Biosignal-Based Recognition of Cognitive Load: A Systematic Review of Public Datasets and Classifiers

Cognitive load is a user state intensively researched in the NeuroIS community. Recently, the interest in designing neuro-adaptive information systems (IS) which react to the user’s current state of cognitive load has increased. However, its measurement through surveys is cumbersome and impractical. Alternatively, it was shown that by collecting biosignals and analysing them with supervised machine learning, it is possible to recognize cognitive load less obtrusively. However, data collection and classifier training are challenging. Specifically, large amounts of data are required to train a high-quality classifier. To serve this need and increase transparency in research, more and more datasets are publicly available. In this paper, we present our results of a systematic review of public datasets and corresponding classifiers recognizing cognitive load using biosignals. Thereby, we want to stimulate a discussion in the NeuroIS community on the role and potential of public datasets and classifiers for designing neuro-adaptive IS.