Thought experiments have played a starring role in physics. They seem, sometimes, to pluck knowledge out of thin air. This is the starting point for my discussion this week with the philosopher Harald Wiltsche: what are thought experiments?How do they function — are they platonic laboratories with no moorings in observations or a way of supercharging our reasoning about phenomena?What do they deliver? Much emphasis has been put on the paradigm-shattering insights of Einstein where thought experiments appear like midwives in the production of new theories. But they can also function in explaining those theories, in ensuring they are understood.This leads to the question: what is understanding? Harald argues that it’s the ability to manipulate or deploy knowledge backed up by a mental model. He reasons that thought experiments can help us to make sense of the abstraction of mathematical models.We discuss many topics: the mathematization of science with Galileo, where thought experiments go wrong, transcendental arguments, the danger of losing sight of physical phenomena, and the links between Husserl and Mach. The theme that we dance to is this: there are lots of forms of reasoning that can work well within physics and we need equal pluralism in our philosophy of science to understand its startling, uncanny success in modeling nature. Harald's website haraldwiltsche.com  where you can find many of his papers and the introduction to his book with Phillipp Berghoffer on Phenomenological Approaches to Physics Twitter @haraldwiltsche Show notes on multiverses.xyz