{"id":"dccf5ef4-a13c-4ba1-9e0e-decd00326958","arxiv_id":"2509.01710","paper_version":1,"verdict":"REJECT","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"high","formal_verification":"none","parameter_count":5,"one_line_summary":"More than 85% of 393 intermediate-redshift galaxies show multiple star-forming episodes, and massive galaxies tend to form their stars in one early episode.","lead":"The paper reconstructs the star-formation histories of 393 galaxies at intermediate redshifts and finds that most had more than one episode of star formation, with massive galaxies forming their stars earlier and faster than low-mass ones. It matters because it tests whether the standard picture of galaxy 'downsizing' holds across 6 billion years of cosmic time and connects to JWST discoveries of very early massive galaxies.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Central claim rests on an unvalidated exponential+Gaussian decomposition of a smoothed, light-weighted SFH; with old-SSP degeneracy admitted, the >85% multi-episode fraction is not established.","rationale":"The paper is a good-faith measurement paper: it reports stellar population trends that are consistent with previous work, acknowledges sample-selection biases, and openly states degeneracies. Those parts should be credited. However, the central novel claim—more than 85% of galaxies have multiple star-forming episodes, with the first exponential episode assembling 40–50% of stellar mass by z~3—depends on a decomposition of a smoothed, interpolated SFR into a small number of functional components. The reader's weakest assumption correctly identifies this: the recovered SFH is not independently validated, and the old-SSP degeneracy means the first episode is exactly where the data are least informative. My concern is slightly broader: even before the age degeneracy, the pipeline has no demonstrated calibration. No injection-recovery test is presented, no criterion for choosing the number of Gaussians is given, and the mass-weighting step from light-weighted pPXF weights is not described. The paper's own statements in Sections 4.4 and 4.5 that old SSP spectra are almost indistinguishable and that the initial SFE duration is poorly constrained reinforce the concern. The stacked-spectrum comparison in Appendix A.6 is suggestive but circular, since the stacks are constructed from galaxies already classified by the same decomposition. Therefore the central quantitative claims are not supported as stated. A single injection-recovery test on mock SFHs would settle whether the decomposition is trustworthy; absent that, rejecting the paper's headline claim is the appropriate verdict.","tokens_in":26049,"tokens_out":5340,"duration_ms":62182,"concrete_test":"Generate ~100 mock galaxies with known SFHs covering the claimed regimes—single exponential, exponential+Gaussian, and two closely spaced old bursts—at redshifts and S/N matching the sample (0.1≤z≤0.9, S/N≥5 within 1Reff). Pass each mock through the exact pipeline: E-MILES pPXF with 4th-order multiplicative polynomials, 500 bootstrap realizations, the Section 3.4 median/0.5-Gyr smoothing/interpolation, and the Section 4.4 exponential+Gaussian least-squares SFE fitting with the same selection rule used to count SFEs. The >85% multi-episode claim is supported only if single-episode input SFHs are recovered as single-episode at least as often as the 85% threshold, and if the recovered z>3 mass fraction matches the input to within the quoted uncertainties. Also rerun the SFR reconstruction applying a mass-to-light conversion if one was not applied; if the early mass fraction changes by more","verdict_should_be":"REJECT","load_bearing_attack":"The paper's headline quantitative claims—>85% of galaxies have >1 SFE, and an initial exponential episode builds ~40–50% of stellar mass by z~3—are outputs of Section 4.4's least-squares decomposition of the Section 3.4 SFR into one exponential plus Gaussians. That SFR is derived from pPXF weights, median-filtered and interpolated at 0.5 Gyr resolution. Two load-bearing links are unsecured. (1) The text says the normalized weights are V-band light-weighted fractions (Section 3.3), but no conversion from light-weighted to mass-weighted fractions is described before Eq. (4) uses M⋆(t_i) as a stellar mass. If light weights are used as mass fractions, the reconstructed mass assembly and the early mass fraction are biased and are not the quantity claimed. (2) The number of SFEs is extracted by least-squares fitting without model selection, injection tests, or quoted uncertainties on N_SFE; smoothing/interpolation can create or destroy apparent bumps. The paper itself states in Section 4.4 that SSP spectra at ages ≥5 Gyr are almost indistinguishable and in Section 4.5 that the first SFE duration is poorly constrained and could be multiple bursts. Thus the decomposition in exactly the early regime that sets the ~40–50% mass fraction is degenerate, and the 85% statistic has no demonstrated false-positive control.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper derives stellar population parameters and star formation histories for 393 intermediate-redshift (0.1<z<0.9) galaxies from integrated MUSE spectra within 1 Reff, using pPXF with E-MILES templates and a bootstrap approach. From the reconstructed SFR(t), the authors fit an exponential plus Gaussian components and identify discrete 'star-forming episodes' (SFEs). They report that more than 85% of galaxies have more than one SFE, that massive galaxies have fewer SFEs, and that an initial exponential episode built ~40-50% of stellar mass by z~3. The paper also presents mass-size, metallicity, formation timescale, and environmental trends, and discusses sample selection biases.","tokens_in":26443,"tokens_out":4537,"duration_ms":48974,"significance":"If the episodic decomposition were validated, the paper would provide useful observational constraints on how intermediate-redshift galaxies assemble their stellar mass, with direct implications for downsizing, quenching, and JWST-era high-redshift findings. The sample is moderately large, the MUSE data are of good quality, and the bootstrap treatment of pPXF weights is a positive feature. However, the headline quantitative claims are currently outputs of an ad hoc decomposition of a smoothed, light-weighted SFH. No model selection, no recovery simulations, and no uncertainties on the number of episodes are presented, and the paper itself admits that old SSP spectra are nearly degenerate. As submitted, the central claims are not established.","major_comments":[{"comment":"The pPXF weights are explicitly defined as V-band light-weighted fractional contributions, but Eq. (4) uses M⋆(t_i) as the 'stellar mass assembled at time t_i' with no conversion from light-weighted to mass-weighted fractions. If light weights are used as mass fractions, the reconstructed SFR normalization and all subsequent mass-fraction statements (e.g., the 40-50% assembled by z~3 in Sec. 4.5) are not the claimed quantities. This needs to be corrected, or the claims restricted to light-weighted quantities.","section":"Sections 3.3 and 3.4, Eq. (4)"},{"comment":"The number of SFEs is not measured but is an output of an assumed exponential-plus-Gaussian model fitted by least squares. No model selection criterion (e.g., AIC/BIC), no uncertainties on N_SFE, and no injection-recovery tests are provided. The SFR was median-filtered and interpolated on 0.5 Gyr bins (Sec. 3.4), which can create or suppress local maxima. The >85% multi-episode statistic therefore lacks demonstrated false-positive control.","section":"Section 4.4, Eq. (5)"},{"comment":"The paper states that SSP spectra for ages >=5 Gyr are almost indistinguishable and that the duration of the first SFE is poorly constrained and could consist of multiple bursts. The initial exponential component and its claimed ~40-50% mass fraction by z~3 lie exactly in this degenerate regime. Without mock-SFH recovery tests or a clear demonstration that the decomposition is unique in this regime, the headline conclusions are not supported.","section":"Sections 4.4 and 4.5"}],"minor_comments":[{"comment":"The caption says panels (a) and (b) are colored by 'weight fraction or bolometric luminosity'; clarify which quantity is plotted and how it relates to the V-band light-weighting discussed in Sec. 3.3.","section":"Figure 2 caption"},{"comment":"Figure 1 cites the main sequence from Whitaker et al. (2017), while Figure 11 cites Whitaker et al. (2012). Harmonize the references.","section":"Figures 1 and 11"},{"comment":"The smoothing and interpolation procedure is described only in words ('median SFR for the last 0.5 Gyr', 'interpolated the values using bins of 0.5 Gyr'). Specify the exact algorithm, kernel, and interpolation method so the processing is reproducible.","section":"Section 3.4"},{"comment":"The abstract gives 'more than 85%'; provide the exact number and fraction, including the uncertainty, in the text and Figure 8.","section":"Abstract and Section 4.4"}],"recommendation":"major_revision","confidential_remarks":"The central claims are strong but currently rest on an unvalidated decomposition and on an apparent light-weighted-to-mass conversion gap. If the authors can supply recovery simulations, model selection, and a proper mass conversion, the paper may be publishable; otherwise, the quantitative conclusions should be removed or heavily qualified."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this is a useful data paper with a headline claim that doesn't hold up. The 393 MUSE galaxies are well selected, the stellar population measurements (ages, metallicities, t90) are presented carefully, and the confirmation of local scaling relations (age–σ, mass–metallicity, mass–t90) at 0.1<z<0.9 is a genuine service to the field. The environment analysis (field vs group) is also reasonable, with appropriate caveats about sample biases.\n\nThe problem is Section 4.4. The star-forming episode decomposition is a least-squares fit of one exponential plus Gaussians to a smoothed, interpolated SFR. There is no model selection, no recovery simulation, no error bars on N_SFE. The SFR itself is derived from V-band light-weighted pPXF weights, and I don't see a conversion to mass-weighted fractions before Eq. (4) treats them as stellar masses. If the weights are light-weighted, the early mass fractions are biased because old stars are fainter per unit mass in V. The paper's own admission in Section 4.4 that SSP spectra at ages ≥5 Gyr are nearly indistinguishable, and in Section 4.5 that the first SFE duration is poorly constrained (could be multiple bursts), means the early exponential component sits exactly where the degeneracy is worst. So the >85% multi-episode fraction and the 40–50% mass by z~3 are not established.\n\nThat said, the broad qualitative picture (massive galaxies assemble fast, have fewer episodes) is probably robust, and the paper is honest about limitations. The JWST comparison is contextual and not overhyped.\n\nI would not desk-reject. The data are valuable and the claim is testable; a good referee could ask for injection tests, a light-to-mass conversion check, and model selection. I would not cite the episode statistics in my own work, but I would cite the stellar population trends. Worth a serious referee, though the authors should expect the central claim to be scaled back.","headline":"Solid stellar population data, but the headline episode-count fractions are not established by the current fit.","tokens_in":26973,"tokens_out":2719,"would_cite":false,"duration_ms":28981,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"More than 85% of intermediate-redshift galaxies formed stars in multiple episodes, with the first episode building roughly half the stellar mass by z≈3.","keywords":["star formation history","galaxy evolution","intermediate redshift","MUSE spectroscopy","stellar populations","spectral fitting","star-forming episodes","galaxy environment"],"falsifier":"Build mock galaxy spectra from known two-episode star-formation histories in which both bursts are older than 5 Gyr, run the pPXF+bootstrap pipeline on them with the same masking and smoothing, and compare the recovered SFE count to the input. If the pipeline returns a single exponential episode for most two-burst inputs, the claim that more than 85% of real galaxies have multiple resolved episodes is an artifact of limited age resolution rather than a property of the galaxies.","tokens_in":25991,"feed_emoji":"🌌","tokens_out":7891,"duration_ms":74591,"temperature":0.7,"pith_summary":"The paper derives star formation histories for 393 galaxies at redshifts 0.1–0.9 from VLT/MUSE spectra integrated within one effective radius, then models each history as an exponential decay plus Gaussian peaks. It claims that more than 85% of these galaxies experienced more than one distinct star-forming episode, and that the first episode—detected as a long exponential decay—was typically complete by z≈3 and built about 40–50% of the final stellar mass. It also claims that more massive galaxies have fewer episodes, assemble their mass faster, and reach quiescence earlier than low-mass systems. If correct, galaxy assembly at these redshifts is dominated by an early, dominant event followed by later, lower-mass episodes, with field and group galaxies differing in their stellar-population properties.","feed_headline":"Most galaxies formed stars in repeated bursts","feed_subtitle":"A 393-galaxy MUSE study traces the first burst to z~3, where it builds half of today's stellar mass.","key_machinery":"The 'star-forming episode' (SFE) decomposition. After computing a smoothed SFR as a function of cosmic time from bootstrapped pPXF weight fractions, the paper fits the curve as a sum of an exponential decay (the initial event) and one or more Gaussian peaks (later episodes). Counting the fitted components yields the number of episodes per galaxy, and the amplitudes, durations, and mass fractions of these components carry the mass-assembly and quenching arguments.","core_discovery":"On the paper's own terms: the star formation history of an ordinary intermediate-redshift galaxy is not a single smooth event. Using full-spectrum fitting of E-MILES simple stellar population templates with a bootstrapping approach, the paper reconstructs SFR(t) at 0.5-Gyr resolution for 393 galaxies and decomposes each history into star-forming episodes. It finds that more than 85% of the galaxies have more than one episode; the initial episode is recovered as an exponentially decaying SFR that is longer than later episodes, is complete by z~3, and contributes ~40% of the stellar mass in low-mass galaxies and >50% in high-mass systems. Massive galaxies typically show only one or two episode","pith_inferences":["If the SFE decomposition is accepted, the 'initial exponential event' is still not necessarily a single burst: the paper itself notes that SSP spectra older than ~5 Gyr are nearly indistinguishable, so the exponential may blend several sub-Gyr bursts; true burst complexity could be higher than the counted 1–5 episodes.","A direct calibration test would run the same pipeline on mock spectra with two known bursts at ages >5 Gyr; the rate at which the fit merges them into one exponential would turn the >85% statistic into a completeness-corrected episode fraction.","Applying the decomposition to spatially resolved IFU data beyond 1 Reff could test whether later Gaussian episodes correspond to accreting satellites or tidal debris, tying episode counts directly to merger activity.","The field–group differences, if they persist after careful mass matching, could serve as a probe of pre-processing in galaxy groups, with the η>0.4 t90 difference as a first sign of environmentally triggered recent formation."],"forward_implications":["If more than 85% of galaxies have multiple SFEs, episodic assembly—not a single smooth history—is the normal mode of galaxy growth at 0.1<z<0.9 in the sampled mass range.","The first exponential episode ending by z~3 and contributing 40–50% of the final stellar mass implies that the early Universe sets the dominant mass scale of today's galaxies; later episodes are comparatively minor increments.","Massive galaxies having fewer, earlier episodes supports 'downsizing' at the level of resolved star formation histories: massive systems form first, quench fast, and stay quiescent.","Field and group galaxies differ in stellar mass, metallicity, age, and t90, whereas within-group accretion-stage differences (η) are mostly absent, pointing to environmental effects acting before or at group entry.","The early dominant episode links the sample to massive quiescent galaxies at z>3 found by JWST: many galaxies now observed at z~0.7 may have passed through a quiescent, little-red-dot-like phase at z~3 before later rejuvenation."],"supporting_citations":[{"why":"Supplies the pPXF full-spectrum fitting code and template-weighting method used to derive stellar populations and SFHs.","marker":"Cappellari 2017"},{"why":"Provides the E-MILES SSP template set that defines the age–metallicity grid for the spectral fits.","marker":"Sánchez-Blázquez et al. 2006"},{"why":"Completes the E-MILES library used for fitting; the paper selects templates at MUSE-like resolution.","marker":"Falcón-Barroso et al. 2011"},{"why":"Defines the MUSE-Wide survey whose observations supply most of the sample's field-galaxy spectra.","marker":"Urrutia et al. 2019"},{"why":"Presents the MUSE-Deep (HUDF) observations that provide the deep spectra in the sample.","marker":"Bacon et al. 2017"},{"why":"Provides the MAGIC survey galaxy spectra and the group-environment catalogue with SFRs and stellar masses.","marker":"Epinat et al. 2024"},{"why":"Supplies the 3D-HST photometric catalogue used for effective radii, photometry, and SED-based masses.","marker":"Skelton et al. 2014"},{"why":"Supplies the exponential-plus-Gaussian decomposition approach that the paper adapts to count star-forming episodes.","marker":"Wang et al. 2024"}],"fun_headline_variants":["85% of galaxies had multiple star-forming episodes","Early burst by z=3 built half of galaxies' mass","Massive galaxies form stars in fewer, quicker bursts","Most galaxies' star formation is a series of bursts"],"cache_read_input_tokens":2688,"weakest_assumption_plain":"A single integrated spectrum inside one effective radius, fitted with E-MILES templates and fourth-order polynomials, yields a star-formation history at 0.5-Gyr resolution reliable enough to separate distinct episodes—even though the paper notes that SSP spectra older than about 5 Gyr are nearly indistinguishable, so overlapping early episodes could be counted as one.","fun_headline_variants_meta":{"raw":{"variants":["85% of galaxies had multiple star-forming episodes","Early burst by z=3 built half of galaxies' mass","Massive galaxies form stars in fewer, quicker bursts","Most galaxies' star formation is a series of bursts"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000298,"raw_usage":{"total_tokens":1639,"prompt_tokens":895,"completion_tokens":744,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":639,"completion_tokens_details":{"reasoning_tokens":692}},"tokens_in":639,"tokens_out":744,"duration_ms":9136,"temperature":1.0,"reasoning_tokens":692,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T12:15:39.261405+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Build mock galaxy spectra from known two-episode star-formation histories in which both bursts are older than 5 Gyr, run the pPXF+bootstrap pipeline on them with the same masking and smoothing, and compare the recovered SFE count to the input. If the pipeline returns a single exponential episode for most two-burst inputs, the claim that more than 85% of real galaxies have multiple resolved episodes is an artifact of limited age resolution rather than a property of the galaxies.","supporting_citations":[{"cited_title":"2017, MNRAS, 466,","cited_arxiv_id":null,"evidence_quote":"Supplies the pPXF full-spectrum fitting code and template-weighting method used to derive stellar populations and SFHs."},{"cited_title":"2019, A&A, 624, A141","cited_arxiv_id":null,"evidence_quote":"Defines the MUSE-Wide survey whose observations supply most of the sample's field-galaxy spectra."},{"cited_title":"E., Whitaker, K","cited_arxiv_id":null,"evidence_quote":"Supplies the 3D-HST photometric catalogue used for effective radii, photometry, and SED-based masses."}],"review_version":1}