DISSOCIATING EXOGENOUS AND ENDOGENOUS INFLUENCES OF EMOTIONAL PROSODY ON NEURAL SPEECH TRACKING

Emotional prosody, the acoustic dimension carrying emotional tones, plays a central role in human communication. Yet it remains unclear how the brain detects emotions from these acoustic properties. One challenge is the bi-directional relationship between sound and emotion processing: While emotions can be detected from the sound acoustics (exogenous influence), the detection of a certain emotion may change how the sounds are processed (endogenous influence). Here, we aim to measure how emotional tones are represented in the human brain, disentangling that representation from the neural reaction to acoustic changes. Electroencephalography (EEG) signals were recorded from adult individuals as they were presented with continuous speech monologues and dialogues with happy, sad, angry, and fear emotional tones. Acoustic features were derived using the extended Geneva Minimalistic Acoustic Parameter Set (eGeMAPS). The features were grouped into emotion-sensitive (ES) or emotion-insensitive (EI) based on an emotion discrimination index derived via a linear discriminant analysis. EEG representation of ES and EI features were derived using forward encoding models (temporal response functions). The neural representation of ES was expected to change with the emotional tone, as a result of acoustic differences. Crucially, we tested the possibility that also the neural representation of EI changes with the emotional tone. Since EI represents acoustically emotion-insensitive features, such an outcome would reflect endogenous neural representation of emotion. The results reveal that emotion modulates both EI and ES neural representations, suggesting the possibility of probing endogenous affective modulation in isolation from exogenous factors.