Intentions, Errors, and Experience

Intentions are schematic states of working memory that instantiate goals by specifying desired outcomes, actions, and the objects of those actions, thus serving a role in realtime control of behavior. Fluent performance may be achieved in part by concise representation of intentions, which may entail representing some elements deictically-specifying only the time and place at which information is available-rather than semantically. This chapter reviews theoretical considerations and empirical results that support this deictic specification hypothesis, demonstrating that deictic specification has both benefits for performance and costs in terms of limits on metacognition. When regularities in the performance environment, such as consistency in the time and place at which information is available, allow deictic specification, performance is superior but is characterized by predictable error patterns and poor error monitoring. Studies of event counting illustrate these phenomena. This analysis of intentions provides a basis for integrating theoretical understanding of cognitive control and of metacognition.