Schedule of Events | Symposia

Sketchpad Series

Decoding of arousal and valence from fMRI data obtained during emotion inductions.

Poster Session D - Monday, March 31, 2025, 8:00 – 10:00 am EDT, Back Bay Ballroom/Republic Ballroom

Joel White¹ (jsw82@duke.edu), Yaohui Ding¹, Nathan Muncy², John Graner¹, Leonard Faul³, Kevin LaBar¹; ¹Duke University, ²University of Nebraska-Lincoln, ³Boston College

We are investigating the neural representations of valence and arousal, two central constructs in affective neuroscience signifying pleasantness and level of activation, respectively. Previous machine learning studies have successfully identified valence states from brain activation patterns observed during task-based fMRI, but the results have varied across studies, and the decoding of arousal states has been less successful. Moreover, prior studies have been limited in sample size, emotion induction methods, and extent of emotions sampled. Here we plan to decode arousal and valence from whole-brain fMRI data collected from 115 participants (67 female, mean age =30.21 (8.47)) during exposure to 300 unique emotional stimuli, presented in blocks of five stimuli. These stimuli were originally classified into one of 15 intended emotions, which span a broad range of emotional experience, and they were evenly distributed between short movies (Cowen & Keltner, 2017) and text scenarios (Faul, Baumann, & LaBar, 2023) taken from normed databases. Following MRI scanning, participants rated arousal and valence of the stimuli. Mass univariate general linear models will identify block-level activation (emotion stimuli > washout) from all grey matter voxels. Multivariate regression analysis will be used to predict post-scan valence and arousal ratings based on these grey matter activations. Preliminary results from 83 participants using partial least squares regression show promise in decoding both valence and arousal. Future work will utilize data from all 115 participants, further test and tune a greater range of parameters, and identify the voxels implicated in the neural representations of arousal and valence.

Topic Area: EMOTION & SOCIAL: Other