Speaker: Adam Bjorndahl (Carnegie Mellon University)
Date and Time: Thursday, 30th May, 2024, 16:30-18:00
Venue: ILLC seminar room F1.15 in Science Park 107 and online.
Title: Belief and Causation
Abstract: We develop a formal framework for reasoning simultaneously about belief and causation. There is a long and rich history of mathematical models for these concepts, yet no work we are aware of that fully integrates the two. The combined model we propose is of interest from at least two perspectives. First, it allows us to represent beliefs about causal structure — for example, an agent who believes (perhaps falsely) that raising interest rates will cause inflation to cool. Second, it allows us to encode potential causal relationships involving beliefs — for example, an agent whose belief that it rained last night is a causal consequence of their lawn being wet. Notably, we can also capture potential causal relationships between beliefs — for instance, an agent whose belief in p is a causal consequence of their belief in q. This framework therefore allows us to deploy the machinery of causal models to implement a kind of belief revision, such as an agent who learns q thereby coming to believe p, where “learns q” in this context takes the form of a (causal) intervention setting the “variable” Bq to true.
This is joint work(-in-progress) with Joseph Halpern.