Learning from ingroup experiences changes intergroup impressions
Poster No:
820
Submission Type:
Abstract Submission
Authors:
Yuqing Zhou1, Björn Lindström2, Alexander Soutschek3, Pyungwon Kang4, Philippe Tobler4, Grit Hein5
Institutions:
1Chinese academy of Sciences, Beijing, Beijing, 2Karolinska Institutet, Stockholm, Sweden, 3Ludwig Maximilian University, Munich, Germany, 4University of Zurich, Zurich, Switzerland, 5University of Würzburg, Würzburg, Germany
First Author:
Co-Author(s):
Introduction:
In multicultural societies, people encounter individuals from their own social group (ingroup members) and from different social groups (outgroup members), and form impressions towards these ingroup and outgroup individuals. As impressions predict behaviors, it is important to understand the mechanisms that shape impressions in intergroup contexts.
One basic mechanism that shapes impressions is learning. According to learning theory, positive experiences with a person establish positive associations towards this individual, while negative experiences have the opposite effect, driven by unpredicted positive or negative experiences that elicit prediction errors. Importantly, in complex social environments, individuals learn from intermixed experiences with both ingroup and outgroup members. However, it remains unclear how such ingroup and outgroup experiences dynamically shape neural learning processes and learning-related changes in impressions towards in- and outgroups.
To address this question, we designed an experiment in which identical experiences with the in- and outgroup could dynamically affect the impressions towards both groups.
One basic mechanism that shapes impressions is learning. According to learning theory, positive experiences with a person establish positive associations towards this individual, while negative experiences have the opposite effect, driven by unpredicted positive or negative experiences that elicit prediction errors. Importantly, in complex social environments, individuals learn from intermixed experiences with both ingroup and outgroup members. However, it remains unclear how such ingroup and outgroup experiences dynamically shape neural learning processes and learning-related changes in impressions towards in- and outgroups.
To address this question, we designed an experiment in which identical experiences with the in- and outgroup could dynamically affect the impressions towards both groups.
Methods:
30 healthy swiss males volunteered to take part in this fMRI study. Participants were about to receive shock in each trial while the individual from the ingroup or the outgroup ostensibly could give up money to save the participant from pain.
At the beginning of each trial, the participants rated their closeness toward the ingroup and outgroup on separate rating scales that were presented in randomized order. Next, participants rated their expectancy of receiving shock, followed by the symbol that represented the ostensible decision of the other person. In reality, participants were relieved from pain in 75% both from ingroup and outgroup individuals. Before and after learning, we used an impression scale to assess participants' impressions towards ingroup (Swiss) and outgroup (Middle Eastern) individuals.
We modeled participants' trial-by-trial expectancy ratings using a standard Rescorla-Wagner (Rescorla & Wagner, 1972) reinforcement learning (RL) algorithm. We then fitted the trial-by-trial closeness ratings regarding the ingroup or outgroup as a linear function of previous prediction errors derived from the reinforcement learning models.
At the beginning of each trial, the participants rated their closeness toward the ingroup and outgroup on separate rating scales that were presented in randomized order. Next, participants rated their expectancy of receiving shock, followed by the symbol that represented the ostensible decision of the other person. In reality, participants were relieved from pain in 75% both from ingroup and outgroup individuals. Before and after learning, we used an impression scale to assess participants' impressions towards ingroup (Swiss) and outgroup (Middle Eastern) individuals.
We modeled participants' trial-by-trial expectancy ratings using a standard Rescorla-Wagner (Rescorla & Wagner, 1972) reinforcement learning (RL) algorithm. We then fitted the trial-by-trial closeness ratings regarding the ingroup or outgroup as a linear function of previous prediction errors derived from the reinforcement learning models.
Results:
We first found a significant group × time interaction (χ2(1) = 7.34, p = 0.007) in impression rating, showing the reduction of ingroup favoritism in the course of learning. Next, we built up computational models to test how ingroup and outgroup prediction errors affected trial-by-trial changes in ingroup and outgroup closeness ratings. In our models, we assumed that changes in closeness to group i (i.e., the ingroup or the outgroup) were a linear function of the time-discounted sum of previous prediction errors to outcomes from i. Having built up the respective models that capture the change of closeness towards the in- and the outgroup, we extracted the model parameters and associated the parameters with the change of intergroup impressions.
The results revealed that the reduction of the ingroup bias in impression ratings was uniquely predicted by the weight given to ingroup prediction errors (β = 0.74, SE = 0.29, t = 2.56, p = 0.017). Importantly, the more strongly an individual weighted ingroup prediction errors, the more pronounced the reduction of ingroup bias in impression ratings was.
Neurally, the individual weight for ingroup prediction errors was related to the coupling between the left inferior parietal lobule (IPL) , the region specifically encoding the negative prediction errors from the ingroup members, and the left anterior insula (AI), which, in turn, predicted learning-related changes in intergroup impressions.
The results revealed that the reduction of the ingroup bias in impression ratings was uniquely predicted by the weight given to ingroup prediction errors (β = 0.74, SE = 0.29, t = 2.56, p = 0.017). Importantly, the more strongly an individual weighted ingroup prediction errors, the more pronounced the reduction of ingroup bias in impression ratings was.
Neurally, the individual weight for ingroup prediction errors was related to the coupling between the left inferior parietal lobule (IPL) , the region specifically encoding the negative prediction errors from the ingroup members, and the left anterior insula (AI), which, in turn, predicted learning-related changes in intergroup impressions.
Conclusions:
Our findings provide computational and neural evidence for ingroup-focused theories, highlighting the importance of ingroup experiences in shaping social impressions in intergroup settings.
Emotion, Motivation and Social Neuroscience:
Emotional Learning 2
Social Interaction 1
Social Neuroscience Other
Modeling and Analysis Methods:
Activation (eg. BOLD task-fMRI)
Novel Imaging Acquisition Methods:
BOLD fMRI
Keywords:
Learning
Social Interactions
Other - intergroup relations
1|2Indicates the priority used for review
Provide references using author date format
Rescorla, R. A., & Wagner, A. R. (1972). A theory of Pavlovian conditioning: Variations in the effectiveness of reinforcement and nonreinforcement. In A. H. Black & W. F. Prokasy (Eds.), Classical Conditioning II: Current Research and Theory (pp. 64-99). Appleton-Century-Crofts
Hein, G., Engelmann, J. B., Vollberg, M. C., & Tobler, P. N. (2016). How learning shapes the empathic brain. Proc. Natl Acad. Sci. USA, 113, 80-85