Naturalness arguments and dimensional regularization? How do issues of naturalness arise when regularizing QFT using dimensional regularization?  I can only recall ever seeing naturalness arguments (hierarchy problem, cosmological constant problem, etc.) phrased in terms regularizing with a cutoff, where naturalness issues arise when physical quantities are quadratically divergent in the cutoff scale.
Is it hard to see how the same naturalness issues are addressed using dimensional regularization? Are there some hidden assumptions involved in using dimensional regularization? Do you reach the same conclusions as you do using a cutoff, but only after also using the RG equations?
I recall being told that when dimensional regularization is used to remove power law divergences there is additionally some optimistic assumption being made about the UV physics, but I don't know if that's correct or relevant to this problem.  
 A: Check this review on the cosmological constant problem for a nice discussion of what hierarchies look like in different regularizations.
Here is the rough idea. In cutoff or Pauli-Villars regularization the counterterms are sensitive to the cutoff scale(s) $\Lambda$. But there is no such scale when using dim.reg. (only the renormalization point $\mu$, which is usually a low scale). So what do you get instead? You need to add new physics.
Consider adding a heavy particle to your theory with mass $M$. This contributes through loops to the counterterms for the cosmological constant ($\propto M^4$), masses ($\delta m^2\propto M^2$) and couplings ($\propto \ln(M/\mu)$). So when talking about, say, the mass term of the Higgs boson, we say it is quadratically sensitive to the scale of new physics.
Note that this is a completely physical statement - no unphysical regulator scale appears in the argument. This would be the same if people made the argument properly in cutoff regularization: adding new particles and then doing the renormalization to get rid of the regulator scale. Talking about the cutoff dependence of quantities is really a nasty, unphysical shortcut.
