Adaptive Self-Preservation in Flocking Systems: How Dynamic Risk Thresholds Shape Collective Safety

Jiaman Li, William G. Kennedy

Keywords: Adaptive self-preservation, Flocking, Emergent heterogeneity, Phase transitions, Agent-based modeling

Abstract:

Can simple local experiences, such as a near-collision, shape how risk is distributed within a group, even when agents have no memory, communication, or global coordination? Traditional flocking models assume fixed behavioral parameters, yet real organisms often become
more cautious after close encounters and relax only gradually when conditions feel safe. We formalize a memoryless threshold adaptation rule where agents increase their collision threshold by α following a near-miss but decay it by β during safety, with the α/β ratio acting as a control parameter. Through systematic experiments on NetLogo’s flocking model (N = 540 runs), we identify three distinct behavioral regimes determined by α/β: homogeneous risk-aversion (α/β ≥ 10), stable differentiated equilibrium (α/β ≈ 5), and collapse to risk-neutrality (α/β ≤ 2.5). In particular, effective self-organization arises primarily within a narrow parameter window, where identical agents develop heterogeneous risk tolerances through interaction alone. This reveals a fragility in adaptive decentralized systems: simple local rules can generate emergent safety structures, with maximum heterogeneity (SD = 0.71) at regime boundaries, but stable differentiation occurs only within tightly bounded parameter regimes, challenging the assumption that minimal mechanisms guarantee robustness.

Email Sharing: We authorize the sharing of all author emails with Program Chairs.

Data Release: We authorize the release of our submission and author names to the public in the event of acceptance.

Submission Number: 15