A Bayesian model of category-specific emotional brain responses.
Level 3 - non-randomized controlled study
Meta-analysis of observational neuroimaging studies (graded by design analogy, not clinical CEBM).
PubMed 25853490 · doi:10.1371/journal.pcbi.1004066
What was done
The authors analyzed human brain activity patterns compiled from 148 neuroimaging studies of emotion categories (comprising 2,159 total participants) using a hierarchical Bayesian model. The model was evaluated on its ability to classify brain responses into five discrete emotion categories: fear, anger, disgust, sadness, and happiness, and to characterize the underlying cortical and subcortical spatial configurations.
What was found
The model classified the target emotion categories with 66% overall accuracy, with accuracy across individual categories ranging from 43% to 86%. Analyses revealed that each emotion category was associated with unique, prototypical patterns of activity across multiple brain networks—including neocortical systems, the thalamus, and the amygdala—rather than being localized to any single dedicated structure or system.
Why it matters
These findings challenge theories positing dedicated or purely subcortical emotion centers. Instead, they provide empirical support for constructionist and componential frameworks proposing that discrete emotional states emerge from distributed cortical-subcortical network interactions.
Limits
The abstract does not report the specific classification accuracy breakdown for each individual emotion category (reporting only the 43–86% range) nor the baseline chance performance. The analysis is constrained by the aggregated, coordinate-based nature of meta-analytic neuroimaging data rather than direct participant-level continuous recordings.
Cited by
- supports Neuroimaging research by Tor Wager and Lisa Feldman Barrett demonstrates reliable, distributed patterns of brain activation corresponding to basic emotions such as anger, fear, and sadness.