Danger signals for untrustworthy thought experiments

Abstract A key question in contemporary metaphilosophy of thought experiments is the “wheat from chaff” problem: How can we separate the good and trustworthy thought experiments from the untrustworthy ones? This article examines this problem by viewing thought experimentation as a form of mental simulation. It argues that we should approach the limitations of thought experiments in light of the general shortcomings of our capacity to run mental simulations. Furthermore, the article proposes an answer to the wheat from chaff problem by presenting three danger signals for untrustworthy thought experiments. These are (1) high counterfactuality of the imagined scenario, (2) complexity of the imagined scenario, and (3) a large psychological distance to the imagined scenario.