Figure 1.

A minimal collective system of two AIF agents (adapted from McGregor et al.). We implement two agents (Agent A and Agent B) that have one common target position (Shared Target) and one individual target position (A’s Target; B’s Target). All targets are encoded with equal desirability. This figure is notional: our simulation environment contains 60 cells instead of the 12 depicted here. Note: we randomize the location of the shared target while preserving relative distances to unshared targets to ensure that the agents’ behavior is not an artefact of its location in the sensory environment.