The ant species, Eciton hamatum, can build “living bridges” across gaps in the forest floor, creating a short cut for foraging ants to look for food. The bridges are made of dozens of ant bodies clinging together, with antennas and legs sticking out in all directions. They're incredible to look at (though they make the hair on my arms stand up):
When researchers at Princeton University and the University of Sydney studied this ability in 2015, they did so to examine an example of collective intelligence, the concept that one ant is not very smart on its own, but together, a colony can accomplish incredible things.
“They're hardwired to coordinate by these physiological mechanisms, and are able to do very impressive things such as building ant hills, locating sources of food, or carrying things that are many multiples their collective body weight,” says Anita Woolley, an associate professor of organizational behavior and theory at Carnegie Mellon Tepper School of Business. “You don't think about a smart ant colony as being a function of having smart ants. Better ant colonies are better at coordinating, so it's a function of their ability to coordinate with each other. That's really the idea that we're thinking about when it comes to collective intelligence in teams."
Collective intelligence is seen throughout the natural world, like in flocks of birds or schools of fish. But in humans, up until recently, Woolley says that the intelligence of a group was thought of in mostly in terms of the intelligence of the individual groups members (like thinking about the smartness of a single ant). When she was in graduate school, she began to wonder: "Is there something independent of that," that might explain why some groups were smarter than others? Did teams of humans have collective intelligence too?
There are two overall types of intelligence. Specialized intelligence is the ability to do a specific task in a specific environment, says Thomas Malone, a professor at the M.I.T. Sloan School of Management and the director of the M.I.T. Center for Collective Intelligence. This is like a complicated math problem or repairing a car. General intelligence is the ability to do a wide range of tasks in different environments, even if the tasks don't seem related (IQ tests measure general intelligence).
Woolley, Malone and several collaborators decided to measure the general intelligence of a group, to see if together, people got smarter and why. Their first study, published in Science in 2010, looked at 699 people working in groups of two to five. The tasks included things like "solving visual puzzles, brainstorming, making collective moral judgments, and negotiating over limited resources," the paper notes.
Before they began the research, Malone says that they worried group intelligence in humans would be simple: that collective intelligence would just be the average of the intelligence levels of the group. And they did find that some groups were more collectively intelligent than others, a measurement they called the "c factor."
But interestingly, the average and highest intelligence scores of individuals in the groups weren’t significantly correlated with collective intelligence, meaning that it didn’t make a big difference if everyone in the group was smart, or if there was one genius driving everyone’s behavior. Instead there were very specific qualities that made a group intelligent, and Malone says, "I don't think we expected that, certainly not in the relatively stark form we found it.”
The first finding was that the proportion of women in the group was strongly associated with collective intelligence, so the more women in the group the smarter the group was (though groups with a few men did better than groups of all women).
When they looked closer, they saw that the more social intelligence or perceptiveness the members in a group had, the greater their collective intelligence. They measured that using a test called “Reading the Mind in the Eyes." It measures people’s ability to infer what others are thinking based on non-verbal cues, like facial expressions. Women, on average, score higher on this test then men, and it’s this capability that Woolley thinks leads to the importance of women in the group for collective intelligence. “You've basically raised the average level of ability to read subtle non-verbal cues," Woolley says, about the groups with more women.
It makes sense why this might be important for a group’s functioning—group members are more able to intuit how others are feeling and thinking, which helps the group overall complete tasks. And it's related to another group feature they found to be important: communication. “Groups that talk a lot and also talk more equally—meaning that you don't have one or a few people dominating the conversation— also tend to be a lot more collectively intelligent," Woolley says. "And if you have people who are higher in social perceptiveness they're more likely to have a lot of communication and more equal communication.”
Malone uses the analogy of a brain: Just because you have a bunch of smart neurons doesn’t mean you’ll have a smart brain. It depends on how the neurons are connected to each other. “At some level, I think the same is true for people in groups," he tells me. "Just because you have a bunch of smart people doesn't mean that they can work effectively together. And in some groups, of course, even with very smart people, they work together so poorly that the group as a whole is not nearly as intelligent as its members.”
Because of the importance of a group’s interconnectedness, Malone has even been able to take the brain metaphor one step further. He analyzed the groups through the same mathematical metrics that neuroscientists use to measure brain interactions that may be correlated with consciousness.
Malone tells me there are two important components of a successful group: differentiation and integration. Differentiation means that you have people or groups that can do different things, like in a company: people on the marketing team have very different jobs than engineers. But these people also need to be able to integrate their work together to achieve an overall goal. “This combination of integration and differentiation is important in many kinds of group behavior," he says. "Interestingly, it's exactly those same two things that's the primary model that neuroscientists have developed for measuring consciousness. It includes those same two things.”
Malone and others found that when a group has more of this property, called integrated information, or the greek letter "phi," it was correlated with a group’s performance. (They also found a similar result when they applied this to groups of Wikipedia editors: Editors whose interactions were higher in integrated information produced higher-quality Wikipedia articles.)
New research that Woolley and her collaborators are working on now show that in humans there might also be physiological indicators of collective intelligence. Thinking back to the ants: Their collective intelligence is driven physiologically, through different hormones they emit that help them coordinate. When Woolley and others have looked at humans, they have found some physiological signs of collective intelligence too, in the form of physical synchrony. Preliminary research has seen that people’s facial expression, vocal pitches, and even heart rates start to match up, and more synchrony is associated with a higher level of collective intelligence. “Groups fall into this very quickly," Woolley says. "And the degree to which they can achieve things seems to enhance with the degree to which they coordinate,” Woolley says.
This is all happening at a subconscious level, she assures me. It’s not as if people intentionally match their faces (or heartbeats) to each other. But it’s an interesting phenomenon, and something she’s interested in trying to modulate in the future. The next step might be to ask: How could we intentionally enhance synchrony?
"Can we get everybody to do jumping jacks and raise their heart rate so that we'll enhance collective intelligence?" Woolley suggests. "I don't know. We haven't gotten that far yet, but that's something we're looking at."
The researchers also found a correlation between collective intelligence and groups that had ethnic diversity and cognitive-style diversity. Groups that had at least a moderate level of overall diversity were able to reach higher levels of collective intelligence.
This can be applied to thinking about, say, hiring for certain kinds of companies, he says. We've moved from a largely production-based economy—"hiring people to put nuts on bolts, or chop down trees or work on an assembly line"—to what he calls a "cognitive" economy. These are teams of people coming up with "creative information products" which include conceptual elements, and require lots of different kinds of people to execute.
Scott Page, a professor of complex systems, political science, and economics at The University of Michigan Ann Arbor, has spent a lot of time studying the benefits of diversity. He says if you approach the question of group performance without bias or ideology, but purely from a scientific perspective, you’ll still find that diversity leads to better outcomes. Why? Because of something he calls a diversity bonus. Because people have different approaches, different ways of thinking, different creativity, different backgrounds, sometimes those unique traits can combine together to be better than just the average of their intelligence.
There’s even a theorem, he tells me, to calculate the wisdom of a crowd. It, somewhat densely, goes: "The squared error of the collective prediction equals the average squared error minus the predictive diversity.” Hard to parse, yes, but the bit he emphasizes to me is that last part: minus the predictive diversity. “What you get is these nice theorems that basically show that it's a mathematical fact that the amount by which you do better is gonna be how much more diverse you are," he tells me. He says that of course, the individual ability of group members is important. But if everyone is similarly abled, their overlap will be high, and the team won’t be as good.
Groups with more women, more cognitive diversity, more ethnic diversity, who take turns speaking—it’s hard not to be aware that these are not the components we see in lots of groups that are making important decisions, like, say, the American government. If these qualities improve a group’s intelligence and ability to do a variety of tasks, why don’t we see them more?
There have some analyses that found that women in Congress were more effective at getting their bills through, and at working with the opposite party. The latter could be a sign that they’re more collaborative, which according to Woolley's findings, is important for the collective intelligence in a group overall. But their findings are not enough, Woolley says. “There are so many ways in which our system is currently broken and people are not necessarily motivated for the group to work well together," she tells me. Women only hold 19 percent of Congress seats.
Unfortunately, collective intelligence doesn’t relate to measures of what’s called “group climate,” which is if members in the group are satisfied, how cohesive they feel with each other, or how psychologically safe they feel.
“There's kind of a broader finding in the literature that supports this, where bringing people together who are different can make us a little uncomfortable,” Woolley says. Ironically, this can be a good thing for a lot of group-performance settings. You think more carefully about information, you pay more attention to what other people are saying, and you're more careful about what you say when you're brought together with people who are a little bit different than you, compared to being with people who are just like you.
Page thinks that this is crucial for us to develop. If we don’t, we’re limiting the talent we have exposure to. What is a mathematical fact, he says, is that if there are five models out there, or five ways of thinking of a problem, and you've only picked one, you're not going to do as well as if you pick all five.
“If you're a pragmatic business person, if you're a bottom-line pragmatist, there are bonuses galore here from being able to work effectively in diverse groups," he says. But it’s not a magic button. What you’re creating with diversity is the opportunity to bring different ideas and ways of thinking together, but people still have to listen and do the work that the group has been set up to do.
Malone thinks that’s why you can't just create groups that are optimally set up to have more collective intelligence. You have to consider motivation. “You could have a whole bunch of smart people in the group and if nobody's motivated to do anything that's relevant to the group's goals, then nothing much good is going to happen,” he says.
This is why it’s important for researchers to study the ways that groups behave together, and how they can perform better. Malone recently wrote a book called Superminds, which he defines as a group of individuals acting together in ways that seem intelligent (just another way of describing collective intelligence).
Malone thinks that with this broad definition, superminds have been running the show for thousands of years. Families are superminds, so are countries, armies, scientific communities, and neighborhoods. A lot of his work now is thinking about how technology is going to influence the creation and functioning of these superminds.
Malone think that artificial intelligence has the potential to make human groups smarter, but perhaps in different ways than we think. Woolley says that machine intelligence could improve group coordination, or make its members aware when there’s an imbalance in how much each person is getting to contribute. It could also help match an individual’s specialized skills to certain tasks, and make sure each person is contributing in the way that’s most helpful.
The growth of superminds, to Malone, is inevitable, and so is, probably, their ability to be smart. “What I don't think is inevitable is that they'll be wise," he says. "They may or may not use their intelligence to achieve what we would consider good outcomes. In a certain sense, for instance, terrorist networks like ISIS and al Qaeda are pretty smart, but they've been doing things many people would say are evil. “
Malone says that it’s tempting to feel powerless in the face of superminds, but to remember that part of their very definition is that they rely on individuals. The superminds that you join and support can make a difference in which superminds become the most powerful,he says.
And we don't even need to wait for more advancements in AI or machine learning. Malone thinks technology's ability to provide us hyperconnectivity is an overlooked asset, and critical for the growth and development of collective intelligence—"that is, connecting people to other people, and also to computers, at a scale and in rich, new ways that were never possible before,” he says.
"If you're like most people in the world, you use computers for things like word processing and e-mail and browsing the web. Those are almost always ways not where the computers are using much artificial intelligence, but where the main thing the computers are doing is connecting you to other people, helping you communicate with other people. That's a very important way that computers will help make human groups smarter, by making it feasible to have much larger groups of people effectively collaborating and making it possible for them to collaborate in new ways that were never feasible before."
Sign up for our newsletter to get the best of Tonic delivered to your inbox.