Monitoring Attentional State with Functional Near Infrared Spectroscopy

Abstract

Functional Near Infrared Spectroscopy (fNIRS) is a technique for quantifying hemodynamic activity in the brain. Its portability allows application in real world operational contexts. The ability to distinguish levels of task engagement in safety-critical situations is important for detecting and preventing attentional performance decrement. We therefore investigated whether fNIRS can be used to distinguish between high and low levels of task engagement during the performance of a selective attention task, and validated these results using functional magnetic resonance imaging (fMRI) as a gold standard. Participants performed the multi-source interference task (MSIT) while we recorded brain activity with fNIRS from two brain regions. One was a key region of the “task-positive” network, which is associated with relatively high levels of task engagement. The second was a key region of the “task-negative” network, which is associated with relatively low levels of task engagement (e.g., resting and not performing a task). Using activity in these regions as inputs to a multivariate pattern classifier, we were able to predict above chance levels whether participants were engaged in performing the MSIT or resting. Classifier input features were selected from an array of probe channels at each of the two locations based on the fit to a model of expected task activity, or on training data. Standard linear regression was implemented with both static and adaptive general linear models to remove concurrently measured physiological noise. Two types of models were used to process the fNIRS signals. One employed knowledge of the task being performed to determine the system’s best capability. The other did not, for a realistic characterization. We were also able to replicate prior findings from fMRI indicating that activity in “task-positive” and “task-negative” regions is negatively correlated during task performance. Finally, data from both companion and simultaneous fMRI experimental trials verified our assumptions about the sources of brain activity in the fNIRS experiment, established a upper bound on classification accuracy expectations for response to the MSIT, and served to validate our fNIRS classification results. Together, our findings suggest that fNIRS could prove quite useful for monitoring cognitive state in real-world settings.