Iterative Experimental Design to Mature Cognitive State Classification Techniques from the Laboratory to Field Settings

This paper summarizes the research conducted on two parallel and complementary thrusts: cognitive state assessment (CSA) and mitigations development for augmented Cognition Systems. Cognitive workload classification research has been largely been limited to laboratory contexts with well defined tasks. However, as the technology matures, it is important to evaluate real-time, EEG- and ECG-based cognitive workload classification techniques when they are fully subject to the noise, motion, weather, and wide range of physical and cognitive tasks inherent in dismounted operational environments. The hardware and software infrastructure must reliably collect clean sensor data, mobile processing is needed to log and process the data, and the experimental design must reliably put participants in the cognitive states of interest. As mitigations mature, automation etiquette plays an important role in ameliorating the negative effects of strongly adaptive systems. In addition to the practical and system configuration challenges faced when moving from the laboratory to field studies, there are issues of experimental control and the characterization of cognitive state in less constrained task environments. A key component of an evaluation is to compare the classification results with "ground truth, " i.e. the actual workload experienced by the participant. While in the laboratory it is possible to develop simple tasks where workload is manipulated precisely and consistently, in an unconstrained field environment it becomes substantially harder to manipulate workloa