Perseverative cognition and vagally mediated heart rate variability in laboratory studies: A series of meta-analyses
Chung Xiann Lim · Emily Mueller · William H. O’Brien
International Journal of Psychophysiology · 2026
What the research tells us
28 laboratory studies contributed 60 effects to six analyses, separated by phase and comparison. Resting correlations yielded g=0.46 (95% CI 0.26–0.66), within-person reactivity 0.22 (0.11–0.33), and within-person recovery 0.19 (0.03–0.36). All three between-group pooled effects were nonsignificant. Positive g was coded as lower vagally mediated HRV (vmHRV), not an increase in HRV.
Scope & interpretation
The review was not preregistered. Several analyses had substantial unexplained heterogeneity; resting correlations and within-person reactivity showed publication-bias signals. Removing one recovery outlier reduced g from 0.19 to 0.12 (95% CI 0.00–0.23; p=0.05). Within-person changes do not automatically exclude time or task confounding; nonsignificant between-group results do not prove absence of an effect. Laboratory vmHRV is not a measure of whole-person recovery, a diagnosis of perseverative cognition, an allostatic-load score or a long-term disease outcome.
Study at a glance
- POPULATION & SAMPLE
- 28 studies, 60 effects. Resting correlations k=14/n=2,296; resting between-group k=5/n=246; reactivity between-group k=9/n=610 and within-person k=18/n=818; recovery between-group k=5/n=264 and within-person k=9/n=349. Here k counts effects within each analysis; studies and samples can overlap.
- EXPOSURE / INTERVENTION / SUBJECT
- Self-reported perseverative cognition or laboratory worry/rumination induction, distinguishing rest, induction reactivity and the period after participants were told induction had ended. Between-group analyses combine high/low cognition groups and induction/control groups; resting correlations are separate.
- OUTCOMES
- Standardized vmHRV differences. Within-person recovery compares the post-induction period with the original resting baseline; it is not directly a recovery-speed estimate. The three principal estimates and CIs appear in the findings; between-group estimates all crossed zero.
- TIME FRAME
- Phase-specific vmHRV recordings lasted 2–15 minutes, with a median of five minutes. These observations do not establish effects over months or years.
Cited on this site
Recovery Is Not Rest
How Is Recovery from Stress Measured?
Perseverative Cognition: Why Can Stress Outlast an Event?
What Can HRV Tell Us About Recovery?
Why Can Stress Responses Continue After Work Ends?
Why Does HRV Drop? How to Read a Low Value in Context
Editorial & review record
Record for v1.0 · Updated
Editorial responsibility: TAP Editorial Team
Editorial note
Under the editorial responsibility of the TAP Editorial Team, this publication used AI assistance for content organization and editing.
Download detailed editorial record (JSON) ↓
How responsibilities and review are recorded ↗Scope of this reading note
Read the 55-page v2 accepted manuscript posted by the authors on PsyArXiv on 22 March 2026: methods, six main analyses, sensitivity results, limitations and figures/tables; compared with the published 2026 abstract and bibliographic record. This accepted manuscript is distinguished from earlier preprints. Separately cited tables A1–A6 and supplements S1–S4 were not obtained; pooled effects were not independently recomputed.
- Study funding
- No separate funding statement was located in the accepted manuscript consulted. This does not establish absence of funding.
- Competing interests
- On the accepted manuscript title page, the authors declare no conflict of interest.
Project support
Founded and supported by Resilio.
Read the editorial standards ↗