Nonlinear neural representation of emotional feelings elicited by dynamic naturalistic stimulation

Emotions guide organisms in constantly changing dynamic environments. While the majority of previous studies have assessed neural basis of emotions elicited by impoverished stimuli (e.g., pictures or words), it has been shown that dynamic naturalistic stimuli (e.g., movies and stories) elicit stronger subjective emotional experiences. Here, we studied the neural basis of subjective feelings of pleasantness and unpleasantness (i.e., emotional valence) and emotional arousal by presenting sixteen participants with thirteen emotional and neutral movie clips during 3-Tesla fMRI scanning. After scanning, the participants viewed the clips again and rated their moment-to-moment subjective experience of valence and arousal. These participant-wise experiential time series were used to model each participant’s brain activity during viewing of the movies, while factoring out contributions of physical stimulus features. The results suggest that a set of cortical and subcortical areas encode subjective feelings in nonlinear fashion; we observed a quadratic (U-shaped) dependency between the experienced emotional valence and BOLD signal in insula, somatosensory cortex, anterior and middle cingulate cortices, hypothalamus and thalamus, inferior frontal gyrus, and in medial prefrontal and orbitofrontal cortex. We propose that this network of brain areas represents the subjectively experienced valence as deviations from the neutral emotional state.