REVIEW
All Quiet on the Vector Front: Scalar Growth and Perturbative Control in Horndeski--Weyl-squared Gravity
General Relativity and Quantum Cosmology · Submitted 14 Oct 2026 · 75 references
Pith's verdict: Credible generalization of the Starobinsky Weyl-ghost scalar-growth result to Horndeski backgrounds, with the physical interpretation honestly gated on the unproven M<Λ_EFT hierarchy.
T0 · AI review
Why Pith says this
Highlight any text to comment on it.
Referee report
The paper
The paper claims
Adding a Weyl-squared term to Horndeski gravity makes at least one Bardeen potential grow on superhorizon scales for every expanding background studied, a growth that cannot be removed by a coordinate choice.
The reading
What carries the argument
Two objects carry the argument. The first is the Weyl-induced mass scale $M^2=G_H/(4\alpha)$ read off from the vector-sector action: positivity of $M^2$ — equivalently $\alpha$ and $G_H$ sharing a sign — excludes tachyonic vector modes and controls the decay rates of the massive scalar, vector, and tensor branches. The second is the finite-$k$ two-field reduction: once the non-dynamical lapse $\psi$ is eliminated through eq. (7.4) (assuming $4\alpha k^4\neq 3a^4\Sigma$ and $G_H\Sigma+3\Theta^2\neq 0$), the scalar dynamics is carried by the curvature perturbation $\zeta$ and the normalized shear $s=(k^2/a^2)\chi$, whose leading-order long-wavelength system closes and whose kinetic coefficients stay regular as $k\to 0$. The pivot is the reconstruction $\Phi=-\zeta-(a^2/k^2)Hs$, $\Psi=\psi+(a^2/k^2)(\dot{s}+2Hs)$: the inverse-gradient factor $a^2/k^2$ converts the decaying massive branches of $s$ into growing branches of the Bardeen potentials, so the same linear solution is bounded in one representation and growing in the other.
Extended reading notes
Core claim
The paper's central claim is that the classical linear scalar sector of Horndeski–Weyl-squared gravity is generically unstable on superhorizon scales: on every expanding background considered — shift-symmetric de Sitter and power-law FLRW with $p>0$ — the unitary-gauge shear potential $\chi$ contains a growing branch, and at least one of the two gauge-invariant Bardeen potentials $\Phi$ and $\Psi$ inherits that growth. In de Sitter the dynamical pair $\zeta$ and $s=(k^2/a^2)\chi$ obeys a characteristic equation whose nonzero roots all have negative real parts for $M^2>0$, so the physical fields themselves decay; the growth appears only in the reconstruction $\Phi=-\zeta-(a^2/k^2)Hs$ and $\Psi=\psi+(a^2/k^2)(\dot{s}+2Hs)$, where the inverse-gradient factor $a^2/k^2$ turns the decaying massive shear branches into growing potentials. Because that factor reflects the fixed metric normalization of the Bardeen potentials and cannot be rescaled away, the growth is a genuine first-order property of gauge-invariant quantities rather than a coordinate artifact; the authors confirm it by solving the unitary-gauge equations, by solving the closed fourth-order equations for $\Phi$ and $\Psi$ directly, and numerically. The same mechanism leaves $\zeta$ bounded, cancels the direct gradient-free shear contribution to the clock-matched tidal field, and makes the electric part of the Weyl tensor decay, so the paper distinguishes growth of the potentials from growth of local observables and concludes that the standard Newtonian-gauge linear expansion has a limited domain of validity.
Load-bearing premise
The load-bearing premise is that the Weyl-squared term can be treated as a genuine dynamical degree of freedom inside a valid effective field theory, with the Weyl-induced mass $M$ and all relevant momenta and curvatures below the cutoff $\Lambda_{\rm EFT}$ — if $M$ sits at or above the cutoff the pole should be integrated out and the growing Bardeen-potential branch would not belong to the low-energy theory, a restriction the paper itself states in Section II and repeats in Section VIII — together with the secondary technical conditions that the lapse elimination is nondegenerate ($4\alpha k^4\neq 3a^4\Sigma$) and that the two-field reduction requires $G_H\Sigma+3\Theta^2\neq 0$.
Editorial extensions
If this is right
- The growth is generic across the studied classes: in shift-symmetric de Sitter both Bardeen potentials grow, while in power-law backgrounds $\Psi$ grows for every $p>0$ and $\Phi$ is bounded only for $1/3<p\leq 2$; in the inflationary window $1<p\leq 2$, boundedness of $\Phi$ cannot be combined with ghost-free subluminal scalar propagation, which requires $p>3$.
- Perturbative control in Newtonian gauge is finite: the weak-metric condition $|\Phi|,|\Psi|\ll 1$ fails after $\Delta N<\sigma^{-1}\ln(\varepsilon_{\rm met}/A_*)$ e-foldings once the growing mode is excited.
- The growth is not a coordinate artifact: it appears in both Newtonian and fully fixed unitary gauges, and it matches the growing branch of the dimensionless gauge-invariant combination $H\chi^{\rm GI}_{\rm PZK}=\mathcal{R}-\Phi$ in the comparison with Ref. [49].
- Growth of the potentials is distinct from growth of local observables: the direct gradient-free shear contribution to the clock-matched tidal field cancels, the electric part of the Weyl tensor decays, and the leading shear contributions to $R$, $K$, and $C^2$ decay at fixed small $k$.
- The curvature perturbation $\zeta$ (and hence the uniform-field curvature $\mathcal{R}=-\zeta$) stays bounded on superhorizon scales for $M^2>0$ and $p>1/3$, a necessary condition for a viable primordial spectrum, while the identification of which scalar combination is observed is left open.
Reading between the lines
- If the inverse-gradient mechanism is generic, the same superhorizon growth should appear in any inflationary effective field theory that retains a light Weyl-induced spin-2 pole, not only in the two Horndeski subclasses studied here; a direct test would be repeating the two-field reduction for alternative quadratic-curvature formulations such as Palatini or Gauss–Bonnet-extended versions.
- A matching calculation between a concrete UV completion and the truncated action could decide whether the growing branch is physically relevant at all: if the Weyl pole mass $M$ sits at or above the EFT cutoff, the pole should be integrated out and the growth would vanish from the low-energy theory, making the paper's conclusion conditional on the hierarchy $M<\Lambda_{\rm EFT}$ that no current ca
- The requirement that subluminal, ghost-free, gradient-stable accelerating power-law solutions need $p>3$, combined with $\Psi$ growing for all $p>0$, suggests that power-law inflation is doubly constrained; computing the curvature power spectrum and non-Gaussianity in the surviving region would reveal whether the second scalar mode leaves an observable imprint.
- A second-order calculation of the induced gravitational-wave strain, sourced by gradients of the growing $\Psi$, is a natural arbiter of whether the weak-metric breakdown is a physical instability or only a failure of the Newtonian-gauge expansion; the paper identifies this mechanism but does not perform the calculation.
Original abstract
Higher-order curvature corrections to general relativity arise naturally in many fundamental theories and provide phenomenologically interesting low-energy effective field theories. However, higher-derivative terms can introduce ghost modes, which call the validity of such models into question. Here, we thoroughly investigate the classical linear stability of the most general second-order scalar-tensor theory, namely the Horndeski theory of gravity, augmented by a Weyl-squared term in the action. We consider shift-symmetric de Sitter and power-law FLRW backgrounds and study the evolution of linear perturbations. Vector perturbations are free of gradient and tachyonic instabilities for a positive Weyl-induced mass squared. Tensor perturbations necessarily contain a ghost; their superhorizon branches are bounded, while a frozen-background calculation identifies a separate finite-momentum stability condition. The scalar sector instead contains a growing long-wavelength branch for all expanding backgrounds we consider, which is inherited by at least one of the gauge-invariant Bardeen potentials. We show that this growth enters the Bardeen potentials through an inverse-gradient factor $a^2/k^2$: the dynamical scalar fields remain bounded on the inflationary backgrounds, the direct contribution of the growing shear to the clock-matched tidal field cancels, and the electric part of the Weyl tensor decays. We discuss the implications of our results for perturbative control and for the phenomenological viability of theories including quadratic curvature corrections.
The compiled record
The paper's own sentences, arranged, kept beside the review and separate from it. Each label is drawn from a closed vocabulary; the number beside it is the probability the classifier gave that label. Numbers in the tables are the authors', checked by code. No sentence in this record was written by a model.
The abstract, sentence by sentence
Higher-order curvature corrections to general relativity arise naturally in many fundamental theories and provide phenomenologically interesting low-energy effective field theories. However, higher-derivative terms can introduce ghost modes, which call the validity of such models into question. Here, we thoroughly investigate the classical linear stability of the most general second-order scalar-tensor theory, namely the Horndeski theory of gravity, augmented by a Weyl-squared term in the action. We consider shift-symmetric de Sitter and power-law FLRW backgrounds and study the evolution of linear perturbations. Vector perturbations are free of gradient and tachyonic instabilities for a positive Weyl-induced mass squared. Tensor perturbations necessarily contain a ghost; their superhorizon branches are bounded, while a frozen-background calculation identifies a separate finite-momentum stability condition. The scalar sector instead contains a growing long-wavelength branch for all expanding backgrounds we consider, which is inherited by at least one of the gauge-invariant Bardeen potentials. We show that this growth enters the Bardeen potentials through an inverse-gradient factor $a^2/k^2$: the dynamical scalar fields remain bounded on the inflationary backgrounds, the direct contribution of the growing shear to the clock-matched tidal field cancels, and the electric part of the Weyl tensor decays. We discuss the implications of our results for perturbative control and for the phenomenological viability of theories including quadratic curvature corrections.
background contribution method result interpretation main claim
| # | sentence | role | the authors' own work | states the main conclusion | strength |
|---|---|---|---|---|---|
| s0 | Higher-order curvature corrections to general relativity arise naturally in many fundamental theories and provide phenomenologically interesting low-energy effective field theories. | background 1.00 | 0.04 | 0.02 | asserted 0.97 |
| s1 | However, higher-derivative terms can introduce ghost modes, which call the validity of such models into question. | background 0.98 | 0.04 | 0.02 | hedged possibility 0.63 |
| s2 | Here, we thoroughly investigate the classical linear stability of the most general second-order scalar-tensor theory, namely the Horndeski theory of gravity, augmented by a Weyl-squared term in the action. | contribution 0.99 | 0.94 | 0.05 | asserted 0.93 |
| s3 | We consider shift-symmetric de Sitter and power-law FLRW backgrounds and study the evolution of linear perturbations. | method 0.81 | 0.93 | 0.03 | asserted 0.72 |
| s4 | Vector perturbations are free of gradient and tachyonic instabilities for a positive Weyl-induced mass squared. | result 1.00 | 0.88 | 0.12 | asserted 1.00 |
| s5 | Tensor perturbations necessarily contain a ghost; their superhorizon branches are bounded, while a frozen-background calculation identifies a separate finite-momentum stability condition. | result 1.00 | 0.92 | 0.19 | asserted 1.00 |
| s6 | The scalar sector instead contains a growing long-wavelength branch for all expanding backgrounds we consider, which is inherited by at least one of the gauge-invariant Bardeen potentials. | result 1.00 | 0.92 | 0.33 | asserted 1.00 |
| s7 | We show that this growth enters the Bardeen potentials through an inverse-gradient factor $a^2/k^2$: the dynamical scalar fields remain bounded on the inflationary backgrounds, the direct contribution of the growing shear to the clock-matched tidal field cancels, and the electric part of the Weyl tensor decays. | result 0.84 | 0.94 | 0.29 | asserted 1.00 |
| s8 | We discuss the implications of our results for perturbative control and for the phenomenological viability of theories including quadratic curvature corrections. | interpretation 0.98 | 0.58 | 0.06 | unclear 0.80 |
What was examined, and what was not
- arXiv listing abstract v1
- full text
- figures and tables
- references and citing sentences
- relations to other papers
- P abstract
- S sections
- T body sentences
- C citations
- R cross-paper relations
Figures
Figures reproduced from arXiv:2610.20156 by the authors.
References
Discussion
Disagree with the review, or know something it missed? Leave a comment, highlight any passage above to comment on it, or challenge this review formally.