Mind-wandering—the spontaneous shift in attention away from the external task to internal thoughts (including daydreaming, fantasizing, rumination, and worrying)—is negatively associated with performance across a variety of tasks including the sustained attention to response task, the Stroop task, tasks of working memory, choice reaction time, visual search, as well as more ecologically related tasks like reading comprehension and mathematics. There has also been promising evidence suggesting a potential link between mind-wandering, functional connectivity of the canonical networks of the brain, and Alzheimer’s disease (AD). However, no study has directly examined the relationship between neural correlates of mind-wandering and AD pathogenesis. In prior work, our lab has identified a whole-brain, functional connectivity-based marker of mind-wandering—the mwCPM—which predicted response time variability in older adults. In this study, we sought to evaluate the ability of this mind wandering CPM, derived from response time variability, to predict CSF p-tau/Aß42 ratio in 289 older adults from the Alzheimer’s Disease Neuroimaging Initiative. We hypothesized that the combined mind-wandering model including functional connections that predict high mind-wandering and functional edges that predict stability in attention, would predict AD pathology.

Resting-state functional MRI data from 289 older adults (147 healthy older adults, 111 individuals with mild cognitive impairment, and 31 older adults with AD) from the Alzheimer’s Disease NeuroImaging Initiative was analyzed for the current study. Participants were only included in the analyses if they had resting-state fMRI data, CSF measures of amyloid beta and tau pathology, and performance on cognitive composites of global cognition, episodic memory, and executive functioning. Using the well-established methodology of connectome-based predictive modeling, the mind-wandering model was applied to the resting-state fMRI data to predict CSF-based biomarker levels of p-tau and Aß42. Moreover, we also examined if this mind-wandering model predicted individual differences in composite measures of global cognition, episodic memory, and executive functioning

The high mwCPM model successfully predicted measured CSF p-tau/Aß ratios (high model: p = .137, p = .0196), controlling for mean framewise displacement. However, the combined network and the low MW network were not significant (combined model: p = .0731, p = .216; low model: p = -.0027, p = .960). We next examined the association between connectivity strengths of the high mwCPM and cognitive functioning in the domains of general cognition, episodic memory, and executive functioning. Connectivity strength in the high mwCPM—functional edges that were associated with high behavioral variability—were negatively associated with all three cognitive composites (global cognition: r = -.239, p < .0001; episodic memory: r = -.208, p = < .0001; executive functioning: r = -.178, p < .0001).

This study provides the first empirical support for a link between a neuromarker of mind-wandering and AD pathophysiology. Moreover, mind-wandering also has downstream consequential effects for key domains of cognitive functioning in older adults. Interventions targeted at reducing mind-wandering, particularly before the onset of AD pathogenesis, may make a significant contribution to the prevention of AD-related cognitive decline.