Advanced Computational Approaches (ACA)
What is the optimal way to decode, track, and fuse neural and non-neural sources of information to infer state?
Current tools for monitoring human performance and assessing human state have, to date, suffered from a lack of robustness and that this lack of robustness is due in large part to the underlying variability of the human, which is not being adequately captured by current modeling and analysis techniques. Rather than disregard these failures as merely the by-product of the variability and non-stationarities that exist within and across individuals, we explore novel advanced computational approaches that explicitly target this variability in order to measure it, categorize it, and ultimately quantify it.
Technical Barriers To Current Computational Approaches:
The lack of mathematical modeling methods and software to find statistical relationships between moment-to-moment variations in environmental, behavioral, and functional brain.
The lack of sufficient data archives and resources to systematically study relationships between individualized models derived for cognitive monitoring and individual differences in performance across diversity of tasks.
Topic Area 1
Identify robust algorithmic representations of the internal states that impact human performance, in both natural contexts and in the context of human-agent interaction.
Topic Area 2
Quantify how augmentation signals, like new physiological or behavioral features, map onto or facilitate these algorithmic representations of performance states and develop computational tools to facilitate this mapping.
Topic Area 3
Investigate and develop novel algorithms to measure the changes in internal states across time, context, and individuals.