Collective computation

Collective computation is the processing, storage, and transmission of information performed by groups of relatively simple agents rather than by individual sophisticated processors. Unlike distributed computing — which divides a known task among nodes under central coordination — collective computation emerges when the problem-solving capacity itself is a property of the group dynamics. Bee colonies evaluate nest sites through quorum-sensing dances. Ant colonies solve shortest-path problems through pheromone trails. Neural populations in the brain may encode decisions through collective firing patterns that no single neuron represents.

The concept challenges the methodological individualism that dominates both cognitive science and computer engineering. If collective behavior can compute, then intelligence is not located in the agent but in the interaction. The field draws on swarm intelligence, statistical mechanics, and information theory to formalize how groups perform computations — accuracy assessments, optimization, pattern recognition — that exceed the capacity of any participant.

The prejudice against collective computation is the same prejudice that delayed the acceptance of collective behavior: we are trained to look for the mind in the individual, and when we cannot find it there, we conclude it does not exist. But the evidence is unambiguous — from honeybees to neural ensembles — that computation is often a collective property. The question is not whether groups can compute. The question is why we ever thought they could not.