Â鶹ÒùÔº

When predators strike, fish need to decide fast: dive or stay? It's a life-or-death choice, and getting it wrong, whether by overreacting to harmless disturbances or missing a real threat, can be costly.

Now, researchers from the Cluster of Excellence Science of Intelligence (SCIoI) with Humboldt-Universität zu Berlin and the Leibniz Institute of Freshwater Ecology and Inland Fisheries (IGB), have shown that wild fish shoals become not just faster, but better at making these decisions as they grow in size. The findings, in Science Advances under the title "Better and faster decisions by larger fish shoals in the wild," offer the first field-based evidence that large animal collectives can overcome two classic trade-offs: detecting real threats without reacting to every shadow, and making fast decisions without sacrificing accuracy.

The key insight? Large shoals of sulfur mollies (Poecilia sulphuraria) are able to distinguish real bird attacks from harmless flybys with increasing accuracy and speed, something that has proved exceptionally hard to demonstrate especially under natural conditions.

"We've known that animal groups can make impressive collective decisions," says lead researcher Korbinian Pacher, "but theoretical and lab approaches can only bring us so far."

"What we wanted to know is whether collective intelligence is present where it really matters, under messy, noisy, real-life conditions. To investigate collective decisions under circumstances where a wrong decision has real consequences is almost impossible in the lab, which makes taking this back to the field all the more important," adds Jens Krause, senior author of the study.

Split-second decisions, thousands of fish

The team studied shoals of sulfur mollies living in the steamy, sulfurous waters of the El Azufre river in Tabasco, Mexico. These fish survive in extreme conditions: low oxygen, high temperatures, and the constant threat of aerial predators like kingfishers and great kiskadees.

What makes these shoals particularly interesting is their unusual anti-predator strategy: when a threat (such as a bird) is detected, the fish perform a collective dive, creating visible ripples across the water surface. If the fish decide that the disturbance is an actual attack, they continue with repeated wave-like dives for up to several minutes. If not, they stop.

This behavior gave researchers a rare opportunity: a window into the moment in which a prey group decides how dangerous a disturbance really is.

Discerning danger, collectively

The challenge for a single fish is obvious: in order to never miss a predator, you might end up diving for every passing shadow. But groups have options. In theory, they can combine information from multiple individuals to detect real threats more reliably. Still, whether such collective cognition works in the chaos of the wild has remained unclear—until now.

In over two hundred documented events, the researchers compared responses to real predator attacks and harmless bird flybys. They focused on one particularly sneaky predator: the great kiskadee. Unlike splashy kingfisher dives, kiskadees perform near-silent overflight attacks where only their beak skims the water, visually similar to harmless disturbances like rustling branches or other bird movements.

The results were striking: larger shoals were much better at telling the difference between these ambiguous attacks and false alarms. Their responses to true threats increased with shoal size, while responses to harmless flybys stayed constant. That means bigger shoals did not just become more sensitive, but more discerning. An essential improvement in decision quality.

Smarter groups, not just bigger ones

In detection theory, decision-making often involves a trade-off: act fast, and you might make more mistakes. Wait longer, and you risk missing your chance. But the fish shoals in this study didn't just get more accurate, they also got faster. Larger groups took less time between the first sign of a disturbance and the collective choice to continue evasive action.

This ability to break two fundamental trade-offs between true and false positives, and between speed and accuracy is what sets these findings apart. "In the biggest shoals, detection rates were nearly perfect—almost 100% of kiskadee attacks were identified correctly," says Korbinian. "That level of performance would be impossible for a single fish."

From molly shoals to human crowds

While previous models have explained group decision-making using ideas like quorum rules, where individuals only act once a threshold number of neighbors do, the sheer scale of these shoals (tens to hundreds of thousands of individuals) makes such explanations harder to apply. It's unlikely that every fish is watching every other one. Instead, the researchers suspect a more complex, self-organized process might be at play.

"We're starting to think of these fish shoals like neural networks," says Korbinian. "They might operate at something called criticality: a state that optimizes information flow in large systems, from brains to crowds."

Understanding how large animal collectives process information so effectively could inspire future models in both biology and artificial intelligence. It also helps answer a core evolutionary question: Why do animals live in groups in the first place?

A window into natural intelligence

This study adds to a growing body of evidence that animal collectives can act as more than the sum of their parts, especially under pressure. By cleverly combining individual inputs into fast, accurate decisions, groups like these sulfur mollies offer insights into the evolution of intelligence itself.

"For me, the most exciting part is that we caught a glimpse of real, wild collective cognition in action," says Korbinian. These fish are solving a genuinely hard cognitive problem together, and they're doing it better than we ever imagined."