Why do Computational Templates Work Across Scientific Disciplines?

Mariana Seabra (LanCog, Centre of Philosophy, University of Lisbon)

8 March 2024, 16:00 (Lisbon Time – WET)

Faculdade de Letras de Lisboa

Sala Mattos Romão [C201.J] (Departamento de Filosofia)

Abstract: Computational methods and their subsidiary models (including both physics-driven methods that appeal to differential equations, and data-driven computational methods that allow the extraction of meaningful patterns from data sets, often without explicit appeal to laws of nature or theories) are used to perform diagnosis, characterization and prediction in systems of interest. Furthermore, these computational methods are applied successfully across scientific disciplines, that is, the same computational structures, termed computational templates, are employed to solve problems in a wide range of scientific domains, from physics to biology, neurology, economics, and so forth. In this talk I try to explain the success of computational templates across different science fields, constructing a scientifically informed version of the ‘fudging solution’ for the applicability of mathematics as it arises for computational templates. I argue that the various templates available, from differential calculus to statistical models, capture change or changing tendencies in a system of interest. Corrections performed within the various stages of model construction not only concern updates in the formal structure of computational templates, but also progressively update the ontology of interest. What is perceived as the applicability of templates to physical reality is already the result of many such corrections, in which models are tailored to the system at hand.