Feed

Human emotions are inherently complex, often manifesting intricate patterns of co-occurrence and variations across different contexts. Previous research has employed numerous indices to capture the complexity of emotional experience, yet these indices are usually mathematically redundant, leading to challenges in interpreting the multifaceted nature of emotions and their relevance to mental health. To address this, we propose a data-driven method to derive indices independently capturing co-occurring patterns of people's emotions across various contexts, which we term "emotion profiles," and evaluated it across three online community samples. Study 1 utilized video stimuli as emotion-eliciting contexts; 514 participants reported their experiences of four positive and four negative emotions in each context, which formed their emotion profiles. Parallel principal component analysis identified four significant principal components of emotion profiles (EP-PCs): context non-specificity, positive emotion co-occurrence, negative emotion co-occurrence, and mixed emotion. Study 2 recruited another 509 participants and assessed their emotion profiles twice over a 2-week interval. The four EP-PCs were replicated with different participants and different stimuli (r = .75 to .86), with EP-PC1 and EP-PC2 also showing good test-retest reliability. Study 3 employed an imagined scenario paradigm with 310 participants, confirming that the first three EP-PCs generalized across paradigms. Moreover, exploratory analysis revealed that EP-PCs mediated the associations between emotional granularity and mental health outcomes, clarifying distinct predictive pathways linking positive and negative emotional granularity to depressive and anxiety symptoms. This illustrates how the proposed method may inform emotion-mental health research.