom-right). This is internally consistent only if one fully accepts the random forest feature-importance ranking. Since that ranking is the very point at issue, the dismissal is circular in practice. The manuscript should either present independent evidence against halo mass as a fundamental driver (e.g., direct comparisons of quenching fractions at fixed sigma across halo mass, or causal-inference analyses) or frame the §5.4 discussion as an open tension rather than a resolved one.","section":"§5.4 and §4.4"}],"minor_comments":[{"comment":"Typo: 'image form the LSST' should be 'image from the LSST.'","section":"§1.1"},{"comment":"Typo: 'casual links' should be 'causal links.'","section":"§4.4"},{"comment":"Typo: 'it’s central, disordered kinematics' should be 'its central, disordered kinematics.'","section":"§4.6"},{"comment":"Typo: 'spacially flat' should be 'spatially flat.'","section":"§1.6"},{"comment":"The sentence 'However, the best tests of the dependence of quenching on structure come from kinematics' is a strong assertion; consider adding a citation or softening it, since the preceding discussion is about photometric structure.","section":"§4.5.1"}],"recommendation":"major_revision","confidential_remarks":"The manuscript relies heavily on the author's own prior work (Bluck et al., Brownson et al.) for the central synthesis, including several reproduced figures. This is not itself inappropriate for a review, but the balance of cited evidence for the central claim could be improved by engaging more prominently with independent groups and with the methodological literature on random forest feature importance. The stress-test concern about collinearity is real and should be addressed in the revision; the paper's own caveats already acknowledge but do not resolve it."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The paper is a review, explicitly, so judge it as one. It does a genuinely good job of organizing a sprawling literature: the three-problems framing (§1.2) is useful for teaching, the definition of quenching in §1.3 is sensible and standard, and the coverage of SFR tracers, the SFMS, morphology, environment, resolved IFU work, gas physics, and JWST results is broad and mostly accurate. The figures are well chosen and the caveats about random forests are stated openly in the text. As a synthesis and entry point, it serves a real purpose.\n\nWhat is new is organizational, not scientific. There are no new measurements, derivations, or predictions, which is fine for a review but means the paper cannot resolve the questions it frames. Its weakest load-bearing section is §4.4–§4.6, where the claim that central velocity dispersion is the only fundamental intrinsic correlator rests on random-forest feature importances from the author's own prior work and Brownson et al. The paper does caution that RF results are conditional on included observables, but then draws a categorical conclusion: halo mass and stellar mass receive zero importance. This is exactly the collinearity trap that impurity-based importances are known to suffer (Strobl et al. 2008). The stress-test worry is legitimate: the three-problems narrative hangs on that synthesis, and the synthesis is correlational, not causal. I would not call it fatal — the paper is upfront about its own limits and the broader literature (e.g., Woo et al. 2013) disagrees on halo mass — but the reader should treat that central claim as an interpretation of one statistical method, not a settled result.\n\nCitation pattern is heavy on self-citations, but those are published and relevant. The math checks out against standard references. The writing is clear, if long.\n\nWho gets value: graduate students and researchers entering quenching who want a map of the field and a bibliography. It deserves a serious referee because it will be widely read and cited, and a referee can flag the over-interpretation of RF importance before it hardens into consensus. I would send it out, with the expectation of minor revision.","headline":"A solid, self-aware review that is worth reading for newcomers, but its central claim that velocity dispersion is the unique driver rests on random-forest rankings that the paper itself admits are conditional.","tokens_in":132,"tokens_out":560,"would_cite":true,"duration_ms":17358,"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":"Galaxy quenching—the long-term shutdown of star formation—is the unifying process behind three major failures of ΛCDM galaxy formation, and this review's observational synthesis identifies central velocity dispersion as the fundamental…","keywords":["galaxy quenching","star formation","galaxy bimodality","central velocity dispersion","galaxy environment","ΛCDM galaxy formation","star-forming main sequence","JWST high-redshift galaxies"],"falsifier":"Take central galaxies matched in $\\sigma_c$ and bulge mass and split them by local overdensity or halo mass: if quenched fraction differs significantly across the split, the claim that centrals quench independently of environment fails. Alternatively, rerun the random forest with cold gas mass and black-hole accretion luminosity among the predictors; if $\\sigma_c$ loses importance, the primacy ranking is an artifact of omitted variables.","tokens_in":50877,"feed_emoji":"🌌","tokens_out":8720,"duration_ms":87017,"temperature":0.7,"pith_summary":"This review argues that galaxy quenching—the long-term suppression of star formation below what is typical at the same mass and epoch—is not one phenomenon among many but the central process needed to reconcile the observed galaxy population with ΛCDM. Three failures motivate it: too few baryons end up in stars at both low and high halo masses, hot gas haloes in clusters should cool and collapse but do not, and galaxies split into two distinct red and blue populations. Drawing on two decades of survey data, the review's observational synthesis places central velocity dispersion (and closely tied bulge mass) as the fundamental intrinsic correlator of quenching in central galaxies, with environment—halo mass, local density, and position within the halo—dominating for satellites. If this hierarchy is right, then stellar mass, morphology, and halo mass are proxies, and the field's next step is to find the physics that ties star formation to the central, disordered kinematics of a galaxy.","feed_headline":"Star-formation death tracks velocity dispersion, not mass","feed_subtitle":"An observational synthesis pins central-galaxy quiescence to one kinematic property, with environment dominant for satellites.","key_machinery":"The load-bearing tool is the random forest classification analysis applied to SDSS and MaNGA samples: parameters compete to reduce impurity in separating star-forming from quenched galaxies, so a parameter's importance score measures whether it adds predictive power after controlling for everything else. This machine identified central velocity dispersion as the dominant predictor for centrals and environment for satellites, and also showed, at resolved scales, that global galaxy properties predict the quenched or star-forming state of a region while local surface density predicts the actual star formation rate in star-forming regions. Central velocity dispersion, defined as the luminosity-weighted line-of-sight velocity dispersion within the central kiloparsec, is a dynamical proxy for the depth of the galaxy's potential well and, via the $M_{\\rm BH}$–$\\sigma_c$ relation, for black hole mass.","core_discovery":"The review's central claim is that quenching is the process whose mass- and environment-dependent action explains the shape of the stellar mass function, the stability of hot cluster atmospheres, and the red sequence-blue cloud bimodality. Observational results assembled here single out central velocity dispersion as the one galaxy property that survives control of nuisance variables in predicting whether a central galaxy is quenched: bulge mass and central density correlate strongly, while stellar mass, morphology, disc properties, halo mass, and local density lose predictive power once dispersion is known. For satellite galaxies the same data identify environment as the dominant driver, with quenched fraction rising toward the centers of massive haloes. The review therefore proposes that the longstanding 'mass versus environment' dichotomy maps onto a deeper distinction: intrinsic, core-regulated quenching for centrals versus externally imposed quenching for satellites, with the two producing opposite radial signatures (inside-out for centrals, outside-in for low-mass satellites).","pith_inferences":["Not in the paper, but if the argument holds: gas content should be downstream of $\\sigma_c$, so one could predict the molecular-gas depletion time and star-forming efficiency profile of a central galaxy from $\\sigma_c$ alone.","Not in the paper, but if the argument holds: compact, high-$\\sigma_c$ satellites in low-density environments should remain star forming, which would distinguish core-driven from environment-driven quenching in a way the assembled data do not explicitly isolate.","Not in the paper, but if the argument holds: galaxy-formation models could decouple the two processes—a single halo-wide throttle plus local ISM regulation may reproduce both the bimodality and the resolved star-forming main sequence without sub-kpc quenching recipes.","Not in the paper, but if the argument holds: massive quiescent galaxies at $z\\gtrsim4$ should have high $\\sigma_c$ at fixed mass, and their quenched fraction should be independent of environment."],"forward_implications":["If central velocity dispersion is fundamental, surveys should prioritize kinematic measurements (e.g., $\\sigma_c$) over morphology or stellar mass for predicting quiescence.","Quenching of centrals and satellites are distinct channels: any simulation must reproduce environment-independence for centrals and environment-dominance for satellites, with opposite radial signatures during transition.","Observed quenched-fraction mass relations and the shape of the stellar mass function can be reinterpreted as consequences of a quenching threshold set by central potential rather than total mass.","If quenching is a global process, spatially resolved studies should find that regions in quenched systems are quiescent everywhere, while star formation in star-forming systems is regulated locally.","At high redshift, JWST's massive quiescent galaxies provide a test: their $\\sigma_c$ (or bulge mass) should predict quenching as at low redshift, and environment effects should be weak for centrals."],"supporting_citations":[{"why":"This random forest analysis on SDSS centrals identifies bulge mass and central velocity dispersion as dominant predictors and shows that halo mass and local density carry no importance.","marker":"Bluck et al. 2022"},{"why":"This random forest on MaNGA kinematics separates disordered from ordered motion and finds velocity dispersion overwhelmingly predictive of quenching.","marker":"Brownson et al. 2022"},{"why":"This work establishes mass and environment as separable quenching channels via the red-fraction map in stellar mass-overdensity space.","marker":"Peng et al. 2010"},{"why":"This work establishes the central-satellite dichotomy: centrals quench independently of environment, while satellites show strong environmental dependence.","marker":"Peng et al. 2012"},{"why":"This work shows that halo mass and location within the halo correlate with quenching for centrals and satellites, providing the environmental baseline.","marker":"Woo et al. 2013"},{"why":"This ATLAS3D analysis connects quenching to dynamical mass and velocity dispersion rather than size, supporting the primacy of central kinematics.","marker":"Cappellari et al. 2013b"},{"why":"This CANDELS bulge/disc decomposition shows that bulge mass is the tightest predictor of quenched fraction up to $z\\sim2$.","marker":"Lang et al. 2014"},{"why":"This MaNGA resolved analysis links supermassive black hole mass to inside-out quenching in centrals and environment to outside-in quenching in low-mass satellites.","marker":"Bluck et al. 2020a"}],"fun_headline_variants":["Galaxy death: dispersion beats mass for centrals","Quenching's key: velocity dispersion for centrals","Why galaxies stop forming stars: dispersion drives it","Centrals die by dispersion; satellites by environment","Star-formation halt: dispersion, not mass, decides"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire ranking of central velocity dispersion as the fundamental driver rests on which parameters are put into the random forest; the review itself notes that the analyses are conditional on included observables and sample selection, so if cold gas content or current black-hole activity were added, the importance scores could change.","fun_headline_variants_meta":{"raw":{"variants":["Galaxy death: dispersion beats mass for centrals","Quenching's key: velocity dispersion for centrals","Why galaxies stop forming stars: dispersion drives it","Centrals die by dispersion; satellites by environment","Star-formation halt: dispersion, not mass, decides"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000278,"raw_usage":{"total_tokens":1647,"prompt_tokens":935,"completion_tokens":712,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":551,"completion_tokens_details":{"reasoning_tokens":637}},"tokens_in":551,"tokens_out":712,"duration_ms":7339,"temperature":1.0,"reasoning_tokens":637,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-10T12:28:33.435686+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take central galaxies matched in $\\sigma_c$ and bulge mass and split them by local overdensity or halo mass: if quenched fraction differs significantly across the split, the claim that centrals quench independently of environment fails. Alternatively, rerun the random forest with cold gas mass and black-hole accretion luminosity among the predictors; if $\\sigma_c$ loses importance, the primacy ranking is an artifact of omitted variables.","supporting_citations":[],"review_version":1}