Jinwoo Lee
Research
Awe is characterized as an ambivalent experience in the human behavior
and cortex: integrated VR-EEG study
Motivation
Awe is a multifaceted emotion that encompasses both positive and negative feelings. Although qualitative descriptions of awe often highlight its ambivalent nature, it has not been fully investigated. This gap exists because the dominant measurement of affective valence in contemporary affective science - the 1D biploar continuum scale - does not accommodate ambivalent responses. Consequently, empirical studies of awe typically divide it into two types based on dominant valence (positive or negative awe). However, this dichotomy fails to capture the mixed nature of awe. Therefore, we examined whether awe experiences are more accurately characterized by ambivalence than by simple positive or negative states, both behaviorally and neurologically.
Behavioral findings
A total of 43 participants watched three awe-inducing VR clips and one positively-valenced control clip while their EEG signals were recorded. During the viewing task, participants reported their valence states in real-time using keypresses (i.e., positive, negative, mixed and neutral). After each trial, they also reported their overall valence, arousal, awe intensity, and motion sickness (fig 2a).
Our findings revealed that the awe-VR clips elicited more intense awe experiences than the control clip, as intended (fig 2b). Additionally, ambivalent feelings lasted longer and were more vivid during the awe conditions compared to the control clip (fig 2c & 2d). Univariate and multivariate modeling showed that only ambivalence-related metrics significantly predicted awe intensity scores, unlike other positivity/negativity metrics (fig 2e, 2f, 2g, & 2h).
Using the key-pressed valence sequence, we examined how the four valence states are distinctively represented in cortical regions. We constructed latent valence-cortical spaces at the individual level using a contrastive learning approach, which aims to repel different valence categories and attract the same ones as much as possible (fig 3a).
We found notable individual variability in how ambivalent states are distinctively represented from single-valence ones in the cortex. This variability in cortical distinctiveness specifically predicted awe intensity (fig 3b). Finally, we used perturbation-based XAI techniques to determine which EEG features were most involved in differentiating valence states during VR viewing (fig 3c). We obser ved that frontal delta power played a crucial role in distinguishing valence states in the brain (fig 3d).
© 2024 Jinwoo Lee. All rights are reserved.