On Making Assumptions about Auxiliary Assumptions

Abstract

In my original article (Trafimow, 2009) I argued that although absolute falsification is impossible, more attention to auxiliary assumptions renders it possible to derive predictions that might be wrong. In their accurate and well-written Comment, Wallach and Wallach (2010) agree with this but also suggest that failed predictions will not overturn a theory if the level of confidence in the theory exceeds that placed on the auxiliary assumptions. Although Wallach and Wallach are doubtless correct about this possibility, it also is possible to have extremely convincing auxiliary assumptions. Furthermore, even in the absence of extremely convincing auxiliary assumptions, falsification can take place via the accrual of several failed predictions. Consequently, Wallach and Wallach’s argument does not weaken my original proposal.