Siegel · Psychological bulletin 2018 · Multilevel meta-analysis and multivariate pattern classification · n=202 studies

Emotion fingerprints or emotion populations? A meta-analytic investigation of autonomic features of emotion categories.

Cited 506 times in the scientific literature.

Level 1 - systematic review of randomized trials

Systematic review and meta-analysis of experimental laboratory studies (graded by design analogy).

PubMed 29389177 · doi:10.1037/bul0000128 · record verified 2026-08-26

What was done

Authors conducted a random-effects, multilevel meta-analysis and multivariate pattern classification across 202 studies measuring autonomic nervous system (ANS) reactivity during laboratory emotion inductions in nonclinical adult samples. The analysis tested whether discrete emotion categories possess distinct physiological profiles ("fingerprints") versus highly variable responses ("populations"), assessing experimental moderators (such as induction modality) and publication bias.

What was found

Increases in mean effect size occurred for 59.4% of ANS variables across emotion categories, but overall effect size patterns did not clearly differentiate one emotion category from another. Substantial within-category heterogeneity was observed, accounting for moderate-to-substantial variability (I² ≥ 30%) in 54% of effect sizes. Induction type (films vs. imagery) did not account for a large portion of variability, and correcting for publication bias further reduced estimated effect sizes while increasing heterogeneity for certain categories.

Why it matters

This review challenges the classical hypothesis that distinct emotion categories have unique autonomic signatures. It provides broad empirical support for constructionist models viewing emotions as context-specific, biologically variable populations rather than fixed physiological archetypes.

Limits

The abstract does not report the total participant sample size, specific physiological channels analyzed, or exact effect sizes and confidence intervals. Findings are restricted to nonclinical adult populations in laboratory settings, which may not capture real-world emotional contexts, and were impacted by unexplained heterogeneity and publication bias.

Cited by