Abstract
Assessment of mind wandering commonly entails the use of experience sampling to capture episodes in the midst of their occurrence. Time series sequences of responses to experience sampling probes have been shown to exhibit consequential temporal dynamics, including power-law scaling behavior in their fluctuations across increasing time scales. This scaling behavior implies long-range temporal correlations in the time series, in which mind wandering occurring far in the past can exert some influence on future episodes over increasingly large temporal intervals. Yet, it is known that short-range correlated Markov processes can also exhibit similar scaling behavior, making it critical that apparent long-range temporal correlations in time series measurements of mind wandering are evaluated with respect to short-range correlated null hypotheses. In the present study, we examine long-range temporal correlations in sequences of mind wandering probe ratings as well as in simulated sequences generated by short-range correlated Markov processes. To evaluate whether mind wandering probe ratings display genuine long-range temporal dependence, we contrast patterns of autodependence in mind wandering probe rating data with confidence intervals derived from short-range correlated Markov models and sequences temporally permuted to remove any temporal structure. We find that autodependence generally extended over short and intermediate ranges but was undifferentiated from short-range correlated processes later in time. This suggests that sequences of mind wandering probe ratings on average have finite memory with dynamics operating at shorter-range characteristic scales beyond which autodependence decays to that of a short-range correlated process.