Philosophy as theory:
Other views of philosophy: therapy, analysis, politics, history, poetry.
Beliefs are graded, with levels of confidence expressed by probabilities between 0 and 1.
To be rational, beliefs must be consistent with the laws of probability. Otherwise, the Dutch book argument shows that they will lose.
Bayes theorem: the probability of a hypothesis given the evidence is equal to the probability of the hypothesis times the probability of the evidence given the hypothesis, all divided by the probability of the evidence.
Bayesian confirmation: evidence E supports hypothesis H if P(H/E) > P(H).
Learning is changing probabilities in accord with Bayes theorem.
Beliefs do not have probabiliies: few grades.
Probability is useful for statistical phenomena (frequencies), but not for subjective degrees of belief.
Hence the Dutch book argument is irrelevant.
Bayes' theorem is a mathematical truth, but hard to apply in real life.
Explanatory coherence provides a more plausible account of scientific inference.
Human learning is non-probabilistic.
Bayesian epistemology is normative, not descriptive.
Agents do and should make decisions that maximize expected utility.
Expected utility (action) = SUM Probability (outcome) * Utility (outcome).
People rarely have information about probabilties and utilities.
People's preferences do not conform to the axioms of utility theory.
Decision making uses multiple criteria, not just utility.
Decision theory promotes immoral decisions: Kant vs. utilitarianism.
Well-behaved preferences can be represented by utility functions, and rationality is maximizing utility.
Games have players, strategies, and payoff functions. E.g. prisoner's dilemma.
Game players jointly maximize expected utility, seeking equilibrium.
Game theory can explain morals by agreement.
Rationality isn't maximizing utility.
The games described in game theory are highly artificial.
Social interactions are appropriately emotional, e.g. the ultimatum game.
The social contract is historically and psychologically irrelevant.
Computational Epistemology Laboratory.
This page updated Oct. 7, 2005.