Emotion Dynamics in Reciprocity:Deciphering the Role of Prosocial Emotions in Social Decision-making
Poster No:
707
Submission Type:
Abstract Submission
Authors:
Jaewon Kim1, Su Hyun Bong1, Dayoung Yoon1, Bumseok Jeong1
Institutions:
1Korea Advanced Institute of Science and Technology, Daejeon, Daejeon
First Author:
Co-Author(s):
Introduction:
Social decision-making is frequently investigated using an ultimatum game (UG). When responder rejects, none of the players receive the reward; hence, the proposer should act more prosocially.
Recently, predictive emotions, in the form of the emotion prediction error (EPE), were also reported to predict the choices of UG responders (Heffner, Son et al. 2021): Participants demand more prosocial interaction by punishing their partners when they receive smaller rewards or feel less pleased and more emotionally aroused than expected. Therefore, emotions in response to a proposer's offer could be better explored using basic and prosocial affect dimensions.
By adopting a dynamics approach and unsupervised neural network classification algorithm, we aim to investigate trajectory of responders' predictive emotions and social decision-making during UG. We predict that there will be groups that show distinct pattern of social decision-making, as well as related experiences of reward expectation, predictive emotions.
Recently, predictive emotions, in the form of the emotion prediction error (EPE), were also reported to predict the choices of UG responders (Heffner, Son et al. 2021): Participants demand more prosocial interaction by punishing their partners when they receive smaller rewards or feel less pleased and more emotionally aroused than expected. Therefore, emotions in response to a proposer's offer could be better explored using basic and prosocial affect dimensions.
By adopting a dynamics approach and unsupervised neural network classification algorithm, we aim to investigate trajectory of responders' predictive emotions and social decision-making during UG. We predict that there will be groups that show distinct pattern of social decision-making, as well as related experiences of reward expectation, predictive emotions.
Methods:
A total of 476 participants participated in an ultimatum game (UG). The rewards in each block were pseudo-randomised values with three fairness level. During each trial, partcipants responded expected amount of offer, decisions to either accept or reject, and emotions before and after offer.
K-means clustering was applied to the entire time series of participants' expected reward and reward acceptance, rather than individual time points. To identify groups of distinct emotion trajectories with fairness level changes, we applied t-distributed stochastic neighbor embedding (t-SNE) and a deep neural network unsupervised classifier. Then UMAP dimensionality reduction algorithm was applied to the time series of expected and experienced emotions for visualization (Figure). The kernel density estimation plot was visualised separately for each emotion group.
K-means clustering was applied to the entire time series of participants' expected reward and reward acceptance, rather than individual time points. To identify groups of distinct emotion trajectories with fairness level changes, we applied t-distributed stochastic neighbor embedding (t-SNE) and a deep neural network unsupervised classifier. Then UMAP dimensionality reduction algorithm was applied to the time series of expected and experienced emotions for visualization (Figure). The kernel density estimation plot was visualised separately for each emotion group.
Results:
Clustering of the participants based on the trajectories of their expected reward, reward acceptance, and expected and experienced emotions resulted in solutions with equal numbers of components. Clustering for all components resulted in solutions wherein the cluster centroids showed three distinct stationary trajectories with consistently low, middle, and high values and one distinct dynamic trajectory that chased the actual reward. UMAP embedding results were plotted, and each emotion group's kernel density estimation plot was overlaid separately (Figure). The results supported our hypothesis that individuals will be grouped into those with distinct pattern of reward expectation, social decision, and emotion experience.
Conclusions:
We identified inherent subsets of participants who show distinct temporal pattern of reward expectation, social decision making, and emotion experiences and propose a novel algorithm that can identify these clusters.
Emotion, Motivation and Social Neuroscience:
Reward and Punishment
Social Interaction 2
Social Neuroscience Other
Emotion and Motivation Other 1
Higher Cognitive Functions:
Decision Making
Keywords:
Affective Disorders
Computational Neuroscience
Emotions
Social Interactions
1|2Indicates the priority used for review
Provide references using author date format
Russell, J. A. and A. Mehrabian (1977). "Evidence for a three-factor theory of emotions." Journal of research in Personality 11(3): 273-294.
Tangney, J. P., J. Stuewig and D. J. Mashek (2007). "Moral emotions and moral behavior." Annu. Rev. Psychol. 58: 345-372.