Warrants for prescription: analytically and empirically based approaches to improving decision making.

Efforts to improve decision making must appeal to some source of warrant - that is, specific criteria or models for guiding and evaluating decision-making performance. We examine and compare the warrants for two approaches to decision aids, decision training, and consulting: analytically based prescription, which obtains warrant from formal models, and empirically based prescription, which obtains warrant from descriptive models of successful performance. We argue that empirically based warrants can provide a meaningful and valid basis for prescriptive intervention without committing the naturalistic fallacy (i.e., confusing what is with what ought to be) and without the use of formal deduction from first principles. We describe points of divergence as well as convergence in the types of warrant appealed to by naturalistic decision making and decision analysis, letting each approach shed light on the other, and explore the application of empirically based prescription to cognitive engineering. Actual or potential applications of this research include the development of training programs to improve various aspects of naturalistic decision making.