Skip to main content
Back

Researchers Build a Unified AI Model Where Creativity Shapes Social Norms

Kentaro Nomura and Takato Horii publish a multi-agent active inference model showing how social norms and creative acts co-construct shared reality

Pro

Paper, Full Access
  • Kentaro Nomura and Takato Horii published a preprint modeling how social norms and creative acts jointly construct shared reality.
  • The paper was accepted for an oral presentation at ALIFE 2026 in Waterloo, Canada (August 17-21).
  • Each agent runs active inference: building social priors from neighbors, generating creative acts, and selectively absorbing others' creations.
  • Key finding: informational clusters emerge spontaneously, mirroring the underlying network topology without explicit coordination.
  • Creative artifacts spread selectively and heterogeneously, unlike norms which diffuse stably -- producing cultural niches within a single network.
  • No code has been released; the model is currently limited to 14 agents on a synthetic network.

What makes a social norm stick? And what makes a creative act spread? These two questions have long lived in separate silos -- social psychology on one side, cultural evolution on the other. A new preprint by Kentaro Nomura and Takato Horii takes a unified computational swing at both, using active inference to model how conformity and creativity jointly construct social reality from the ground up.

The paper, posted to arXiv and announced by co-author Takato Horii after it was accepted for an oral presentation slot at ALIFE 2026, sits at the intersection of multi-agent systems, cognitive science, and computational social science. The work is led by Nomura, an M2 (master's second-year) student, with Horii as supervising co-author.

The gap they are filling

Social agents both internalize collective norms and reshape them through creative action, yet computational models have not captured this bidirectional process within a unified framework. Most prior work treats agents as passive observers: they receive signals from their environment and update their beliefs, but they do not push back. Agents in these frameworks are passive observers that infer representations from observations they receive, leaving the feedback pathway from individual creative acts to social representation formation unaddressed. In the present work, social representations are instead placed inside each agent and allowed to differ across individuals, enabling representational alignment to emerge naturally from local interaction. Agents are also endowed with the capacity for active creation, closing the feedback loop between individual creativity and collective norm formation.

Active inference, briefly