Investigation of fNIRS brain sensing as input to information filtering systems

Abstract

Today's users interact with an increasing amount of information, demanding a similar increase in attention and cognition. To help cope with information overload, recommendation engines direct users' attention to content that is most relevant to them. We suggest that functional near-infrared spectroscopy (fNIRS) brain measures can be used as an additional channel to information filtering systems. Using fNIRS, we acquire an implicit measure that correlates with user preference, thus avoiding the cognitive interruption that accompanies explicit preference ratings. We explore the use of fNIRS in information filtering systems by building and evaluating a brain-computer movie recommender. We find that our system recommends movies that are rated higher than in a control condition, improves recommendations with increased interaction with the system, and provides recommendations that are unique to each individual.