The Situational Rationality of Task Performance: Artifacts of Belief in Human Problem-Solving and Artificial Intelligence
Understanding the artificiality of belief is crucial for the image it generates regarding task performance in the course of problem-solving. This paper examines the contribution of the artificial as a model of human intervention and its relationship to a model of human participation. Specifically, it details the logical differences underlying how belief operationalizes the perception of complexity and its effects on task performance in human and machine problem-solving. Three configurations of artificiality are presented to explain these differences and their effects on the relationship between representation and computation in problem-solving. The first describes the “natural artifact” that arises from a symbolic model of intelligence and the design of a maze. The second describes the “natural artifact” that arises from a sub-symbolic model of intelligence and the design of a mold. The third examines how a hybrid model of intelligence requires “socio-cognitive artifacts” in which a means of adaptation is primarily mediated by discourse rather than design. In doing so, the paper examines how momentary beliefs explain the situational rationality of task performance. The paper concludes with a commentary on the requisites of artificial intelligibility in machine problem-solving.