The metatheoretical calculus of connectionism — or: how we reason over artificial neural networks
Abstract:
I will demonstrate a way of understanding our reasoning as cognitive, neuro, and psychological scientists. This is done through creating a formal account of our adjudication over theories, a metatheoretical calculus. The example I will walk through is of connectionism, a highly influential cognitive scientific framework, that deeply implicates artificial neural networks. I will analyze these models: when used as scientific instruments of study and when functioning as emergent arbiters of the zeitgeist in the cognitive, computational, and neural sciences. Through these lenses, we can charactise two distinct tendencies within connectionism — classical and modern — with divergent properties (e.g. goals, mechanisms, and scientific questions). Ultimately, by engaging more deeply with the metatheoretical calculi we deploy when we think about artificial neural networks, our fields can obviate and reflect on the cycle of artificial intelligence winters and summers.
Relevant work:
- Guest, O. & Martin, A. E. (2021). How Computational Modeling Can Force Theory Building in Psychological Science. Perspectives on Psychological Science. https://doi.org/10.1177/1745691620970585
- Guest, O. & Martin, A. E. (2023). On Logical Inference over Brains, Behaviour, and Artificial Neural Networks. Computational Brain & Behavior. https://doi.org/10.1007/s42113-022-00166-x
- Guest, O. (2024). What Makes a Good Theory, and How Do We Make a Theory Good?. Computational Brain & Behavior. https://doi.org/10.1007/s42113-023-00193-2
- Guest, O. & Martin, A. E. (2025). A Metatheory of Classical and Modern Connectionism. Psychological Review. https://doi.org/10.1037/rev0000591
Passcode: 809383