Fish can recognize human faces. Should that change how we think of them?

One of the many astounding things that humans can do is recognize faces. Even though most faces we encounter are basically similar — a rough oval shape, two eyes, a nose, a mouth — most of us can pick out subtle differences and distinguish among hundreds of mugs. In an instant. Monkeys and other primates are similarly adept.

For a long time, scientists thought this feat required sophisticated brain machinery — a combination of perception and memory. Brain scans show that a part of our neocortex known as the fusiform gyrus becomes activated when we look at faces. People with damage to the fusiform gyrus lose their ability to recognize faces, a disorder known as prosopagnosia. Surely this was a special talent.

But perhaps not! A fascinating new paper in Scientific Reports offers evidence that archerfish — a tropical fish that spits jets of water to stun prey — can be trained to recognize human faces with surprising accuracy, even though they lack the same complex brain structures. If true, it's yet another piece of evidence that fish are much, much smarter than we think.

How to teach a fish to recognize human faces

In the study, led by Cait Newport of Oxford University, the researchers trained four archerfish to spit water at an image of a particular human face in order to receive a food prize. They found the archerfish could then distinguish that face from 44 other faces with surprising accuracy (81 percent of the time, at peak).

Here's a video from LiveScience below showing them in action:

As always with animal studies, it's tough to be 100 percent confident that the fish are actually recognizing faces. We don't have access to their inner lives, only their behavior. But it sure seems like they do. The researchers tried again with a fresh set of four archerfish, standardizing face shape and switching to black and white images. The fish could still reliably distinguish among 18 faces, 86 percent of the time.

Now, fish aren't the only animals that can do this. Pigeons have also demonstrated an ability to distinguish between different human faces. But bird brains have neocortex-like structures — long thought to be important to cognition — similar to those in human brains. Fish don't have anything like a neocortex. That's what's so surprising.

I asked Culum Brown, an Australian biologist who studies fish intelligence but wasn't involved with this study, what he made of these results. He argued that it should make us rethink some of our ideas about the brain:

I think humans are overly wrapped up in the importance of the neocortex with respect to its role in cognition. The human neocortex has taken over many of the functions that take place in other parts of the brain in other animals. This recent work on object/human face recognition in archerfish is just another example. Indeed much of the visual processing that takes place in the optic lobes and cerebellum in fish occurs in the neocortex in humans. This is also true of pain perception.

The result clearly indicate that the neocortex is not the grail of intelligence.

To put this another way: Yes, scientists can see our fusiform gyrus light up in brain scans when we recognize faces. But that doesn't mean this brain structure is absolutely necessary. It might just mean that this is the spot where humans happen to process that information. Other brains have different ways of accomplishing the same task.

This is more evidence that fish are smarter than we think

(Tambako)

Brown also noted that the facial recognition study jibes with quite a bit with what we've been learning lately about fish intelligence. "Most fish species are social and have the capacity for individual recognition," he wrote. "We have known that for more than 20 years. We also know that fish can use facial features to categorize predator and non predators. Fish are much smarter than people think."

A few years ago, Brown published a review essay called "Fish intelligence, sentience and ethics" in which he laid out the considerable body of research that fish are surprisingly intelligent — no less than, say, pigs or cows. He also expounded on this argument at length in this Q&A with Joseph Stromberg for Vox.

Fish can learn from each other. They can recognize other fish they've spent time with previously. They know their place within fish social hierarchies. They remember complex spatial maps of their surroundings. (Also it's a myth that goldfish have a three-second memory.)

"Anybody who feeds fish will tell you this," Brown said. "In the morning, at the appropriate time, the fish will gather at the right end of the tank, expecting to be fed. That's called time-place learning — they're learning a place and associated it with a time."

If fish really are intelligent, should that change how we think of them?

The really provocative question is whether fish intelligence has any big ethical implications. Some people go vegetarian because they think it's morally wrong to kill and eat a sentient cow or pig. But a subset of vegetarians will make an exception for fish — perhaps because they might think of fish as "lesser" creatures. So what if that's wrong?

Similarly, while animal rights groups will often campaign for better living conditions for cows or pigs or chickens, you don't quite see the same activism around the welfare of fish caught in the wild or raised in aquaculture. Here's Brown:

[B]ack in the '50s and '60s for terrestrial commercial farms, we started to think about things like moving pigs on the backs of trucks, and whether chickens had access to the real world. That revolution stopped at the water. Every major commercial agricultural system has some ethical laws, except for fish. Nobody's ever asked the questions: "What does a fish want? What does a fish need?"

Part of the problem comes back to the question of whether fish feel pain. But for the last 30 years, the neurophysiologists have known that they do, and haven't even argued about it. And from an evolutionary perspective, our pain perception systems — and the systems of all terrestrial vertebrates — come from a fish-like ancestor. Whether they're in the water, or on the land, they all have the same pain receptors. But for some reason, a lot of people refuse to believe that fish can feel pain. …

I think, ultimately, the revolution will come. But it'll be slow, because the implications are huge. For example, I can't think of a way to possibly catch fish from the open ocean in a massive commercial way to meet demand that would be anyway near our standards for ethics if we think of them like other animals. Currently, you go out, you catch a bunch of fish, you crush most of them to death in a net, you trawl them up from the bottom of the sea — which causes barotrauma for most of them — you dump them on a deck, half suffocate to death, the ones you don't want get thrown overboard and die anyway, and the ones you keep go on ice, just to preserve the flesh for market reasons. How do you do that in a way that has the fish's interests involved to any degree? You can't.

So it's not surprising that there is some fierce opposition to this idea. It would mean a massive change in the way we do things.

It's a provocative argument. Obviously not everyone will find it convincing. But it's certainly a challenge to many of our usual notions of animal rights.

Further reading:

• Here are two good interviews with Oxford's Cait Newport about the fish recognition study. Here's what she told CNN on the question of whether aquarium fish might recognize their owners: "When strangers walk into her lab, the fish 'act skittish,' she said. "When I walk in, they start spitting at me — many cases right in the eye.'"
• In the New York Review of Books, Elizabeth Kolbert has a nice essay about how we're constantly searching for the trait that makes humans "special" — and usually find that other animals have it to at least some degree. (Which isn't so surprising, evolutionarily speaking.)
• The full-length interview with Culum Brown about fish intelligence.

Explanatory journalism is a public good

At Vox, we believe that everyone deserves access to information that helps them understand and shape the world they live in. That's why we keep our work free. Support our mission and help keep Vox free for all by making a financial contribution to Vox today.

One-Time Monthly Annual

$95/year

$120/year

$250/year

$350/year

Other

$

Yes, I'll give $120/year

Yes, I'll give $120/year

We accept credit card, Apple Pay, and Google Pay. You can also contribute via