EEG Based Decoding of the Perception and Regulation of Taboo Words

Abstract

Emotional meaning is often conveyed through language, and certain word categories, such as taboo words, can elicit strong affective responses that may require regulation. While taboo words are socially and psychologically salient, their neural processing remains to be fully explored. In this study, we investigated whether EEG signals can be used to predict neutral, negative, and taboo words, and whether this prediction is preserved under conditions of emotion regulation. Forty native Italian speakers viewed 240 words across these categories while EEG was recorded. Participants completed two conditions: a Look condition (passive observation) and an Accept condition (emotion regulation via acceptance). Using support vector machine (SVM) classifiers applied to event-related potentials (ERPs), we found that word categories could be reliably decoded from the late positive potential (LPP) in central-parietal-occipital and anterior right regions between 450 and 850 ms. Notably, classification remained above chance but largely reduced even during the Accept condition, suggesting that word-related affective information persists at the neural level despite the regulation effort. These findings advance our understanding of emotional language processing and highlight the utility of machine learning for decoding subtle neural representations.