The neural representation of an individualized relational affective space

Humans experience emotions every day. Traditionally, psychology has described emotions through discrete labels (e.g. happy, afraid) or standardized affective dimensions (e.g. valence, arousal), and neuroscience has more recently sought the neurobiological basis of emotions via functional neuroimaging. However, by treating emotions similarly among everyone, we neglect that emotions are individualized; thus the overall relational structure of an individual's emotion information may be vital in understanding how the brain represents emotions. Combining behavioral and functional MRI experiments with similarity analyses, we demonstrate that neural activity patterns in the left insula correspond to the multi-dimensional arrangement of individuals’ affective spaces, despite interindividual differences, better than to a group-averaged model of affective space, a standardized valence-arousal space, a semantic category space, and a visual similarity space. This finding suggests that the insula may underlie individual-level affective information processing that is specific to one's own affective states, which offers new opportunities for functional neuroimaging to inform clinical approaches of disorders involving emotion dysregulation. •Participants free-sorted emotionally-charged images, yielding “emotion spaces”.•We used fMRI to measure brain activity while participants viewed the images.•Several models of emotion spaces were correlated with patterns of brain activity.•Individual emotion spaces explained activity in the insula better than other models.