Generative Models Simulate Attitude Change Theories
A new study introduces a generative actor-based modeling workflow that renders classic attitude change theories—cognitive dissonance (Festinger 1957), self-consistency (Aronson 1969), and self-perception (Bem 1972)—as runnable simulations. Using the Concordia library, actors operate by predictive pattern completion on natural language strings, generating actions from memory and observation prefixes. This approach transforms verbal mechanism sketches into executable systems, enabling computational testing of competing theories. The work is published on arXiv (2604.19791) and represents a methodological advance in computational social psychology.
Key facts
- arXiv paper 2604.19791 presents generative actor-based modeling for attitude change theories.
- Workflow uses Concordia simulation library to render theories as runnable actor-environment simulations.
- Actors operate by predictive pattern completion on natural language strings.
- Theories modeled: cognitive dissonance (Festinger 1957), self-consistency (Aronson 1969), self-perception (Bem 1972).
- Each theory is implemented as distinct decision logic.
- Approach addresses lack of technical specifications in verbal theories.
- Simulations generate action suffixes from memory and observation prefixes.
- Published as arXiv preprint with announcement type 'new'.
Entities
Institutions
- arXiv
- Concordia