Emotional Modulation in Swarm Decision Dynamics
Agent-based model uses emotional valence and arousal to modulate swarm behavior, creating tipping points.
A new research paper by David Freire-Obregón, accepted for ICAART 2026, introduces a novel framework that merges classical swarm intelligence with emotional modeling. The work extends the established 'bee equation'—used to describe honeybee nest-site selection—into an agent-based model where simulated agents possess emotional states defined by valence (positive/negative) and arousal (low/high). These emotional states directly modulate the agents' interaction rules, changing their rates of recruiting others to an option or inhibiting alternative choices. Agents even display simulated facial expressions mapped from their emotional state, allowing researchers to study emotional contagion within the decision-making process.
The study explores three key scenarios: how the combination of valence and arousal affects consensus outcomes and speed, how high arousal can break decision deadlocks when valence is neutral, and the 'snowball effect' where consensus rapidly accelerates after surpassing a critical support threshold. Results demonstrate that even small emotional asymmetries within a group can significantly bias collective decisions and alter convergence times by shifting the effective parameters of the swarm's decision-making engine. Interestingly, the model also shows that intrinsic non-linear dynamics can produce decisive outcomes even in perfectly symmetric emotional conditions, highlighting the complex interplay between emotional modulation and structural tipping points.
This framework provides a flexible tool for understanding the emotional dimensions of collective choice, applicable to both natural systems and artificial ones. For AI developers, it suggests pathways for designing multi-agent systems (swarms or agentic workflows) where simulated emotional states could be used to steer group decisions, manage consensus formation, or model social dynamics more realistically. It bridges a gap between cold, mathematical swarm theory and the messy, affective reality of how groups—whether biological, human, or artificial—actually make choices.
- Extends the classic 'bee equation' for swarm decisions into an agent-based model where emotional valence and arousal modulate interaction rates.
- Shows emotional asymmetries can bias group outcomes and create 'snowball effects' where consensus accelerates after passing 50% support thresholds.
- Provides a framework for designing AI multi-agent systems where simulated emotions can steer collective decisions and model social dynamics.
Why It Matters
This research could lead to more sophisticated and human-like collective decision-making in AI agent swarms and multi-agent systems.