Suppose I have a set of propositions and "exogenous" credence levels for each proposition i.e. how likely I think it is to be true. Then suppose Ι know that certain logical relations hold between the propositions. I could then try to find the assignment of true and false values for the propositions that maximizes the absolute (or squared etc.) error (i.e. 0 or 1 minus the credence) subject to the logical relation constraints.

I'm guessing this sort of thing has a name in formal epistemology. What is it and where can I read about it? Thanks

(I was trying to notate this post a little better but it doesn't seem to support mathjax).

1 Answer 1


This is essentially maximum likelihood estimation. If you replace your credence levels with probability distributions, finding truth values for your propositions that maximize the credence (subject to constraints) is equivalent to finding truth values for the propositions that maximize the joint likelihood.

You may wish to look at energy based models.

  • Interesting thank you. I see maximum likelihood is one approach, though it's only one special case of various potential loss functions. And I'm still pretty sure there must be a general literature about this sort of thing with a name.
    – Smithey
    Jan 4, 2021 at 0:04

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .