Much of the force behind doxastic involuntarism comes from our pre-theoretical judgement that any effort to form a belief simply by intending to form it must remain unsuccessful. However, despite this, ordinary language use of locutions like “chose to believe” are common. In this article, we present new experimental data that shows that the prevalence of ordinary language talk of “chosen beliefs” is no obstacle to doxastic involuntarism in a standard sense (pace Turri et al. 2018). While we employ the methods of experimental philosophy, our argument also concerns the fundamental question as to what those methods can achieve: as is typical of philosophical debates, the debate over doxastic voluntarism involves a refined theoretical concept, specifically a refined concept of voluntary belief. In such debates, we cannot determine our pre-theoretical position by conducting surveys on those who may not yet own the requisite theoretical concepts.