Abstract: | I have previously argued that anti-realism cannot use the claim that theories have empirically equivalent rivals (EE) to establish that theories are underdetermined by evidence (UD). Douven objects that the auxiliary hypotheses needed to generate observational equivalence need not as I claimed be themselves underdetermined if (EE) is true, because they may be either compatible with observation or acceptable in the constructive-empiricist sense. I reply that these conditions are not strong enough to establish (EE): the auxiliaries have to be epistemically warranted if they are to fix a theory's observational commitments unambiguously. The support they require need not presuppose their respective observational consequences, because there are more holistic sources of warrant. |