Modeling Epistemic Agents

Suppose you could define an ideal (maybe just a median) Bayesian agent. Then you could make definitions along these lines:

Testimony Skeptic: One who is more reluctant to accept testimony than the ideal Bayesian agent.

Testimony Denier: One who will not accept testimonial evidence at all.

You could start to model different epistemological types by defining scalars on outcomes of probability assessments for given classes of evidence, to get a piecewise function.  Like this:

P_α(h|e&b&c) =  P_ideal(h|e&b&c) for e not in B

                           P_ideal(h|e&b&c)*s for e in B

To get the definitions of skeptic and denier given above you let B be the class {evidence based testimony}. But you could let it be {evidence based testimony about __________}. You could have any number of classes defined in different ways and model all sorts of individuals this way.