User Cognitive Modeling to Enhance Task Execution

Procedures are commonly used by organizations to specify, document, and disseminate prescribed methods for performing tasks efficiently and effectively. However, even well-trained personnel can make errors when carrying out procedures. The risk of these errors increases when task loads are too low or too high, when multi-tasking or switching between tasks, when interrupted, when complex team coordination or handovers are required, and/or during stressful situations. In these situations, users can become fatigued or complacent, or they can lose situation awareness due to overloaded working memory, automaticity, loss of vigilance, cognitive tunneling, ineffective information scanning, or susceptibility to confirmation and other biases. When these cognitive states occur, users are prone to committing errors such as wrong steps, skipped steps, mode errors, completion errors, default errors, and perseveration.The goal of this project is to develop an intelligent assistant that monitors users, such as crewmembers performing procedural tasks, maintains estimates of the crewmembers' cognitive states (including situation awareness and affective state), identifies situations where the user is at risk of making errors, and selects appropriate interventions that reduce the likelihood of errors. Depending upon the situation and the cognitive state of the user, the assistant will select an intervention that increases user awareness of important situational elements that the user may be missing; and by changing the level of automation, the assistant will reduce user workload.The assistant's modular architecture facilitates plugging in of different data models and algorithms required to monitor user performance, assess situation awareness and cognitive states, identify states that might lead to errors, and intervene to prevent those errors.