Inducing the Lovelock action

We re-analyze a possible ambiguity in the application of dimensional regularization to Einstein-Gauss-Bonnet gravity, arising from the way one treats the Gauss-Bonnet term. It is demonstrated that the addition of such a term to the action gives rise to a non-minimal graviton wave operator, but does not produce new on shell divergences at one loop order in d=4. However, from a d-dimensional perspective the Gauss-Bonnet term is shown to generate new divergences in the form of the six-dimensional Euler density. The conjecture that one would next produce the eight-dimensional Euler term is shown to be false.