Publisher's Synopsis
The authors argue that a new theoretical foundation for artificial intelligence can be constructed in which rationality is a property of "programs" within a finite architecture, and their behavior over time in the task environment, rather than a property of individual decisions.