This chapter distinguishes the work of thought experiments in exemplifying concepts from their role in aiding cognition of regulative ideas. Contemporary interest in thought experiments as a “method of cases” treats thought experiments as providing instances or exemplars. For regulative ideas as Kant understands them, however, no observable instances or exemplars are possible. Nevertheless, thought experiments can direct attention toward regulative ideas negatively (by distinguishing them from what is observable) or positively by indicating a direction for extrapolation or ongoing inquiry. These positive uses are forms of cognition. The distinction between regulative and constitutive concepts matters for thought experiments that deal with regulative ideas such as the self – for example, for thought experiments about personal identity, where a number of objections to the use of thought experiments have been raised. I argue in this chapter that some of these objections can be answered by distinguishing regulative from constitutive concepts.

In philosophy of science, Mach’s account of thought experiments is more often described as relevant for contemporary usage than Ørsted’s. In this chapter, I survey recent Kantian accounts of thought experiment, arguing that the leading views inspired by Kant in philosophy of science remain broadly empiricist. This tendency may be due to their focus primarily on the role of thought experiments in the sciences. In later chapters, I will argue – against recent Kantian views – that Kant understood cognition more broadly to include not only sensory perception but also mathematical construction. Acknowledging that cognition does not always require empirical fulfillment opens new ways of understanding how thought experiments work in philosophy, which may rightly differ from their use in the sciences.