Context-aware experience sampling reveals the scale of variation in affective experience

Katie Hoemann, Zulqarnain Khan, Mallory J. Feldman, Catie Nielson, Madeleine Devlin, Jennifer Dy, Lisa Feldman Barrett, Jolie B. Wormwood, Karen S. Quigley

Abstract

Emotion research typically searches for consistency and specificity in physiological activity across instances of an emotion category, such as anger or fear, yet studies to date have observed more variation than expected. In the present study, we adopt an alternative approach, searching inductively for structure within variation, both within and across participants. Following a novel, physiologically-triggered experience sampling procedure, participants' self-reports and peripheral physiological activity were recorded when substantial changes in cardiac activity occurred in the absence of movement. Unsupervised clustering analyses revealed variability in the number and nature of patterns of physiological activity that recurred within individuals, as well as in the affect ratings and emotion labels associated with each pattern. There were also broad patterns that recurred across individuals. These findings support a constructionist account of emotion which, drawing on Darwin, proposes that emotion categories are populations of variable instances tied to situation-specific needs.

Introduction

The search for physiological "fingerprints" of emotion has long dominated affective science. This classical view posits that categories like anger or fear correspond to specific, diagnostic patterns of autonomic nervous system (ANS) activity. However, laboratory findings have consistently demonstrated significant variation rather than uniformity. Critics suggest this lack of specificity stems from the artificiality of lab settings. This study addresses the debate by employing "context-aware experience sampling" to investigate whether physiological consistency emerges in naturalistic, real-world environments.

Methodology

The study followed 46 healthy participants over a 14-day period, accumulating over 400 hours of physiological data. Researchers utilized the MESA smartphone application, which triggered surveys only when detecting a substantial, sustained change in heart period (Interbeat Interval, IBI) while the participant remained physically still. This ensured that physiological shifts were likely psychological in origin rather than physical.

Physiological data included electrocardiography (ECG) and impedance cardiography (ICG). Participants provided self-reports on valence and arousal (100-point scales) and free-text emotion labels. Data were analyzed using Dirichlet Process-Gaussian Mixture Modeling (DP-GMM), an unsupervised machine learning technique that allowed physiological clusters to emerge inductively for each individual.

Results and Discussion

The analysis identified 219 distinct physiological clusters across the sample. Participants exhibited between 3 and 8 distinct physiological "modes" (mean = 4.76), representing recurring patterns of autonomic activity.

Physiological clusters for two participants showing distinct modes of autonomic activity.

The Many-to-Many Mapping

A key finding was the "many-to-many" mapping between physiological states and emotional experiences. A single physiological cluster often served as the biological substrate for diverse emotion labels. For instance, a single pattern might be labeled as "happy," "calm," or "annoyed" at different times. Conversely, the same emotion word (e.g., "tired") frequently appeared across multiple, distinct physiological clusters.

Distribution of self-generated emotion words across physiological clusters.

Statistical tests (Kruskal–Wallis) confirmed that for the vast majority of participants, there were no significant differences in felt affect (valence and arousal) between their different physiological clusters. This suggests that the ANS is primarily engaged in allostasis—the predictive regulation of metabolic resources—rather than signaling discrete emotion categories.

Conclusion

These findings support a constructionist, "population" view of emotion. Rather than universal fingerprints, emotion categories represent populations of variable instances tailored to specific situational demands. While individuals show unique physiological modes, 44 broad patterns recurred across the sample, suggesting a finite biological "menu" that the brain interprets based on context. Future research should incorporate broader contextual variables, such as social environment and GPS data, to further elucidate how the brain constructs emotional meaning from a combination of internal signals and external demands.

Summary updated

Abstract

Emotion research typically searches for consistency and specificity in physiological activity across instances of an emotion category, such as anger or fear, yet studies to date have observed more variation than expected. In the present study, we adopt an alternative approach, searching inductively for structure within variation, both within and across participants. Following a novel, physiologically-triggered experience sampling procedure, participants' self-reports and peripheral physiological activity were recorded when substantial changes in cardiac activity occurred in the absence of movement. Unsupervised clustering analyses revealed variability in the number and nature of patterns of physiological activity that recurred within individuals, as well as in the affect ratings and emotion labels associated with each pattern. There were also broad patterns that recurred across individuals. These findings support a constructionist account of emotion which, drawing on Darwin, proposes that emotion categories are populations of variable instances tied to situation-specific needs.
BibTeX
@inproceedings{hoemann2020context,
  author = {Katie Hoemann and Zulqarnain Khan and Mallory J. Feldman and Catie Nielson and Madeleine Devlin and Jennifer Dy and Lisa Feldman Barrett and Jolie B. Wormwood and Karen S. Quigley},
  title = {{Context-aware experience sampling reveals the scale of variation in affective experience}},
  booktitle = {Scientific},
  year = {2020},
  doi = {10.1038/s41598-020-69180-y},
}
Figure