Automatic Detection of Mind Wandering from Multimodal Datastreams: A Survey of State-of-the-art Methods

Mind wandering is a cognitive phenomenon whereby a person’s focus of attention involuntarily drifts from a task that he or she is currently engaged in, to unrelated thoughts. This “Zoning-out” is quite common, but can occasionally have negative side-effects. For example, while engaging in tasks like reading or listening to a lecture, mind wandering is associated with a loss of meaningful comprehension.

For this reason, there exists a growing interest in enabling intelligent applications to automatically detect such episodes of mind wandering in their users, providing a window for timely interventions. To achieve this, such systems typically rely on machine learning to identify patterns in physiological measures and/or the behavior displayed by users during interactions.

The aim of this literature review is to provide the reader with a structured overview of the state-of-the-art for the detection of mind wandering from such multimodal data streams.

---

Survey performed in April 2019