By Mark Jones
The experiment is brilliant. Researchers came up with a way to compare the intelligence of ants to humans. It didn’t go so well for us. Some of us perform worse at a cognitive test than an ant. Disturbing as that is, the study also shows that when we cooperate in a group, it doesn’t get better.
The radio show Car Talk was a weekly display of human cognition. In real time, without the benefit of preparation, Click and Clack applied their intellect to solve problems. On one show, Tom and Ray received a question about electric brakes. Each clearly stated they’d never heard of electric brakes. They then proceeded to answer the question despite their professed total lack of knowledge. Their answer prompted the famous Andy Letter, shared in a later show. Andy concluded the hosts “proved that even in a case where one person might know nothing about a subject, it is possible for two people to know even less!”
The concept that two or more people making poorer decisions, being dumber, than each of the individuals in a group, stuck in my head. I’ve looked for confirmation in academic literature but never found a definitive study until now. The Proceedings of the National Academy of Sciences recently published “Comparing cooperative geometric puzzle solving in ants versus humans.” It’s an interesting read, but the accompanying videos are what make it fascinating. The authors managed to construct a puzzle where ants and humans are pitted against each other and where performance can be directly compared.
The experiment is a variation of the “piano mover” puzzle. In piano mover puzzles, the time required to move an awkward load through obstacles is measured. The challenge here consists of three rooms in a line with narrow doorways connecting the rooms and a T-shaped load. The load starts in the left room and must be moved to the right room. The top of the T is longer than the width of the doorways and must be maneuvered into the middle room in a particular way to clear the last doorway. It is a purely geometrical puzzle and, as a result, can be scaled in size. A small version works for a single ant, and a much larger version works for a group of people. Human and ant puzzles maintained a similar ratio of body size to load size. Individuals can be tested. The cooperative effort of a large group can be tested. Humans were tested with unlimited communication and under conditions blocking all communication. Ants aren’t able to communicate in a meaningful way to solve the puzzle. Communication blocking mimics the ant experience. The study is brilliant.
Ant brains are about one million times smaller than human brains. Pitting humans against ants in a cognitive challenge is like having the Kansas City Chiefs play a junior high football team. We should slaughter them. That’s not what happened.
On average, individual humans did better than individual ants, but the best ants were better than some humans. Ouch. Teams of ants did considerably better than individual ants. For humans, it was a mixed bag. On equal footing with ants, with no communication, teams of humans did considerably worse than individuals. Ouch again. With communication, human teams and individuals perform about the same. The researchers observed that groups tend to follow a randomly selected leader. The group could get a leader worse than an ant.
I’ve got new proof that the Andy Letter was right, at least for non-communicating groups. Proof a group can be collectively dumber than any individual in the group. Group decision-making in humans isn’t always better. Worse outcomes are possible. The wisdom of crowds is questioned, as is self-directed work. It shows the perils of groupthink. And some of us can be cognitively outperformed by a longhorn crazy ant (Paratrechina longicornis).
You may also like:
Filed Under: Commentaries • insights • Technical thinking
