Thought contents during rest account for functional connectivity-behavior associations

Resting-state functional connectivity (rsFC) is widely studied in relation to behavioral phenotypes. Typically, rsFC-behavior relationships are explained by attributing various psychological processes to the observed patterns of rsFC. Such explanations may seem far-fetched and unrelatable since subjects do not perform tasks that elicit these psychological processes during rest. Instead, they engage mostly in mind wandering. During such experiences, behavioral phenotypes can account for one's resting-state thought contents (rsTC), which, in turn, explain variability in rsFC. This study posits that rsTC can explain the relationship between rsFC and diverse behavioral phenotypes. We analyzed 765 resting-state fMRI runs from 164 subjects (each completed at least 3 runs) in relation to self-reported rsTC ratings and 63 behavioral phenotypic scores; rsTC ratings were collected using the Short Version of the New York Cognition Questionnaire after each run. In a series of linear mixed-effect analyses on the within-subject data, we showed that similarity in rsTC across runs, especially vigilance-related TC, significantly predicted similarity in rsFC fingerprints, and several dimensions of rsTC can be significantly predicted by one or more network-to-network connections. Cross-sectionally, canonical correlation analysis with behavior as the outcome was used to derive canonical variates from the multivariate rsFC and rsTC predictors. For 45 of the behavioral phenotypes examined, their rsTC canonical variates significantly mediated their respective rsFC canonical variate-behavior associations. Overall, we have rigorously shown that rsTC is significantly related to rsFC and behavior, and importantly, rsTC can partially account for most rsFC-behavior relationships examined in this study.