Predicting intentional and unintentional task unrelated thought with EEG

Our attention seldom remains on a singular activity, instead veering off into thoughts unrelated to the task at hand. Studies adopting a component process view of off-task thought have begun to identify the underlying mechanisms and associated electrophysiological correlates underlying ongoing thought. In the present study, we developed subject-independent classification algorithms based on electroencephalographic (EEG) markers to discriminate on-task vs off-task as well as intentional vs unintentional off-task thought. To that end, spatio-temporal and spectral features extracted from EEG activity prior to reports of ongoing thought during a test of sustained attention were ranked according to their discriminative power. Using data collected from 26 participants, average classification accuracies of 83.4% and 71.6% were achieved using a regularized linear model for on-task vs off-task and intentional vs unintentional off-task thought, respectively. Our results identified gamma oscillations as the most discriminative feature to distinguish on-task from off-task states, and alpha synchronization as the most prominent feature when off-task states are engaged in deliberately rather than when experienced as arising spontaneously. Our work represents the first successful attempt at reliably discriminating the degree of intentionality experienced during task-unrelated thought and highlights the importance of recognizing the heterogeneous nature of off-task states.