We sketch a theory of decision that allows us to construct both goals and degrees of belief. Before choosing an action, we create and weight goals. We represent our beliefs about the consequences of each action by constructing a belief function on the set of possible consequences. Then we combine the beliefs and goals to see the value that will be secured and the value that will be excluded by each action. This approach is more constructive than Bayesian decision theory, because as Savage's problem of small worlds teaches us, that theory assumes preferences that antedate deliberation.
All Science Journal Classification (ASJC) codes
- Theoretical Computer Science
- Artificial Intelligence
- Applied Mathematics