We consider the gravitational Wilsonian effective action at low energy when all the particles of the standard model have decoupled. When the ${\cal R}^2$ terms dominate, the theory is equivalent to a scalar-tensor theory with the universal coupling $\beta=1/\sqrt 6$ to matter for which we present strong lower and upper bounds on the scalaron mass $m$ obtained by using results from the E\"ot-Wash experiment on the modification of the inverse-square law, the observations of the hot gas of galaxy clusters and the Planck satellite data on the neutrino masses. In terms of the range of the scalar interaction mediated over a distance of order $m^{-1}$, this leads to the small interval $4\,\mu m \lesssim m^{-1} \lesssim 68\, \mu m$ within reach of future experimental tests of deviations from Newton's gravitational inverse-square law.

We analyse the role, on large cosmological scales and laboratory experiments, of the leading curvature squared contributions to the low energy effective action of gravity. We argue for a natural relationship $c_0\lambda^2\simeq 1$ at low-energy between the ${\cal R}^2$ coefficients $c_0$ of the Ricci scalar squared term in this expansion and the dark energy scale $\Lambda=(\lambda M_{\rm Pl})^4$ in four dimensional Planck mass units. We show how the compatibility between the acceleration of the expansion rate of the Universe, local tests of gravity and the quantum stability of the model all converge to select such a relationship up to a coefficient which should be determined experimentally. When embedding this low energy theory of gravity into candidates for its ultraviolet completion, we find that the proposed relationship is guaranteed in string-inspired supergravity models with modulus stabilisation and supersymmetry breaking leading to de Sitter compactifications. In this case, the scalar degree of freedom of ${\cal R}^2$ gravity is associated to a volume modulus. Once written in terms of a scalar-tensor theory, the effective theory corresponds to a massive scalar field coupled with the universal strength $\beta=1/\sqrt{6}$ to the matter stress-energy tensor. When the relationship $c_0\lambda^2\simeq 1$ is realised we find that on astrophysical scales and in cosmology the scalar field is ultralocal and therefore no effect arises on such large scales. On the other hand, the scalar field mass is tightly constrained by the non-observation of fifth forces in torsion pendulum experiments such as E\"ot-Wash. It turns out that the observation of the dark energy scale in cosmology implies that the scalar field could be detectable by fifth force experiments in the near future.