AFOSR
Many have argued that emotion is fundamental to cognition, so it is perhaps surprising that computational models of emotion have not received much attention from researchers in artificial intelligence and cognitive science until very recently.
Our objective is to model the relationship between emotion and cognition by developing an emotionally evocative interactive social problem-solving task where individuals can take explicit actions leading to measurable outcomes as well as describe their emotional responses.
This test bed will allow us to characterize aggregate trends and individual differences in problem solving behavior. Primarily, this will serve as a basis for validating and improving integrated models of emotion and cognition. Additionally, as subjects can engage in and receive feedback about the problem solving characteristics, it has potential as a training tool for teaching students to self-reflect on their emotional responses to situations and how this may alter their problem solving styles.
In terms of basic science, we would like to use computational models to concretize psychological theories concerning the relationship between emotion, cognition and behavior, and to collect a body of human performance data with which to validate and inform this process. In terms of engineering, we would like to construct a problem solving test bed that will allow direct comparisons between human and simulated problem solving performance. This test bed will have potential value in teaching trainees to better introspect on the influence of emotion on their own problem solving.
This project differs from others in the following ways:
- Emphasizes rigorous evaluation of a computational model of emotion against human performance data.
Goals
- Create a collaborative problem solving test bed that allows human subjects to engage in the emotionally-charged, uncertain, unstructured and socially complex problem solving situations in which emotional influences seem strongest, and which military decision makers are increasingly asked to confront.
- Collect and disseminate data on human performance and individual differences on this task.
- Assess the validity of a computational model of emotion's influence on human problem solving and characterize this model's ability to serve as a proxy for human behavior in training systems.
- Extend the modeling to incorporate the role of cultural and other group differences on emotion and cognition, and to begin to study how the surface form of communication influences beliefs and emotions: for example, what is the impact of text communication vs. audio communication vs. video communication, and the resulting increase of nonverbal signals of emotion.
- Explore the potential of this test bed as a training tool for teaching problem solvers how to reflect upon their own emotional state or the state of others and how this may influence performance on the collaboration.