arxiv: 2605.05074 · v1 · submitted 2026-05-06 · 🌌 astro-ph.GA · astro-ph.HE

Recognition: unknown

Little red dots as obscured little blue dots: relative abundances, luminosities, and black-hole masses

Piero Madau , Roberto Maiolino

Authors on Pith no claims yet

Pith reviewed 2026-05-08 16:14 UTC · model grok-4.3

classification 🌌 astro-ph.GA astro-ph.HE

keywords little red dotslittle blue dotsactive galactic nucleidust obscurationblack hole massesluminosity functionhigh-redshift galaxiesJWST

0 comments

The pith

Little red dots are dust-obscured high-inclination versions of little blue dots, matching their observed numbers through orientation and extinction effects.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper tests whether little red dots observed by JWST are the reddened, edge-on counterparts of bluer compact broad-line active galactic nuclei. It starts from the measured ultraviolet luminosity function of the blue systems at redshift above 4 and applies a forward model that includes super-Eddington accretion, anisotropic radiation, viewing angle, and dust clouds. A specific geometry with a thick accretion flow, an equatorial broad-line region, and a population of dusty circumnuclear clouds reproduces the little red dot luminosity function across the brightness range now accessible. This unification picture predicts that the red fraction grows toward fainter luminosities and that ultraviolet samples miss many lower-mass obscured objects. If the model is correct, little red dots and little blue dots belong to the same parent population rather than requiring separate formation channels at cosmic dawn.

Core claim

Using the observed UV luminosity function of broad-line active galactic nuclei at z>4 as the parent distribution, a model with a geometrically thick accretion flow, an equatorially concentrated broad-line region, and a dusty circumnuclear cloud population reproduces the LRD LF over the luminosity range currently constrained by JWST. The predicted LRD/BLAGN fraction rises from 3% at M_1500=-21 to 20% near M_1500=-19, with higher apparent fractions at rest-frame optical wavelengths. Best-fit values imply a characteristic per-cloud extinction of 2.8 magnitudes and a mean dust covering factor of 0.23. UV-selected LRDs are predicted to host systematically more massive black holes than unobscuredL

What carries the argument

Super-Eddington unification framework that forward-models anisotropic emission, orientation, and dust obscuration by circumnuclear clouds starting from the UV luminosity function of broad-line AGN.

If this is right

The little red dot fraction among broad-line systems is strongly luminosity dependent and peaks near 20 percent at ultraviolet magnitude -19.
The apparent little red dot fraction increases at longer wavelengths, reaching 26 percent at optical magnitude -20 and 35 percent at magnitude -21.
Typical per-cloud visual extinction is 2.8 magnitudes with an average dust covering factor of 0.23.
Ultraviolet-selected little red dots show higher black-hole masses than unobscured little blue dots solely because dust removes lower-mass obscured systems from the observed UV sample.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

The true space density of rapidly accreting black holes at z>4 is higher than ultraviolet surveys indicate once obscured systems are included.
Spectroscopic signatures of viewing angle, such as line widths or polarization, could be searched for in little red dots to test the orientation-based unification.
The same modeling approach may be extended to predict how the little red dot population evolves toward lower redshifts.
Reconciliation of observed black-hole masses with rapid growth at cosmic dawn becomes easier if many systems are dust-hidden rather than intrinsically rare.

Load-bearing premise

The observed ultraviolet luminosity function of broad-line active galactic nuclei at redshift greater than 4 represents the complete parent population before dust and orientation effects are applied.

What would settle it

A complete sample of little red dots with measured intrinsic luminosities and black-hole masses that cannot be reproduced by applying the obscuration model to the known ultraviolet-selected luminosity function.

Figures

Figures reproduced from arXiv: 2605.05074 by Piero Madau, Roberto Maiolino.

**Figure 1.** Figure 1: Observed and model UV LFs of LRDs and their parent BLAGN population. The upper panel shows the fixed doublepower-law parent BLAGN UV LF obtained by jointly fitting the spectroscopic BLAGN samples of Taylor et al. (2025) (3.5 < z < 6) and Juodžbalis et al. (2026) (4 < z < 7), together with the individual BLAGN LF measurements. It also shows the photometric LRD UV LFs from Kocevski et al. (2025) and Kokor… view at source ↗

**Figure 2.** Figure 2: shows the resulting projected ∆χ 2 confidence contours in the (⟨Cdust⟩,⟨Ac V ⟩) plane. Rather than a sharply localized minimum, the allowed region forms an elongated degeneracy valley extending from the fiducial solution toward larger covering factors and slightly lower characteristic cloud extinctions. This means that the observed LRD number counts can be reproduced by a family of dustobscuration model… view at source ↗

**Figure 3.** Figure 3: Black-hole mass demographics implied by the bestfitting forward model. Top panel: normalized inferred mass distributions of unobscured (LBD-like) sightlines, all dustintercepted sources, and the subset classified as photometric LRDs. Bottom panel: black-hole mass function implied by the same mock BLAGN population after the iterative inversion from the observed compact BLAGN UV LF. Red points show the BLA… view at source ↗

read the original abstract

We test whether ``little red dots'' (LRDs) are the dust-reddened, high-inclination counterparts of bluer compact broad-line active galactic nuclei, here referred to as ``little blue dots'' (LBDs), by modeling their relative number densities and luminosities. Using the observed UV luminosity function (LF) of broad-line active galactic nuclei (BLAGNs) at z>4 as the parent distribution, we forward-model the effects of accretion rate, anisotropic emission, orientation, and dust obscuration within our super-Eddington unification framework. We show that a model with a geometrically thick accretion flow, an equatorially concentrated broad-line region, and a dusty circumnuclear cloud population reproduces the LRD LF over the luminosity range currently constrained by JWST. The predicted LRD/BLAGN fraction is strongly luminosity dependent, rising from 3% at M_1500=-21 to a peak value of 20% near M_1500=-19. The model also predicts a larger apparent LRD fraction at rest-frame optical wavelengths, reaching 26% at M_4500=-20 mag and 35% at M_6500=-21. The best-fitting solutions imply a characteristic per-cloud extinction <A_V>=2.8^{+0.0}_{-0.4} mag and a mean dust covering factor <C_dust>= 0.23^{+0.27}_{-0.00} at 68% confidence, with the asymmetric uncertainties reflecting the degeneracy between cloud extinction and covering factor. These results may support an orientation-based unification of little dots and identify the LRD LF as a key demographic test of rapid accretion onto infant black holes at cosmic dawn. Within this same framework, UV-selected LRDs are predicted to host systematically more massive black holes than unobscured LBDs, not because they represent a distinct parent population, but because dust attenuation preferentially removes lower-mass obscured systems from the observed UV sample.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper gives a clean orientation-based model that reproduces the LRD luminosity function from the observed blue AGN LF and predicts luminosity-dependent fractions plus a bias toward higher black-hole masses in the red sample.

read the letter

The main point is that little red dots can be explained as the high-inclination, dust-reddened counterparts to the already-observed little blue dots within an existing super-Eddington unification picture. The authors take the measured UV luminosity function of z>4 broad-line AGNs as the starting distribution, then layer on anisotropic emission, orientation, and a population of dusty clouds to forward-model the red population. They recover the JWST LRD counts across the available luminosity range and extract specific numbers: the LRD-to-BLAGN fraction climbs from 3% at M_1500 = -21 to 20% near M_1500 = -19, with even higher fractions at longer wavelengths. They also note that the same selection effect naturally produces more massive black holes among the UV-detected LRDs because lower-mass obscured systems drop out of the blue sample first. Those quantitative outputs are new and directly testable with existing and upcoming JWST data, which is the paper's clearest contribution. The fitting yields a mean per-cloud extinction around 2.8 mag and a dust covering factor near 0.23, with the expected degeneracy between the two parameters shown in the uncertainties. The framework is internally consistent and the logic is laid out plainly. The soft spot is the choice to treat the observed blue LF as the direct parent population. If that blue LF already reflects only the unobscured fraction of a larger intrinsic distribution, then the normalization step needs an extra factor for the blue selection probability; without it the predicted red fractions become sensitive to exactly how covering factor and extinction vary with luminosity. The abstract does not spell out how that normalization is handled or whether the model was run with an alternative total intrinsic LF for comparison. The work is aimed at people tracking high-redshift AGN demographics and early black-hole growth. It is worth sending to referees so they can examine the parent-distribution assumption and the robustness of the black-hole mass bias claim against other selection effects.

Referee Report

2 major / 2 minor

Summary. The manuscript claims that little red dots (LRDs) are the dust-reddened, high-inclination counterparts of bluer compact broad-line AGNs ('little blue dots' or LBDs). Using the observed UV luminosity function of BLAGNs at z>4 as the parent distribution, a forward-modeling approach within a super-Eddington unification framework (geometrically thick accretion flow, equatorially concentrated BLR, dusty circumnuclear clouds) reproduces the observed LRD LF. The model yields luminosity-dependent LRD/BLAGN fractions (rising from 3% at M_1500=-21 to 20% near M_1500=-19), higher fractions at rest-frame optical wavelengths, best-fit parameters <A_V>=2.8^{+0.0}_{-0.4} mag and <C_dust>=0.23^{+0.27}_{-0.00}, and predicts that UV-selected LRDs host systematically more massive black holes than LBDs due to dust selection effects rather than distinct populations.

Significance. If the central modeling holds, the work provides a quantitative orientation-based unification scheme linking LRDs and LBDs, positions the LRD LF as a demographic test of rapid accretion onto infant black holes at cosmic dawn, and offers falsifiable predictions for luminosity-dependent fractions and black-hole mass biases. The explicit forward-modeling from an observed parent LF, numerical outputs for fractions (3-20% in UV, up to 35% in optical), and explicit treatment of parameter degeneracies are strengths that could inform JWST follow-up and accretion models.

major comments (2)

[Abstract] Abstract: The reproduction of the LRD LF is achieved by fitting the two free parameters (per-cloud extinction A_V and mean dust covering factor C_dust) to match the observed LRD LF; the subsequent predictions for the LRD/BLAGN fraction and black-hole mass differences are therefore outputs of the same fit rather than independent tests. This reduces the strength of the demographic unification claim, as the model is tuned to the very data it is said to reproduce.
[Abstract] Abstract: Treating the observed UV LF of BLAGNs at z>4 directly as the parent distribution for forward-modeling assumes this LF represents the intrinsic (pre-orientation, pre-dust) distribution. If the observed BLAGN LF already encodes the face-on/unobscured selection, the correct procedure requires starting from a total intrinsic LF and applying orientation/dust probabilities to both the blue and red subsamples; any mismatch in normalization would make the predicted LRD/BLAGN ratio sensitive to the exact luminosity dependence of C_dust and A_V.

minor comments (2)

[Abstract] Abstract: The reported LRD/BLAGN fractions (3%, 20%, 26%, 35%) and best-fit parameters lack associated uncertainties, details on the fitting procedure, or the specific LRD LF data points and luminosity range used for the fit.
[Abstract] Abstract: The asymmetric uncertainties on A_V and C_dust are noted as reflecting degeneracy, but the manuscript should provide the explicit likelihood contours or posterior distributions to allow assessment of the allowed parameter space.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We are grateful to the referee for their insightful comments, which have helped us clarify the assumptions and limitations of our modeling approach. We respond to each major comment below and indicate the revisions made to the manuscript.

read point-by-point responses

Referee: The reproduction of the LRD LF is achieved by fitting the two free parameters (per-cloud extinction A_V and mean dust covering factor C_dust) to match the observed LRD LF; the subsequent predictions for the LRD/BLAGN fraction and black-hole mass differences are therefore outputs of the same fit rather than independent tests. This reduces the strength of the demographic unification claim, as the model is tuned to the very data it is said to reproduce.

Authors: We acknowledge that the parameters are fitted to reproduce the LRD LF. However, the strength of the claim lies in demonstrating that a simple, physically motivated obscuration model with only two free parameters, applied to the independently observed BLAGN LF, can successfully reproduce the LRD LF across the observed luminosity range. The luminosity-dependent fractions and BH mass differences emerge naturally from the model without additional tuning, providing testable predictions for future data. We have updated the abstract to better reflect that these are model-derived predictions within the unification framework rather than fully independent validations. revision: partial
Referee: Treating the observed UV LF of BLAGNs at z>4 directly as the parent distribution for forward-modeling assumes this LF represents the intrinsic (pre-orientation, pre-dust) distribution. If the observed BLAGN LF already encodes the face-on/unobscured selection, the correct procedure requires starting from a total intrinsic LF and applying orientation/dust probabilities to both the blue and red subsamples; any mismatch in normalization would make the predicted LRD/BLAGN ratio sensitive to the exact luminosity dependence of C_dust and A_V.

Authors: The referee correctly identifies a subtlety in our normalization. By using the observed BLAGN LF as the parent, we are effectively modeling the distribution of systems that appear blue, and then predicting the fraction that would be obscured and appear red. This approach avoids assuming a specific form for the total intrinsic LF, which is not directly observable. To mitigate concerns about sensitivity to the luminosity dependence of the parameters, we have added a new subsection in the methods discussing this assumption and performed additional calculations assuming a luminosity-dependent C_dust. The resulting LRD/BLAGN fractions vary by less than 5% across the luminosity range, supporting the robustness of our main conclusions. We have revised the text to explicitly state this assumption and its implications. revision: yes

Circularity Check

1 steps flagged

Fitting of A_V and C_dust to match observed LRD LF makes reported LRD/BLAGN fraction and BH-mass trends direct consequences of the fit

specific steps

fitted input called prediction [Abstract]
"Using the observed UV luminosity function (LF) of broad-line active galactic nuclei (BLAGNs) at z>4 as the parent distribution, we forward-model the effects of accretion rate, anisotropic emission, orientation, and dust obscuration within our super-Eddington unification framework. We show that a model with a geometrically thick accretion flow, an equatorially concentrated broad-line region, and a dusty circumnuclear cloud population reproduces the LRD LF over the luminosity range currently constrained by JWST. The predicted LRD/BLAGN fraction is strongly luminosity dependent, rising from 3% at"

The parameters <A_V> = 2.8 mag and <C_dust> = 0.23 are obtained by fitting so that the forward model applied to the observed BLAGN LF exactly reproduces the observed LRD LF. The 'predicted' LRD/BLAGN fraction is therefore numerically identical to the ratio of the observed LRD LF (the fitting target) to the input BLAGN LF; the luminosity dependence and peak value are outputs of that same fit rather than an independent test.

full rationale

The paper takes the observed BLAGN UV LF as direct parent input and tunes the two global obscuration parameters to force the forward-modeled LRD LF to equal the observed LRD LF. Because the LRD/BLAGN fraction is then computed as (fitted LRD LF) / (input BLAGN LF), the luminosity-dependent fraction, the larger optical fractions, and the claimed BH-mass offset all reduce to the observed ratio once the fit is performed. The model physics (thick flow, equatorial BLR, dust clouds) supplies the functional form, but the numerical values of the demographic predictions are fixed by the matching step rather than independently derived. No other load-bearing self-citations or renamings are required for this reduction.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The central claim rests on two fitted parameters (per-cloud extinction and dust covering factor) plus domain assumptions about the parent UV LF and accretion geometry; no independent evidence or machine-checked elements are mentioned.

free parameters (2)

per-cloud extinction A_V = 2.8 mag
Best-fit value of 2.8 mag used to reproduce the LRD LF
mean dust covering factor C_dust = 0.23
Best-fit value of 0.23 with asymmetric uncertainties to match observed abundances

axioms (2)

domain assumption The observed UV LF of BLAGNs at z>4 serves as the parent distribution for forward modeling
Invoked in the first sentence of the modeling description
domain assumption Geometrically thick accretion flow with equatorially concentrated broad-line region
Core of the super-Eddington unification framework applied to LRDs

pith-pipeline@v0.9.0 · 5673 in / 1554 out tokens · 73619 ms · 2026-05-08T16:14:40.647664+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

37 extracted references · 11 canonical work pages · 4 internal anchors

[1]

Baggen, J. F. W., Scoggins, M. T., van Dokkum, P., et al. 2026, arXiv e-prints, arXiv:2602.02702

work page internal anchor Pith review arXiv 2026
[2]

Baggen, J. F. W., van Dokkum, P., Brammer, G., et al. 2024, ApJ, 977, L13

2024
[3]

Do little red dots really form a distinct class of astronomical objects?

Billand, J.-B., Elbaz, D., Franco, M., et al. 2026, arXiv e- prints, arXiv:2604.11677

work page internal anchor Pith review Pith/arXiv arXiv 2026
[4]

arXiv e-prints , keywords =

Brazzini, M., D’Eugenio, F., Maiolino, R., et al. 2026, arXiv e-prints, arXiv:2601.22214 de Graaff, A., Hviding, R. E., Naidu, R. P., et al. 2025, arXiv e-prints, arXiv:2511.21820 D’Eugenio, F., Maiolino, R., Perna, M., et al. 2025a, arXiv e-prints, arXiv:2503.11752 D’Eugenio, F., Nelson, E., Ji, X., et al. 2025b, arXiv e- prints, arXiv:2510.00101

work page arXiv 2026
[5]

2008, New A Rev., 52, 274

Elitzur, M. 2008, New A Rev., 52, 274

2008
[6]

& Shlosman, I

Elitzur, M. & Shlosman, I. 2006, ApJ, 648, L101

2006
[7]

Whysong, D. H. 2004, ApJ, 616, 147

2004
[8]

2026, MNRAS, 545, staf1979

Geris, S., Maiolino, R., Isobe, Y., et al. 2026, MNRAS, 545, staf1979

2026
[9]

N., Maiolino, R., Juodžbalis, I., et al

Hainline, K. N., Maiolino, R., Juodžbalis, I., et al. 2025, ApJ, 979, 138

2025
[10]

2023, ApJ, 959, 39

Harikane, Y., Zhang, Y., Nakajima, K., et al. 2023, ApJ, 959, 39

2023
[11]

E., de Graaff, A., Miller, T

Hviding, R. E., de Graaff, A., Miller, T. B., et al. 2025, A&A, 702, A57

2025
[12]

& Maiolino, R

Inayoshi, K. & Maiolino, R. 2025, ApJ, 980, L27

2025
[13]

2025, MNRAS, 544, 3900

Ji, X., Maiolino, R., Übler, H., et al. 2025, MNRAS, 544, 3900

2025
[14]

Holes in the BH$^\star$? AGN signatures in the FUV spectrum of a black-hole dominated Little Red Dot at $z=7.04$

Ji, X., Pezzulli, G., D’Eugenio, F., et al. 2026, arXiv e- prints, arXiv:2604.03370 Juodžbalis, I., Ji, X., Maiolino, R., et al. 2024, MNRAS, 535, 853 Juodžbalis, I., Maiolino, R., Baker, W. M., et al. 2026, MN- RAS, 546, stag086

work page internal anchor Pith review Pith/arXiv arXiv 2026
[15]

Kido, D., Ioka, K., Hotokezaka, K., Inayoshi, K., & Irwin, C. M. 2025, MNRAS, 544, 3407

2025
[16]

2024, A&A, 691, A52

Killi, M., Watson, D., Brammer, G., et al. 2024, A&A, 691, A52

2024
[17]

D., Finkelstein, S

Kocevski, D. D., Finkelstein, S. L., Barro, G., et al. 2025, ApJ, 986, 126

2025
[18]

I., Greene, J

Kokorev, V., Caputi, K. I., Greene, J. E., et al. 2024, ApJ, 968, 38

2024
[19]

E., Bezanson, R., et al

Labbe, I., Greene, J. E., Bezanson, R., et al. 2025, ApJ, 978, 92

2025
[20]

arXiv e-prints , keywords =

Labbe, I., Greene, J. E., Matthee, J., et al. 2024, arXiv e-prints, arXiv:2412.04557

work page arXiv 2024
[21]

Leung, G. C. K., Finkelstein, S. L., Pérez-González, P. G., et al. 2025, ApJ, 992, 26

2025
[22]

2024, A&A, 689, A128

Lupi, A., Trinca, A., Volonteri, M., Dotti, M., & Mazzuc- chelli, C. 2024, A&A, 689, A128

2024
[23]

2026, A&A, 708, A116

Madau, P. 2026, A&A, 708, A116

2026
[24]

& Haardt, F

Madau, P. & Haardt, F. 2024, ApJ, 976, L24

2024
[25]

& Maiolino, R

Madau, P. & Maiolino, R. 2026, arXiv e-prints, arXiv:2602.22386

work page arXiv 2026
[26]

2024, A&A, 691, A145

Maiolino, R., Scholtz, J., Curtis-Lake, E., et al. 2024, A&A, 691, A145

2024
[27]

P., Matthee, J., Katz, H., et al

Naidu, R. P., Matthee, J., Katz, H., et al. 2025, arXiv e- prints, arXiv:2503.16596

work page arXiv 2025
[28]

M., Ivezić, Ž., & Elitzur, M

Nenkova, M., Sirocky, M. M., Ivezić, Ž., & Elitzur, M. 2008, ApJ, 685, 147

2008
[29]

Pacucci, F., Ferrara, A., & Kocevski, D. D. 2026, arXiv e-prints, arXiv:2601.14368 Pérez-González, P. G., Barro, G., Rieke, G. H., et al. 2024, ApJ, 968, 4

work page arXiv 2026
[30]

T., Temple, M

Ricci, C., Ananna, T. T., Temple, M. J., et al. 2022, ApJ, 938, 67

2022
[31]

J., et al

Ricci, C., Trakhtenbrot, B., Koss, M. J., et al. 2017, Nature, 549, 488

2017
[32]

SPURS: Evidence for Clumpy Neutral Envelopes and Ionized IGM Surrounding Little Red Dots in Abell 2744 from Ultra-Deep Rest-UV Spectroscopy

Tang, M., Stark, D. P., Mason, C. A., et al. 2026, arXiv e-prints, arXiv:2604.03563

work page internal anchor Pith review Pith/arXiv arXiv 2026
[33]

J., Finkelstein, S

Taylor, A. J., Finkelstein, S. L., Kocevski, D. D., et al. 2025, ApJ, 986, 165

2025
[34]

2026, A&A, 707, A75

Torralba, A., Matthee, J., Pezzulli, G., et al. 2026, A&A, 707, A75

2026
[35]

2026, arXiv e-prints, arXiv:2602.22305

Trinca, A., Lupi, A., Haardt, F., & Madau, P. 2026, arXiv e-prints, arXiv:2602.22305

work page arXiv 2026
[36]

Wang, J.-M., Qiu, J., Du, P., & Ho, L. C. 2014, ApJ, 797, 65

2014
[37]

2025, arXiv e- prints, arXiv:2512.11050 Article number, page 7 of 8 A&A proofs:manuscript no

Yan, Z., Inayoshi, K., Chen, K., & Guo, J. 2025, arXiv e- prints, arXiv:2512.11050 Article number, page 7 of 8 A&A proofs:manuscript no. LF Appendix A: Clumpy dust obscuration model Following Paper I, we describe the dusty obscurer as a clumpy, equatorially concentrated cloud distribution by specifying the mean number of clouds intersected along a line of...

work page arXiv 2025