If I want to argue that hypothesis ๐ด is more plausible (i.e., more worthy of credence) than hypothesis ๐ต, I could frame this using Bayesian inference given some evidence ๐ธ by showing that ๐(๐ธโฃ๐ด)โ ๐(๐ด) > ๐(๐ธโฃ๐ต)โ ๐(๐ต).
There are, however, several points worth noting:
- This approach relies on probabilities, effectively reducing the concept of plausibility to that of probability.
- The comparison depends on prior probabilities ๐(๐ด) and ๐(๐ต), which can feel recursive because it seems to assume we already have an idea of which hypothesis seems more plausible to us before considering the evidence, but the whole point of the exercise is to calculate plausibility in the first place.
- Additionally, applying probabilities in certain contexts can seem problematic or out of place, particularly in domains where we lack the ability to develop statistics to validate whether the real world aligns with our probability estimates. For instance, when evaluating the plausibility of competing metaphysical worldviews (see related question), we lack a "metaphysical simulator" to generate frequencies for how often worldview ๐ด versus ๐ต might be true.
This raises the question: are there situations in which plausibility carries a meaning that cannot be fully captured by or reduced to the concept of probability? If so, how do philosophers suggest arguing for plausibility without relying on probabilities?
Or am I overthinking this, and plausibility and probability are essentially interchangeable concepts?
Closely related questions: