Modelling Abstraction

At the risk of oversimplifying, models work by abstracting from the complexity of things: they shed detail and introduce distortion toward ends as numerous as the models themselves. This is as true of neural networks — which can compress huge amounts of online data to an app that can be run on a smartphone — as it is of crash-test dummies, which render the human body in a plastic form better suited for studying high-speed impact. At the same time, models are more than just stripped-down versions of the real thing, instead they are artefacts that produce unexpected social and epistemic effects, and that can end up transforming what they were merely designed to represent.

This event explores the problem of abstraction through the lens of modelling practices, taking as its point of departure the recently published volume Breaking and Making Models (ICI Berlin Press, 2025). The discussion focuses on a central philosophical tension articulated throughout its pages: the role of abstraction in enabling, shaping, and sometimes constraining the epistemic and imaginative scope of scientific and artistic practices. Contributors to the discussion will consider the productive and limiting aspects of abstraction in modelling, from grounding knowledge of hardly accessible events in the world to obscuring and distorting the complex ‘nature’ of those same events.