There is obviously a gap in my knowledge of the origin of effective actions in string theory.
As far as I understand it, the strategy is straightforward (at least in principle):
- Write down the worldsheet action for the superstring.
- Take compactification and (or) localization into account, obtain a low-energy action for the brane.
- Re-interpret this action as a QFT on the brane with a natural cut-off of order $l_s$ (string length).
OK, so that explains why we have a field-theoretic action of some form. However there is no reason (besides our intuition) for it to generate a QFT.
In path integral formulation, the fundamental measure comes from the (supersymmetric version of) Polyakov path integral over the fields on the superstring worldsheet.
It seems to me that it could (in general) generate another low-energy "effective measure", spanning another QFT.
How does one show that this is not the case?