Go your own way: the success of the group depends on it

Scientists from the Weizmann Institute have shown that groups may act in the best way for them when each individual tries to act as differently as possible from the others

How do memes or videos suddenly become viral on social networks? These are all well-known contemporary examples of collective behavior that characterizes communities of humans, just like schools of fish, colonies of bacteria, or networks of neurons. study recently conducted at the Weizmann Institute of Science adds a new layer to the understanding of the relationship between collective behavior and individual behavior, and shows that groups may act in the best way for them when the individuals in the group act in their own way.

This surprising finding stems from the research of research student Ehud Karpas, Prof. Elad Schneidman, and the research student at the time, Adi Shkelresh, in the neurobiology department of the institute. The researchers created mathematical models to answer the question: How does collective behavior result from the decisions of many individuals? These models may constitute new frameworks for understanding collective behavior among groups in the real world.

The research focused on models simulating how long it would take for an individual to locate a "weak" source in a "noisy" environment, similar to an animal searching for food in a forest. The starting point of the research is in a model developed about a decade ago by biophysicists from the University of California in San Diego and Santa Barbara. They showed that searching for information relative to the target (instead of directly addressing its likely location) is a better method in noisy environments, and argued that this is what animals in the wild actually do. This idea is called "infotaxis" (from the Greek word taxis, which means movement in response to a stimulus). These models predict that in their search for information, animals will follow a spiral path (similar to the paths observed in nature), which will allow them to collect evidence and test their estimates regarding the location of food during the search.

Two details "going their own way" 

In the study thatRecently published In "Proceedings of the National Academy of Sciences of the United States of America" (PNAS), Karps, Shklaresh and Prof. Schneidman took the infotaxis model one step further, and asked: How will it apply to groups of individuals? "In a situation like this," says Karps, "each individual can see or hear the other individuals, and use that to get important information."

"If a rabbit who wants to eat a carrot sees another rabbit, he can assume that he thinks and behaves like him. When an unexpected change in the other rabbit's behavior occurs, the first rabbit can conclude that he has discovered a patch of carrots or that he is running away from a madman."

The researchers started building models for two situations - in one, the individuals searching simply ignore each other, and in the other, all the information is shared and decisions are made together. Intuitively, sharing information confers an advantage, and this is indeed what is found in the research. But such sharing of information and decision-making is unlikely in nature. Thus, the scientists looked for an intermediate state, where some of the information is shared, and individuals weigh it with what they already know. This weighted average, Schneidman says, is the method used in many models in physics and biology to deal with noisy signals and uncertainty, as it reduces the noise. However, the middle way did not really help the details in the foraging site model. "It was a serious failure," says Prof. Schneidman. "In most cases, the individuals 'circled' each other, and did not reach the food at all."

The team therefore considered the opposite scenario, where each individual tries to act as differently as possible from the others. Karpas explains: "Basically, we told each 'animal' to solve the problem by thinking differently from the rest. It worked surprisingly well." Adds Prof. Schneidman: "The efficiency of the search was similar to the situation where the entire group had access to all the information, and decisions were made jointly. In fact, the more the individuals made sure to be different from the others, the better the model worked." The reason this model was so effective is that diversity may help the animals cover the area more efficiently.

A simulation of two individuals weighting information from each other 

The scientists then used their model, which they called socialtaxis, to ask questions about the nature of the information that one group member receives from other members. For example, what happens if each animal shares only one reliable piece of information? The model performed even better with less information - presumably because this reduced the noise in the information passed between individuals.

Even without any intentional communication between the animals, the model showed that their very behavior embodies information that others can use. For example, if one individual does something surprising - for example, turns in a new direction - the others can interpret this change in behavior as information. "If a rabbit that wants to eat a carrot sees another rabbit," says Schneidman, "he can assume that he thinks and behaves like him. When an unexpected change in the other rabbit's behavior occurs, the first rabbit can conclude that he has discovered a patch of carrots or that he is running away from a madman." This model worked quite well, although not as well as the one where the information was shared through direct lines of communication between the animals.

Says Prof. Schneidman: "Instead of formulating the question of collective behavior in terms of influence or dependence depending on distance, we can ask what the other members of the group tell me about the world; how, through them, I can improve my knowledge of the world. The socialtaxis model links together individual and group behavior, and indicates that very successful groups can be composed of individuals who 'go their own way'". The model can be applied to many areas - from hoes and schools of fish to antibiotic resistance, which increases as the diversity of bacteria increases. When it comes to group behavior, politics and policy making are good examples, but the model may also apply to more everyday areas, such as problem solving by scientists. For example, when two people approach the same problem, they may use the information they shared to seek separate and different approaches to the solution.

Large groups illustrate the benefits of socialtaxis