Fish have been swimming in schools for millions of years. This has many advantages for them: if you are surrounded by congeners there is a better chance that you yourself will not be caught by predators, and it is well known that the eyes of many see more of them.
“We observe freshwater fish, small species that can easily be kept and reared in the laboratory,” explains Jens Krause, professor at the Leibniz Institute for Freshwater Ecology and Inland Fisheries in Berlin (IGB) . “You can make swarms of them. Then we see how the performance of the swarm grows with its size. “
Smarter in swarm than alone
The biggest achievement of the swarm is protection from predators. However, in order for an orderly mass movement to take place, each fish must follow a few rules, for example, always orient itself towards its direct neighbors and at the same time maintain a constant distance from them. Jens Krause uses cameras to observe the behavior when the swarm is confronted with a replica of a predator.
“You can see that there is only a 55 to 60 percent chance that the individual fish will find the predator and then swim in the safe direction. But if it is a school of 16 fish , there is an 85 to 90 percent chance that they will do it. A huge increase in the choice of the safe route, and the fish do it much faster when they are in the group. “
But why does fish work so well in schools? Could it be due to a lack of individuality? David Bierbach, also from IGB, observed the Amazon Molly self-cloning fish species, in which all fish have the same DNA. He separated individuals and allowed them to grow up under exactly the same conditions.
Result: “They had the same genes, the same environment, and yet they all differ in their behavior. So there may well be personality in fish.
New properties through cooperation
According to Jens Krause, each fish as an individual contributes to the intelligence of the swarm. In the swarm there is what is called the emergence – new properties resulting from the interaction of the elements. “Each fish has different cognitive possibilities. But we see that these possibilities are extremely extensive when the fish is already with one or two others. ”
This explains why the formation of swarms and groups in the animal kingdom is so successful. “It will certainly be one of the factors that people are often in a group.”
Swarm mechanisms in humans
Ralf Kurvers, group leader “Collective Intelligence” at the Max Planck Institute for Human Development, is studying whether similar swarm mechanisms can be found in humans. In a conference, he organizes an experiment with the audience.
One image shows around seventy images of a tuna – and a few images of sharks that are difficult to distinguish. Visitors only see the image for two seconds and then must decide whether to see fewer or more than four sharks – so whether to run away or stay like tunas. In a short time, you can only guess the number of sharks, but you need to make up your mind quickly.
Ten say flee, 37 say stay. The result is communicated to the group, after which everyone can reconsider their decision. Most of them don’t at first.
Swarms are hard to fool
In the second round, however, the information about the behavior of the group prompted some to adjust their opinion in the revision towards the majority, which in this case gave the correct answer.
“We usually do 30-40 of these laps and notice a strong learning effect along the way,” says Ralf Kurvers. “At first people are reluctant to trust social information, but then they realize it’s useful and start following the majority. When we look at the results, we see that individuals with social information are much more likely to escape. After receiving the information, they were also able to filter out the bad decisions made by the majority. The bigger the group, the better it worked. “
Swarms are hard to fool. Jens Krause also observes it in IGB aquariums. If, for example, it lets out a robot fish integrated into the swarm, the swarm does not automatically react by fleeing as well, but only when there is a precisely balanced number of refugees.
“And this threshold is slightly higher than the normal rate for an individual to be mistaken in a swarm. This means that if 20 percent of the fish are often wrong, then the fish will only react when more than 20 percent want to swim away. This would then be an indication that the percentage that wants to leave is higher than the spontaneous error rate in this group. This is a relatively ingenious procedure ”, on which an algorithm is now based, which is supposed to make it possible to rule out medical errors when diagnosing breast cancer.
But swarm intelligence cannot be used everywhere. Attempts on the stock market, for example, with funds that rely on the intelligence of many investors, have failed. The reason: there were agreements between individual investors. But in order to be able to use the wisdom of the swarm, it is important that each individual does not know what the others are deciding.
Small swarm of intelligence on the Internet
Even the internet doesn’t automatically lead to more swarm intelligence, says Pawel Romanczuk, who studies network structures at Humboldt University in Berlin. Social media algorithms, for example, were primarily aimed at generating subscribers for certain content, for example by targeting users with emotional messages.
It’s not about increasing collective well-being – or improving collective decision-making skills. “The structure of the network does the exact opposite of what the match does for the fish. This completely hinders consensus building and makes it much more difficult for a society as a whole to work towards a common goal. “