On the Social Logic of Fish

by F.

As we all know from our days in the high school cafeteria, if Frank is more popular than Ned, and Ned is more popular than Cecil, then Frank is more popular than Cecil. That’s the way hierarchies work, and it doesn’t take much effort for human minds to think like this. Now we know that fish think the same way—at least some fish.

In a widely reported and blogged experiment published in Nature, Grosenick et. al. found that African cichlids can reason (in a sense) about dominance hierarchies.

Basically, the researchers set up mano-a-mano (actually, fisho-a-fisho) fights between cichlids. But here’s the twist: they let another fish watch. This bystander fish would watch some bouts and then would be put into a tank with two fish, both of whom the bystander had seen fight, but not necessarily together. Mr. Bystander then had to choose whom he wanted to spend time with. Generally, cichlids prefer to be with weaker fish (less of a threat, it appears). Even when Mr. Bystander hadn’t seen the two fish fight, he was able to correctly choose the weaker one.

How? By transitive inference. Mr. Bystander may have seen Fish A beat Fish B, and Fish B fight Fish C. So, when Mr. Bystander got a chance to hang out with either Fish A or Fish C, he chose Fish C—the weaker, less threatening fish. Here’s more detail, from a Stanford press release:

The fights were staged in a square tank divided into several compartments. A lone male observer—the “bystander”—was placed in a cubicle in the center of the tank. Surrounding him were five smaller compartments, each with a solitary male rival identified simply as A, B, C, D or E. Researchers made sure that the bystander and his five potential rivals had never met.

Although the bystander remained alone in his cubicle and never swam with the others, he was allowed to observe a series of fights between rival pairs—A vs. B, B vs. C, C vs. D, and D vs. E. Researchers manipulated the fights so that A would dominate B, B would dominate C, and so forth down the line.

“These fights, taken together, imply the dominance hierarchy [A is better than B is better than C is better than D is better than E]” the authors wrote. But did the bystander really comprehend this intricate pecking order, and if so, would he use that knowledge to make logical decisions about the same fish paired in new relationships?

To find out, eight different bystanders were tested in the familiar square tank and in a new setting—a rectangular aquarium with three adjacent compartments. In each test, a bystander was placed in the middle compartment between two sets of rivals that he had never seen together—A and E (AE), and B and D (BD). At this point in the experiment, all the rivals had recovered from earlier losses, so their physical appearance was similar, right down to the eyebar. From the point of view of the bystander, therefore, each rival looked like a winner.

Using a video camera, researchers recorded which rival the bystander approached first, and the overall time he spent next to each of them. “Previous experiments in A. burtoni and other fish have shown that time spent in tank quadrants adjacent to a particular male indicates bystander ‘preference,’ and that bystanders spend more time near the rival they perceive to be weaker,” the authors explained.

Here’s the bottom line:

The results were dramatic. Virtually all of the bystanders swam to the weaker rival first and stayed near him for a significantly longer period of time. In the AE tests, bystanders preferred E, the wimpiest of all the losers, over A, the top fish in the tank. In the more subtle BD tests, most bystanders chose D over B, even though these two rivals were ranked very close together on the dominance hierarchy.

“These results show that fish do, in fact, use transitive inference to figure out where they rank in the social order,” Fernald said. “I was amazed that they could do this through vicarious experience, just by watching other males fight. In Lake Tanganyika, where conditions change all the time, it would be advantageous for a male to know who the new boss is going to be and who his weakest rivals are. Our experiment shows that male cichlids can actually figure out their odds of success by observation alone. From an evolutionary standpoint, transitive inference saves them valuable time and energy.”

Cite: Grosenick, L. et al. (2007). Fish can infer social rank by observation alone. Nature 445: 429-432