ABSTRACT
Today's resource constrained environment has pushed the training community toward more intelligent, automated training systems, which provide an opportunity to reduce instructor requirements and increase training throughput. One challenge with this, however, is that in order for an intelligent training system to achieve individually tailored training that is similar to the capability of a human instructor, the system must be able to assess trainee performance, and how that performance is influenced by a trainee’s affective (e.g., frustration, anxiety) and cognitive (e.g., workload, engagement) learning states. With the emergence of advanced measurement technology (e.g., eye tracking, electroencephalography, etc.) access to these cognitive and affective states by an intelligent automated system is becoming more feasible. This paper reviews behavioral, cognitive, and physiological metrics that have been validated to provide an indication of learning states and thus could be used to derive the diagnostics necessary to drive tailored training solutions.
