Experience sampling during fMRI reveals default network and executive system contributions to mind wandering

Abstract

Although mind wandering occupies a large proportion of our waking life, its neural basis and relation to ongoing behavior remain controversial. We report an fMRI study that used experience sampling to provide an online measure of mind wandering during a concurrent task. Analyses focused on the interval of time immediately preceding experience sampling probes demonstrate activation of default network regions during mind wandering, a finding consistent with theoretical accounts of default network functions. Activation in medial prefrontal default network regions was observed both in association with subjective self-reports of mind wandering and an independent behavioral measure (performance errors on the concurrent task). In addition to default network activation, mind wandering was associated with executive network recruitment, a finding predicted by behavioral theories of off-task thought and its relation to executive resources. Finally, neural recruitment in both default and executive network regions was strongest when subjects were unaware of their own mind wandering, suggesting that mind wandering is most pronounced when it lacks meta-awareness. The observed parallel recruitment of executive and default network regions--two brain systems that so far have been assumed to work in opposition--suggests that mind wandering may evoke a unique mental state that may allow otherwise opposing networks to work in cooperation. The ability of this study to reveal a number of crucial aspects of the neural recruitment associated with mind wandering underscores the value of combining subjective self-reports with online measures of brain function for advancing our understanding of the neurophenomenology of subjective experience.

Fig. 1.

Experimental paradigm. An experience sampling approach was used to collect self-reports about the subjects' focus of attention while their performed a concurrent task (the SART). Also, task accuracy at targets was used as a behavioral index of mind wandering. Analyses focused on the interval of time immediately preceding experience sampling probes to dissociate the effects of mind wandering from the effects of answering a probe.

Fig. 2.

Activations preceding reports of mind wandering (intervals prior to off-task versus on-task probes). Upward green arrows, default network regions; downward blue arrows, executive network regions. Regions of activation included: (A) dorsal ACC (BA 32), (B) ventral ACC (BA 24/32), (C) precuneus (BA 7), (D) bilateral temporoparietal junction (BA 39), and (E) bilateral DLPFC (BA 9). Height threshold P < 0.005, extent threshold k > 5 voxels.

Fig. 3.

Activations preceding SART errors (intervals prior to incorrect versus correct targets). Upward green arrows, default network regions. Regions of activation included: (F) Ventromedial PFC (BA 10/11) and (G) Dorsomedial PFC (BA 9). Height threshold P < 0.005, extent threshold k > 5 voxels.

Fig. 4.

Mind wandering in the absence (a) and presence (b) of meta-awareness. (a) Regions of activation associated with mind wandering in the absence of awareness (intervals prior to off-task unaware vs. on-task probes) included: (A) Dorsal ACC (BA 32), (B) Ventral ACC (BA 32), (C) Precuneus (BA 7), (D) Temporoparietal Junction (BA 39), (E) Dorsal Rostromedial PFC (BA 10), (F) Right Rostrolateral PFC (BA 10), (G) Posterior and Anterior Insula, and (H) Bilateral Temporopolar Cortex. (b) Similar regions were activated during mind wandering with awareness (intervals prior to off-task aware vs. on-task probes), but to a lesser degree, including: (A) Dorsal ACC (BA 32), (B) Ventral ACC (BA 24/32), and (G) Posterior and Anterior insula. Height threshold P < 0.005, extent threshold k > 5 voxels.

Fig. 5.

Regions showing greater activation during mind wandering in the absence of meta-awareness compared to mind wandering with meta-awareness (intervals prior to off-task unaware vs. off-task aware probes). Height threshold P < 0.005, extent threshold k > 5.