One of the many questions that have been asked about thought experiments is the ontological question concerning what thought experiments are, and more specifically, if they are some sort of experiments – perhaps experiments with specific characteristics, but nevertheless experiments. After all, if they were not experiments, how could they bring us new genuine knowledge about the world ? But, on the other hand, since they are just counterfactual scenarios, how could it make sense to say that they are experiments ? Thus the ontological question of what thought experiments are is entangled with the epistemological question of understanding where the knowledge that they bring may come from and what kind of knowledge exactly they can bring. Without dissociating the ontological question and the epistemological question, we will defend in this paper a dialectical account of thought experiments.
In order to defend this dialectical account of thought experiments, we will proceed in the following way :
First, starting from a fairly broad characterization of what thought experiments are in general, we will focus on a tension between two of their characteristics, a tension that puts thought experiments in what we will describe as an ontological state of unstable equilibrium.
Second, having interpreted the epistemological debate between James Robert Brown and John Norton in terms of this ontological unstable equilibrium, we will clarify our position in this debate. To put it bluntly, we agree with Norton when he argues against Brown’s Platonist positions, but we do not agree with him when he maintains that thought experiments are only arguments in the sense of valid arguments. According to us, scenarios that thought experiments involve always including some kind of opacity that can not be immediately reduced, thought experiments can not be assimilated to arguments.
Third, we will argue that, to account for the acquisition of knowledge that is at stake in thought experimenting, one should propose a dialectical account of argumentation.
Finally, we will answer some possible objections.