Whenever we humans invoke goldfish, it is often as the butt of a joke. When we wish to criticize a bozo, we say they have a goldfish's attention span, or a goldfish's brain, or the memory of a goldfish. But this comparison isn't quite fair to the fish, which are able to learn and retain information for long periods of time. A goldfish can remember things for weeks, and some studies have shown goldfish can solve mazes from memory, as this Short Wave segment explains.
In recent years, a spate of studies has helped to redeem the humble goldfish—a domesticated carp—from the realm of the dumb. Some involve food, such as teaching the fish that pushing a paddle of a certain color leads to a food reward, or how to distinguish Stravinsky from Bach. In one study published in January, a team of researchers taught goldfish to drive a "tank on wheels" on land in pursuit of a food reward. This study went viral for its commitment to the bit, although testing a fish's ability to drive on land feels as illuminating a test as teaching an owl to use Seamless; perhaps they can, but why would they ever need to?
In a paper published in October in the journal Proceedings of the Royal Society B, a group of researchers from the University of Oxford put a handful of goldfish to a new cognition test to illuminate how fish navigate and move through space. The task of being aware of where you are in space is crucial for many animal species "to find food, or hide, or find a sexual partner," according to Adelaide Sibeaux, a researcher at the University of Oxford and an author on the new study.
Many terrestrial species, including you, a human, use a process called optic flow, where an animal measures the degree of visual change due to their (your) motion in the environment. Humans and other terrestrial vertebrates use what is called global optic flow, measuring how the angle between their eye and surrounding objects changes as they navigate through their environment. Imagine you are looking out the window of a train whizzing through some beautiful countryside. The trees and mountains you pass appear to move backward, with the trees closer to you appearing to move faster than the mountains in the distance.
As evidenced by the tank-on-wheels study, goldfish can navigate efficiently in their environment. But the mechanisms by which the fish accomplish this were less clear.
To test the goldfish, the scientists first had to train them. Sibeaux's team purchased nine goldfish from an appropriately named local shop called The Goldfish Bowl and named all of them: Coco, Toto, Momo, Sushi, Maki, Sashimi, Tiger, Winnie, and Bubulle. In the lab, the fish ate pellets in the morning, snacked on spinach or bloodworms in the afternoon, and socialized with the one or two other goldfish living in their tanks. And then the learning begun.
"Training with goldfish, and fish in general, is a very long process," Sibeaux said, sighing. The steps were simple and slow. She would drop a goldfish in the experimental tank: a narrow tank with Beetlejuice-like wallpaper of black and white vertical stripes, each two centimeters wide. Once the fish had gotten comfortable, Sibeaux dropped treats to guide the fish to swim to the target distance of 70 centimeters, and then wave at the fish to signal that it's time to swim back to the starting position. After the fish learned this, Sibeaux stopped putting food at the target distance, and then stopped waving to remove any external cue to turn around. Eventually, eight fish learned to swim to the correct distance and back in order to receive their treat.
Coco being trained.
Once the fish mastered this, Sibeaux began altering the background of the experimental tank. When she changed the width of the vertical stripes from two centimeters to one centimeters, effectively doubling the frequency of spatial information in their surroundings and giving the fish the impression they were moving more quickly through space, the fish overestimated the distance they traveled by 36 percent and turned around before they reached the actual target distance. When she changed the tank background to horizontal lines, the fish swam shorter, more inconsistent distances than either of the tests using vertical lines. But when Sibeaux changed the background to a two-centimeter checked pattern, the fish performed just as well as they did with the two-centimeter vertical lines. This outcome indicated the fish were measuring their distance based on the frequency in the spatial information they received, not just the pattern. In other words, goldfish also use optic flow but use a different visual mechanism to process the information.
These results did not surprise Sibeaux, who recently found Picasso triggerfish passed the same test. But Picasso triggerfish live in very visually different environments: a brightly lit coral reef with high contrast poses different navigational challenges to the murky, weedy pond of a goldfish. "I was very happy to find out that they were able to estimate this distance," Sibeaux said, speaking of the goldfish. "If you don't give an individual a chance to show how smart they are, they will never do it," she added.
Coco being tested.
From her ample experience training fish, Sibeaux is very familiar with their "high cognitive abilities." "They are much smarter than we think," she said, adding that she believes this is true for every species that we do not pay attention to, from fish to insects. "Every species is perfectly adapted to their environment. No species is dumb."
Indeed, the eight goldfish that excelled at Sibeaux's test are a testament to their species' ability to perform specific, trainable tasks and remain unflappable in the face of checked walls. But, if you remember, Sibeaux acquired not eight goldfish for their experiment, but nine.
Coco, Toto, Momo, Sushi, Maki, Sashimi, Tiger, and Winnie all passed the test.
Bubulle did not pass the test.
Although Bubulle (think "bubble" in a French accent) did learn how to swim to a set distance and turn around for a treat, Bubulle would constantly swim significantly past the target distance before turning around.
Sibeaux tried sending me videos of Bubulle failing the test, but the files were somehow glitched, which felt quite on brand for Bubulle.
"This fish was just a fish like the others," Sibeaux said, noting that any behavioral experiment will reveal individual differences between the participants, and even the eight goldfish that did pass the test had varying experiences being trained. "When you work with them, you just start to learn about their different personalities. Some of them will do the experiment quicker and learn faster than others, while some others will take longer to learn but be more precise at the end."
"This just shows the individual differences we have in the world, and without this, I think the world would be boring," Sibeaux added.
Even though Bubulle failed by the standards of Sibeaux's test, the fish did still learn all the steps of the experiment, which if not efficient still seems impressive to me in its own way. Maybe Bubulle is simply an overachiever, dreaming of more—more distance, more stripes, more treats. Or maybe Bubulle just isn't very self-aware. There's room for all kinds.
