Original title: Social Discounting Enables Fast and Reliable Collective Escape
Zachary P Kilpatrick · August 2026
Solitary animals face a tough tradeoff when watching for predators: detecting danger faster means suffering more false alarms. Researchers modeled animals in groups as noisy evidence-accumulators that run away when their belief passes a set thresholdThe level of belief at which an animal decides it is dangerous enough to flee.. In groups, an animal's escape signals danger, but an undisturbed neighbor's stillness signals safety. By treating calm neighbors as evidence that things are safe—a strategy called social discountingThe rate at which an animal ignores the lack of alarm from a neighbor as evidence of safety.—groups break the tradeoff and become both faster and more accurate than lone animals. Naive animals that only copy escapes end up triggering more false alarms as group size grows. The team found that discounting lets safety cascades stay low while real threat cascades spread rapidly. Looking at wild sulphur molly shoals attacked by birds, the fish discounted threats at very high rates, which allows their false-alarm rate to remain constant even as shoals grow.
You are at a party and need to know if it is time to leave because something bad is happening. Normally, you look for clear signs like a fire alarm. But if you only react when someone runs for the exit, you might panic over nothing—like if someone just leaves early. Now suppose you also pay attention when people stay calm: if your friend is still chatting, you relax a bit. In this paper, animals use the stillness of neighbors as evidence of safety, which lets the group make better decisions than any single animal could.
When animals flee together, each extra pair of eyes should help—but only if the group can avoid false alarms. This work shows that by actively trusting a calm neighbor, groups escape threats faster and with fewer mistakes. The finding explains why some shoals and flocks stay coordinated even under attack, and offers a way to predict group false-alarm rates from individual behavior alone.
The abstract doesn't explicitly state limitations, but the model is validated only on sulphur molly shoals under bird attack. Generalizability to other species or threat types is untested. The social discounting approximation may not capture all complexities of group decision-making in different contexts.
Ask anything about this paper. Answers come from its abstract, and say so when the abstract doesn't cover your question.
Social Discounting Enables Fast and Reliable Collective Escape — Zachary P Kilpatrick
Abstract licence: CC0
Explanation generated by Synthesis from the abstract — it is not the original paper. This is a preprint, so it hasn't necessarily been peer reviewed.