A wandering mind is a less caring mind: Daily experience sampling during compassion meditation training

Abstract

Mind wandering, or the tendency for attention to drift to task-irrelevant thoughts, has been associated with worse intra- and inter-personal functioning. Utilizing daily experience sampling with 51 adults during 9-weeks of a compassion meditation program, we examined effects on mind wandering (to neutral, pleasant, and unpleasant topics) and caring behaviors for oneself and others. Results indicated that compassion meditation decreased mind wandering to neutral topics and increased caring behaviors towards oneself. When collapsing across topics, mind wandering did not serve as an intermediary between the frequency of compassion meditation practice and caring behaviors, though mind wandering to pleasant and unpleasant topics was linked to both variables. A path analysis revealed that greater frequency of compassion meditation practice was related to reductions in mind wandering to unpleasant topics and increases in mind wandering to pleasant topics, both of which were related to increases in caring behaviors for oneself and others.

Keywords:

• mind wandering
• awareness
• attention
• meditation
• compassion
• caring behavior

Acknowledgments

The authors would like to thank the two anonymous reviewers for their careful reading of the manuscript and for their insightful comments and suggestions.

Disclosure statement

No potential conflict of interest was reported by the authors.

Notes

1. Caring behaviors is an umbrella term referring to a variety actions performed by individuals that are intended to benefit oneself or others.

2. Only participants with a 50% or greater response rate across all assessment points (2 samplings per day × 56 days = 112 potential samples) were included in these analyses, this excluded 9 participants who did not complete the intervention.

3. After being signaled to make a response, participants were instructed to make a response at that moment or if busy (e.g. driving, meditating, etc.), to make a responses as soon as possible after the current activity was completed. 88% responses were done via e-mail (both via non-iPhone smartphone devices and computers), 10% of responses were done via the iPhone application we developed (downloaded via the Apple App Store), and 2% of responses were done using a paper/pencil format (when participants were out of cell and computer access). Mind wandering and caring behavior outcomes did not differ by response format (ps > 0.08). Although perhaps considered a ‘trend’, given the sufficient power to detect differences (i.e. large sample size and large number of repeated-measurements), the lack of significance appears to be a reliable result. Aside from the mind wandering, meditation practice, and caring behaviors questions participants were also asked about emotion regulation attempts and self-efficacy (not reported here).

4. When calculating ICCs for binary outcome variables, the level-1 residual variance is estimated as π²/3 (i.e. the standard deviation of a logit function; Snijders & Bosker, 1999).

5. The direct effect, c’, was also estimated but is not relevant when testing mediation (Hayes, 2009; Rucker, Preacher, Tormala, & Petty, 2011).

6. We used Selig and Preacher's (2008) web-based utility, in which we input the unstandardized path estimates (a₁, a₂, b) as well as the corresponding slope variances and covariances (specific to mediation analyses where all variables are measured at level 1) to simulate the sampling distribution of the indirect effect. We requested a 95% confidence interval (CI) and 20,000 repetitions. If the resulting 95% CI does not include zero, the indirect effect is significant at α = 0.05.

7. Traditional ordinary least squares regression-based methods for assessing mediation (e.g. Baron & Kenny, 1986; MacKinnon, Lockwood, Hoffman, West, & Sheets, 2002) require the assumption of independence of observations and are appropriate when assessing single-level mediation. In the presence of multilevel data, use of these traditional methods yields downwardly biased standard errors (MacKinnon et al., 2002).

8. On average, the percentage of missing data increased slightly from week 1 (17%) to week 8 (31%); however, no common or consistent patterns of day-to-day ‘missingness’ were found throughout CCT.

9. Skewness values were –1.49 and –1.61 for caring for oneself and caring for others, respectively. Kurtosis values were 1.50 and 2.77 for caring for oneself and caring for others, respectively.

10. Because the time predictor reflects the number of assessments/pings during the study period, the estimated odds ratio indicates mico-level changes from one ping to the next. For interpretability, week-to-week odds ratios are reported to correspond with the weekly trajectories depicted in Figure 3.

11. Baseline ratings are minimal (from 0 to 3 ratings), so the estimated likelihood of mind wandering is heavily weighted by the Week 1 ratings (17.2% mind wandering to pleasant). This decrease (from Week 1) is depicted in Figure 3.

Additional information

Funding

This work was supported by the Fetzer Institute and the Center for Compassion and Altruism Research and Education (CCARE).