Life can be hard for a fish. That is why many species don’t go it alone. Instead, they group together in schools that can range from a dozen individuals to millions. Once they’re formed, these groups become like superorganisms with collective behavior that responds to threats and stimuli. Schools can even develop new senses that no one individual fish possesses.
How these complex behaviors and patterns emerge in groups – from schools of fish to flocks of birds and even crowds of people – remains mysterious, but researchers such as Iain Couzin are shedding some light on the phenomenon. Couzin is the director of the Max Planck Institute for Ornithology in the Department of Collective Behavior at the University of Konstanz in Germany. Through machine learning, Couzin and his team can track individual fish in a school, and study how much influence individuals have over the collective behavior of the group. This knowledge, says Couzin, is more than just a scientific curiosity. It can help with marine conservation and fisheries – and even help us understand human society.
Recently, Spine Films created a video for the magazine bioGraphic about Couzin’s latest work. To find out more, Oceans Deeply spoke with Couzin about studying schooling fish and the future of collective behavior research.
Oceans Deeply: You say in the video that we’re still bad at predicting this kind of group behavior. Why?
Iain Couzin: I think that [these groups] are so mysterious because they exhibit this incredibly high degree of behavioral synchronization. I mean, they’re so tightly synchronized it’s almost like a living organism, yet it’s composed of these individual components that somehow coordinate their activity in such a way that they can achieve these near-instantaneous changes of direction. You can see why in the 1950s and 1960s, people believed that to be telepathy within these groups. They exhibit a balance between order and disorder. It’s like a living fluid.
How on earth can simple organisms like fish coordinate their activity with others? Does there need to be a leader to tell others where to go? We’re so used to it in human societies – having a choreographer coordinating complex collective behaviors. Here, what we now know is that there is no such individual. This could be what’s called self-organized, individuals just responding to the movement of local neighbors.
What’s been a big advance in the past three years or so is the development of software that allows us to reconstruct the visual fields of each individual. We can actually reconstruct the world from their perspective to see how they respond to the sensory inputs.
Oceans Deeply: When you have a group of individuals, all of whom are different, do you know yet whether or not those differences essentially wash out in a large group, or are these specific individuals really important?
Couzin: That’s a great question. We simply don’t know. We have a study coming out in Current Biology in a few weeks where we’ve done this but with small groups. There, the differences are really important. The differences in individual behaviors are very important.
Theoretically, in computational models, we can show that these differences should thrive within large groups. We very much expect that to be the case. There could be individuals that have specific positional preferences within groups that may have disproportionate influence over other individuals.
Oceans Deeply: In the video, it almost looks like there’s a shape coming toward the screen. Can you explain what exactly that is doing and what you’re testing for there?
Couzin: That’s called the looming stimulus. It’s meant to represent an object, in this case a predator, approaching rapidly. What we’re doing here is we’re controlling which individual or individuals within this group can actually see that stimulus. The others can’t see it, but they can see those that have responded to it. How do they use social information to make decisions? Does it matter who has seen it? Does it matter if it’s the dominant individual or a subordinate individual? Does it matter where the individual is? How does this information, this meaningful information that there’s a risk present in the environment, how does that spread socially to different members of the group? In this case, the damselfish will go into the coral head for protection.
Oceans Deeply: Do the rules of group behavior change from species to species, or are they pretty constant?
Couzin: I think that’s something that is a very hard question to address. We’ve looked at four or five different species of fish. Broadly speaking, the way that they school is remarkably similar – the general tendency to be attracted to others and to avoid others and so on. However, the exact details, even in terms of which fins they use to propel themselves – those exact details do differ from species to species.
This means that we can create general models, or general simulations, of how schooling occurs. Even though these models are deliberately simplified versions of reality, we can actually make pretty good strong predictions about how real systems behave.
Of course, that’s not to say that there aren’t important differences. For example, in some systems individuals do not recognize others individually. In others, they do. That can have very important consequences [for] collective behavior. That is an example of where there is a very strong difference between two different systems.
Oceans Deeply: What do you think is the most exciting next step in this research?
Couzin: Probably virtual reality. We’ve developed a virtual reality system for freely moving fish. It basically creates a holographic illusion. That means we can have photo-realistic virtual fish that swim exactly like real fish swimming in the tank with a single real fish. This is an incredibly powerful technique that allows us, for the first time, to have a dynamic interaction between a virtual individual and a real individual.
It was just published last week, actually, in the journal Nature Methods. This is a very, very new discovery and new technique. At the moment, it’s working over a very small scale. We want to expand this and develop it, and also link together these different holographic displays so we can have multiple individuals coexisting within the same virtual environment. This, I think, is very futuristic, but it’s real. It’s real. It exists. It really works.
Oceans Deeply: Can any of this understanding of fish schooling be used for conservation or fishing?
Couzin: Yes. We’ve been applying these concepts and ideas to the marine environment using new sonar imaging tools that allow us to track animals underwater. Importantly, working with groups of tens of thousands of individuals is now possible. We’ve been doing experiments with schools of herring, for example, with colleagues in Norway. We’ve also been studying Gulf menhaden in Louisiana.
Oceans Deeply: There are some species whose entire survival is based around schooling in big groups. At a certain point, would a schooling system just collapse if the population gets low enough?
Couzin: We think so, yes. For example, previously we’ve looked at the rule of collective sensing, how a school of fish can be aware of long-range gradients in its environment – things like temperature gradients in the [water column] or gradients of phytoplankton – even though no individual is aware of those gradients. They’ve evolved what’s called an emergence capacity to sense their environment. The danger there is this only works if the group is above a certain size.
With humans also harvesting these populations, we may be harvesting individuals in such a way as we can remove this knowledge, this ability to collectively sense the world.
Oceans Deeply: Were you are able to look at these systems and relate them to human systems, even information systems?
Couzin: Yes. We can relate this to man-made systems like microrobotics. We’ve been working for some years on a new generation of underwater robots that can communicate together and can school together. We are also developing autonomous robots here that swim like fish.
Broadly speaking, understanding these collective phenomena is very useful for controlling unmanned aerial vehicles, but also for understanding human tribe behaviors. We’ve also been studying human tribes and understanding the subconscious transmission of behavior within tribe environments. People aren’t necessarily even aware that they’re copying the behavior of others or moving together in predictable ways, but they are. These general principles can really span a wide range of different scales from how cells communicate and interact all the way to how we communicate and interact.