Using fNIRS for Real-Time Cognitive Workload Assessment.

In this paper, we evaluate the possibility of detecting continuous changes in the user's cognitive workload using functional near-infrared spectroscopy fNIRS. We dissect the source of meaning in a large collection of n-backs and argue that the problem of controlling the content of a participant's mind poses a major problem for calibrating an algorithm using black box machine learning. We therefore suggest that the field simplify its task, and begin to focus on building algorithms that work on specialized subjects, before adapting these to a wider audience.