Information Migration in Multi-Agent Systems: Recovering Hidden Sources from Trajectory Histories After Signal Loss

This paper investigates whether information can remain recoverable after the original signal that generated it has disappeared. Using a simplified multi-agent simulation, a hidden source was allowed to influence agent behavior before being removed from the environment. After removal of both the source and much of the associated field structure, reconstruction was attempted using only recorded agent trajectory histories. Results show that directional records from a passive observer network remained sufficient to localize the original source with high accuracy despite substantial loss of direct field information. The study introduces the concept of information migration: the transfer of recoverable information from one representation (a scalar field) to another (agent trajectories). The work does not claim biological or clinical applicability. Instead, it presents a computational proof-of-concept demonstrating that hidden causes may remain reconstructable through distributed response patterns even after the original signal is no longer directly observable. Potential applications include source localization, swarm intelligence, distributed sensing, inverse problems, fault reconstruction, and complex adaptive systems.