When a bird of prey attacks, every second counts for fish: dive or stay? A wrong decision can be fatal - either because it comes too late or because a threat is mistaken for a harmless noise. A study by researchers from the Science of Intelligence Cluster of Excellence (SCIoI), Humboldt-Universit?t zu Berlin (HU) and the Leibniz-Institute of Freshwater Ecology and Inland Fisheries (IGB) now shows that larger schools of fish not only make such decisions faster, but that they can also make better decisions. The results, which have just been published in the renowned journal Science Advances, provide the first evidence to date under natural conditions that large groups of animals can overcome two classic trade-offs: recognising real dangers without reacting to every disturbance and making quick decisions without losing accuracy.
Lightning-fast decisions with thousands of participants
For their study, behavioural biologists Korbinian Pacher from IGB and Prof. Dr Jens Krause from the Thaer-Institute of Agricultural and Horticultural Sciences and other colleagues from the HU examined schools of sulphur fish (Poecilia sulphuraria) that live in the hot, sulphurous springs of the El Azufre River in the Mexican state of Tabasco. The conditions there are extreme: A lack of oxygen, high temperatures and birds of prey such as kingfishers or kiskadees, which regularly hunt the fish. The collective defence strategy of the sulphur fish is particularly interesting: as soon as a school perceives a potential danger, the fish dive down in synchrony and create visible wave patterns on the surface of the water - like la-ola waves. If it is actually an attack, a series of further "wave dives" follow. If the stimulus is harmless, it remains a one-off dive. For the researchers, this behaviour was a stroke of luck - because for the first time it allows a precise view of how animal groups make decisions about potential dangers.
Recognising danger – as a team
With over two hundred documented events, the researchers compared the behaviour of the flocks during real attacks and harmless bird flights. Their focus was on the kiskadee, a predator that is particularly difficult to recognise: instead of diving in loudly, it attacks in flight, with only its beak briefly touching the water, which can hardly be distinguished visually from the harmless movements of other birds. The result: larger flocks were significantly better at distinguishing between real danger and false alarms. While the reactions to real threats increased with group size, the reactions to harmless stimuli remained constant. The swarms therefore did not become more sensitive, but more precise. A real increase in the quality of decision-making.
"We have known for a long time that animal groups can make impressive collective decisions," says study leader Korbinian Pacher, PhD student at IGB. "But theoretical models and laboratory experiments only take us so far in researching this phenomenon. We were interested in whether collective intelligence also works where it really counts, namely under chaotic, noisy, real environmental conditions."
"Investigating collective decision-making behaviour under conditions where a wrong decision has real consequences is hardly possible in the laboratory, which is precisely why it was so important to take this question back into the field," adds Jens Krause, another author of the study and professor of fish biology and ecology at the HU and head of the "Biology of Fish, Fisheries and Aquaculture" department at the IGB.
Not only bigger, but also smarter
Decision theory is often based on a dilemma: Those who react quickly make more mistakes. If you hesitate too long, you miss the chance to escape. But the swarms of sulphur fish not only became more accurate, they also became faster. The larger the shoal, the shorter the time between the first dive and the collective decision to continue the defence. "In the largest swarms, the detection rates were almost perfect, with almost 100 per cent of kiskadee attacks being correctly identified," says Korbinian Pacher. "That would simply be impossible for a single fish."
From schools of fish to crowds of people
Previous models often explain group decisions with so-called quorum rules: An animal only reacts if a certain number of conspecifics also react. However, in schools of tens of thousands or even hundreds of thousands of fish, it is unlikely that each animal observes all the others. Instead, the researchers suspect a self-organised, more complex mechanism. "You can visualise these schools of fish almost like a neural network," says Pacher. "They could operate in a state that we call 'criticality' - a state that optimises information processing in large systems such as the brain or even crowds of people." A better understanding of such group processes could inspire not only biological but also artificial systems: from robotics to swarm intelligence. And it helps to answer one of the most fundamental questions in evolutionary biology: Why do animals live in groups in the first place?
A window into natural intelligence
The study provides convincing evidence that animal groups are more than the sum of their parts under real-life conditions. By bringing together individual information quickly and precisely, swarms like the sulphur molly show how collective intelligence works in nature and pays off in the form of an evolutionary survival advantage. "For me, the most fascinating thing is that we were able to observe real collective cognition in the wild here," says Pacher. "These fish solve a damn difficult problem together - and they do it better than we would have thought possible."
Further information
Contact
Korbinian Pacher
Leibniz-Institute of Freshwater Ecology and Inland Fisheries (IGB)/Cluster of Excellence Science of Intelligence (SCIoI)
Phone: 0171 1198774
korbinian.pacher? Please insert an @ at this point ?gmail? Please insert a period at this pointcom
Prof. Dr Jens Krause
Leibniz-Institute of Freshwater Ecology and Inland Fisheries (IGB)/Cluster of Excellence Science of Intelligence (SCIoI)
Tel.: 030 64181 610
Email: jens.krause? Please insert an @ at this point ?igb-berlin? Please insert a period at this pointde