Because Steve's emotion states are based on antecedents, it is possible for him to give rich explanations for why he feels the way he does. These in themselves can be useful pedagogical tools: One emotion state may be the result of many different goals, principles, and a large collection of intensity values. The student might, for example, be motivated to ask Steve what makes him happy, or angry. The explanation can serve to inform the user that Steve cares about the students progress, wants the user to learn safety rules, and so forth.
Agents, such as Steve can have multiple, and even conflicting, goals and principles, just as people do (e.g., he might be simultaneously experiencing hope that the user will not have an accident subsequent to training, based on instances of caution, and em fear that the user will have an accident, based on instances of sloppiness).
Although not likely to be stressed in the first pass at this work, theoretically-based causality can result in interesting, yet cohesive, changes in the ways Steve appraises the world. For example, a negative emotion might lead Steve to have ``self-directed attributions of negative worth.'' This in turn can be expressed as a change in the thresholds for variables used in the match process, putting him in a bad mood, or making it easier for him to be saddened, and harder for him to feel happy. Such a technique would allow us to support a tutoring model that includes Steve getting depressed about a student's poor attention, which in turn is consistent with fostering the belief that Steve and the are both in the learning process together.