Reilly Raab

UC Santa Cruz

“The Emergence of Prejudiced Social Norms”

We formalize a simple game of pairwise social coordination informed by a partner’s features, and consider the dynamics of this repeated game in a large population. We rely on numerical experiments and a novel synthesis of machine learning and evolutionary game theory, and our results are highly relevant to the emergence of prejudiced social norms and the construction of “groups” such as race or gender.


We show that self-reinforcing social divisions and prejudiced social norms can result from the inductive biases of human pattern-matching: First, we introduce a simple game in which pairs of individuals must coordinate actions (such as whether to shake hands or bump fists) after observing each other’s features. Next, when individuals are simulated with simple machine-learning algorithms, the population self-organizes, and individuals with similar features tend to learn similar policies for how to coordinate with others, depending on the structure of the social network, distribution of features, and the agents’ tolerance for complexity. Our analysis relies on a novel synthesis of evolutionary game theory, tools from machine learning, and Bayesian inference. Finally, we relate our findings to the construction of “groups”, such as race or gender, with boundaries that agents may disagree on, even when everyone in a population is able to successfully navigate social norms without error.

5 + 7 =