Starobinsky 1980, the foundational paper. An R-squared term drives early-universe inflation.
Starobinsky R-squared Inflation
The original higher-curvature gravity proposal: an R-squared term in the action drives a phase of cosmic inflation in the very early universe. Starobinsky 1980 is the foundational work that originated the higher-curvature gravity research program.
Looping ambient scene for Geometric Modified Gravity. Geometric modified gravity replaces the Einstein-Hilbert Lagrangian, linear in the Ricci scalar R, with an arbitrary function f(R) or a related higher-curvature extension. The framework was developed in inflationary cosmology (Starobinsky 1980 R-squared inflation) and revived for late-time cosmic acceleration in the mid-2000s (Hu-Sawicki 2007, Starobinsky 2007). The unifying claim is that observed dark energy may be a manifestation of modified gravitational geometry rather than a separate energy component. Modern f(R) gravity comes with sharp constraints: solar-system tests, structure-growth measurements, and the chameleon mechanism for screening. The framework is distinct from MOND because the mechanism is geometric (modifying the curvature-action relation) rather than a phenomenological acceleration scale.
§1 · The claim, in one sentence
Alexei Starobinsky proposed in 1980, in a paper published before the arXiv existed, that an R-squared term added to the Einstein-Hilbert action drives a phase of cosmic inflation in the very early universe. The 1980 paper is the foundational work for the entire higher-curvature gravity research program and predates the modern late-time f(R) constructions (sibling variants) by 25 years.
§2 · Why it might be true
Starobinsky's 1980 framework was developed in the context of the early universe, not late-time cosmology. Adding an R-squared term to the gravitational action introduces a higher-derivative dynamics whose perturbative behavior matches the slow-roll inflation paradigm. The constant of proportionality of the R-squared term determines the inflationary phenomenology; current Planck data constrain this constant tightly.
The construction is connected to the late-time f(R) work in the sibling variants: both modify the gravitational action with higher-curvature terms, but with different parameterizations and at different cosmological epochs. The 1980 paper is the conceptual ancestor; the modern work extends and refines the framework to specific cosmological problems.
Starobinsky inflation predicts a specific tensor-to-scalar ratio that has been a benchmark for inflation searches. Current Planck and BICEP-Array data constrain the model tightly; the framework remains viable but the parameter space is restricted. It is one of the most precise inflationary models currently in the running.
The family stance
The accelerating expansion of the universe and the structure of strong-gravity regimes may be explained by replacing Einstein's R Lagrangian with a more general function f(R). The standard dark-energy interpretation (a true [[cosmological constant]] or scalar dark-energy field) becomes unnecessary if the gravitational action itself is the source of the acceleration.
§2.5 · Evidence
- Current Planck 2018 data (with 2020 follow-up analyses) is consistent with Starobinsky inflation in the small-r region of parameter space; the model is among the most successful inflation candidates
- The 1980 paper has been cited extensively (hundreds of citations) and is the canonical reference for higher-curvature inflation
- The framework connects naturally to the modern late-time f(R) work via the higher-curvature gravity research program (sibling variants in this family)
- Independent supergravity completions (no-scale models, alpha-attractor inflation) often reduce to Starobinsky-style attractors in their inflationary regimes; the framework has multiple UV-physics motivations
§3 · What you'd need to test it
- An R-squared term in the gravitational action produces a phase of slow-roll inflation in the very early universe with specific perturbation spectra
- The model predicts a tensor-to-scalar ratio of order 0.003 to 0.004, in tension with some other inflation models that predict larger values
- The predicted shape of primordial density ripples (the scalar spectral index n_s, a measure of how the strength of those ripples changes with scale, and its running, how that slope itself drifts) matches what the Planck satellite's map of the cosmic microwave background sees, under appropriate parameter ranges
- The framework should be testable by future CMB B-mode polarization measurements (LiteBIRD, CMB-S4, Simons Observatory) that probe the tensor-to-scalar ratio
§4 · Where it breaks
- Current data constrains the Starobinsky parameter range tightly; the model is testable rather than free, and next-generation B-mode measurements will either confirm or restrict it sharply
- The R-squared model requires a specific scale for the higher-curvature term that does not have a deep first-principles explanation
- The framework is one of several viable inflation candidates; selecting Starobinsky over other models awaits sharper observational constraints
- The 1980 paper predates the modern formulation of f(R) gravity; some technical subtleties of the inflationary slow-roll evolution were worked out in subsequent literature rather than in the original paper
Go deeper
The R-squared term in the gravitational action has dimension four (in 4D spacetime), so it is renormalizable in the standard power-counting sense. This was originally noted by Stelle in the context of curvature-squared gravity in 1977 (cross-reference: Ch.4 Asymptotic Safety family, Higher-Derivative Gravity Extensions variant). Starobinsky's 1980 paper applied this insight specifically to inflation.
The Starobinsky inflation framework has been extensively connected to UV physics: no-scale supergravity reduces to Starobinsky-style attractors; alpha-attractor inflation models also flow to Starobinsky-like behavior in their slow-roll regimes. The framework has multiple motivated UV completions, which is unusual for inflation models.
▸§5 · Who built it, and when(1 source, 1 established)
Variants in this family
Compare variantsUp next
Spotted an error? Have a source to add?
Prefer email?
You can also send a prefilled email with the variant URL already filled in.