Social Reality Construction via Active Inference: Modeling the Dialectic of Conformity and Creativity
Researchers use active inference to show how conformity and creativity shape our social reality.
Researchers Kentaro Nomura and Takato Horii have published a novel computational model that simulates how social groups collectively construct shared beliefs and cultural norms. Using a multi-agent system grounded in active inference—a theoretical framework where intelligent agents act to minimize prediction error—the study formalizes the dialectical tension between conformity and creativity. Each agent maintains an internal generative model of the world, communicates with neighbors to form social priors, generates novel observations, and selectively incorporates others' creations into its memory. This creates a dynamic, bidirectional process where social reality is both internalized and reshaped.
Simulation experiments on structured social networks revealed three key phenomena. First, informationally cohesive social groups emerged endogenously, with agents' internal representations aligning according to the underlying network's cluster topology. Second, a circular relationship formed between social representations and the distribution of observable creations, maintained through agents' creative acts that project their internal structures onto the external world. Third, the propagation of novel creations followed selective, heterogeneous patterns distinct from the stable diffusion of shared beliefs, indicating that agents construct distinct cultural niches through local interaction dynamics. The work, submitted to the ALIFE 2026 conference, provides a unified computational framework for understanding the emergence of social reality.
- The model uses active inference to simulate how multi-agent systems form shared beliefs through local network interactions.
- Simulations show endogenous emergence of cohesive belief clusters that mirror the underlying social network's topology.
- Creative acts by agents project representational structure onto the world, creating a feedback loop that shapes cultural niches.
Why It Matters
This research bridges AI and social science, offering tools to model cultural evolution, misinformation spread, and organizational dynamics.