Mind wandering in sensory cortices

The recent task-free approach in Cognitive Neuroscience has sparked interest in understanding the brain's default mode network (DMN). One particular mental activity that has been identified to recruit such a network is mind-wandering, which points to the functional aspect of mind-wandering as a default system. However, the phenomenological aspect of mind-wandering has been missing in the literature on brain imaging. To tackle this issue, we adopted online thought sampling while participants underwent a simple fixation task over multiple sessions in the scanner. During 10 h of scanning of each participant, over 200 mind wander episodes were labelled in each participant. With linear support vector machine classification on mind-wandering episodes with exclusive sensory content, we found that decoding accuracy in content-corresponding sensory cortices was significantly higher, indicating the neural bases of the phenomenology of mind-wandering. Unique patterns in classification were revealed in different individuals, pointing to individual variances in our phenomenal experiences.