Reasoning in Physics: The Bayesian Approach

This project addresses the question of what constitutes good scientific reasoning. The starting point is the observation that scientific reasoning takes place in the realm of uncertainty: We do not know whether the theories we posit are true, and the arguments we use to convince each other are usually not valid. So how can we assess scientific reasoning? Here we develop the idea that good reasoning is probabilistic (or “Bayesian”) reasoning: agents assign probabilities to various propositions, they have certain expectations about what will happen, and they rationally change the probabilities as new evidence becomes available. This simple idea not only explains many methodological truisms; it also has the potential to critically analyze forms of scientific reasoning such as we find, not least, in fundamental physics. More than any other field of research, fundamental physics suffers from the fate of having sparse (or nonexistent) empirical data. And yet the scientists involved want to make reliable statements about the world. If this aspiration is not to remain a dream, the question arises whether it is possible to draw conclusions about the world anyway. We pursue this question through a series of case studies and present our findings in a monograph.