One loop in $D=11$ vs $D=10$: a mismatch

Abstract: The one-loop correction to eleven-dimensional supergravity involves a term $t_8t_8R^4$, with four Riemann tensors. With a suitable UV regularization, this term is expected to reduce to a similar one-loop term present in the type IIA effective action. We show that, while this is indeed the case in the NSNS sector, it fails when one includes the RR fields. Specifically, we identify 4-point couplings involving the dilaton, the graviton and two RR one-forms which lead to amplitudes that disagree with one-loop string amplitudes. We conclude that the eleven-dimensional $R^4$ superinvariant must split into more than one upon dimensional reduction. We give a complete analysis of 4-point terms in eleven dimensions computed from superparticle amplitudes, and show that, again, extra terms involving RR fields arise from these. We also present a simplified form for the 4-point couplings in eleven dimensions.