Making Sense of Uncertainty in Science and Policy

Not even wrong - Performativity of Science as an Edge Case for Uncertainty Talk

Jakob Ortmann (Leibniz University Hannover)

Abstract: It is a well-appreciated phenomenon that scientific models and theories (broadly construed) sometimes causally influence their targets; related terms are reactivity, reflexivity or performativity. This is usually cast as yielding epistemic and ontological problems; Ian Hacking, for instance, worried that with 'interactive kinds', there might not exist a stable target to have knowledge about. Many others, especially in economics, voiced concerns about the difficulties of predicting in the face of performativity - people could, the argument goes, always just choose to behave differently if they know about someone else’s prediction. In this talk, I advance the following claim: to regard performativity-induced (in)stability of target systems as a particularly 'epistemic' challenge is, in some ways, a category mistake - once we recognise a model as performative, we are already committed to regarding some salient model-target relations as causal, not epistemic, and thus committed to regarding the epistemic value of the initial model as, at best, incomplete. I argue that this has consequences for how epistemic concepts, such as uncertainty, should be deployed in such contexts. I conclude by outlining a ‘two-models model’ of performativity to help articulate these implications.