
informational influence, not directly addressed to the question of nor-
mative behavior. The algorithm, though, can be seen to consist of two as-
pects: a hard-coded set of behaviors and a flexible process of discovery.
Every particle swarm program tells the particles to maintain a memory of
their previous successes, to evaluate the relative achievements of their
neighbors, and to adjust their trajectories toward previous successes;
their social and cognitive propensities are specified in computer code.
No particle swarm program tells the particles where the optimum is, or
which particular neighbors will have found the best so far, and nothing
tells them ...