CogTracer

From individualized training, to responsive decision-support, and improved human-machine teaming, the ability to accurately predict the cognitive state of an individual in real time would open the door for numerous technologies that would benefit the operational needs of the warfighter. Until now, much of the research using EEG for operational needs has focused on tailoring a system to detect only a few cognitive states. This solution provides severely limited coverage of the space that this technology could be applied to, and is not a realistic path for developing neuroimaging as an operational asset. Soar Technology and Dr. John Anderson at Carnegie Mellon University propose the development of the CogTracer Toolkit. This toolkit supports the prediction and detection of cognitive states from electroencephalography (EEG) signals. The CogTracer toolkit will be integrated with the ACT-R cognitive architecture and use state-of-the-art machine learning approaches applied to neuroimaging data. Our Phase I work formalizes and extends the ongoing research of Dr. Anderson and colleagues for detecting cognitive states from neuroimaging data. Through research and validation, we will extend these methods to support a wide range of tasks, functional capabilities and EEG signal components.