No abstract

Different individuals respond differently to emotional stimuli in their environment. Therefore, to understand how emotions are represented mentally will ultimately require investigations into individual-level information. Here we tasked participants with freely arranging emotionally charged images on a computer screen according to their subjective emotional similarity (yielding a unique affective space for each participant) and subsequently sought external validity of the layout of the individuals' affective spaces through the five-factor personality model (Neuroticism, Extraversion, Openness to Experience, Agreeableness, Conscientiousness) assessed via the NEO Five-Factor Inventory. Applying agglomerative hierarchical clustering to the group-level affective space revealed a set of underlying affective clusters whose within-cluster dissimilarity, per individual, was then correlated with individuals' personality scores. These cluster-based analyses predominantly revealed that the dispersion of the negative cluster showed a positive relationship with Neuroticism and a negative relationship with Conscientiousness, a finding that would be predicted by prior work. Such results demonstrate the non-spurious structure of individualized emotion information revealed by data-driven analyses of a behavioral task (and validated by incorporating psychological measures of personality) and corroborate prior knowledge of the interaction between affect and personality. Future investigations can similarly combine hypothesis-and data-driven methods to extend such findings, potentially yielding new perspectives on underlying cognitive processes, disease susceptibility, or even diagnostic/prognostic markers for mental disorders involving emotion dysregulation.

No abstract

Many investigations into emotion processing contend that emotions can be reduced to a set of lower dimensions (e.g., valence and arousal). Additionally, emotion dysregulation is associated with numerous psychiatric disorders, whose treatment(s) may require inspiration from personalized medicine. To translate emotion research to the clinical domain, one may therefore need to investigate at the individual level, employing datadriven methods and forgoing classical assumptions regarding emotions. To this end, we explored the relative structure of emotion information resulting from 85 participants organizing emotionally-charged images following their own emotional responses to the pictures. Using cluster analyses and multidimensional scaling, we investigated the underlying composition of individuals' emotion spaces. Hierarchical clustering revealed five subtypes that reflect differing layouts of the emotion space; multidimensional scaling of each subtype's representative emotion space demonstrated that, although valence explained the primary organization of all emotion spaces, arousal as a secondary explanatory variable played a reduced role differentially for the subtypes, suggesting intrinsic differences in emotion information processing. Such data-driven methods yield new, unbiased ways of studying emotions and may reveal limitations of classic models or idiosyncrasies of individuals, which can inform future neuroimaging research and offer new approaches for studying emotions and emotion dysfunctions in psychiatric disorders.