Now the Fish Are in on It

As with the earlier research, the study was led by psychobiologist Christian Agrillo of the University of Padova. Agrillo’s team began in the same manner as in its previous experiment: training mosquito fish to swim through doors that divided up the fish tank, with one doorway leading to other fish (an incentive for mosquito fish, which are highly social). The scientists placed varying numbers of geometric shapes by each doorway, then trained the fish to associate the “right” door (the one leading to their friends) with a specific number of shapes. (Which geometric shapes were used varied throughout the experiment, but the number leading to the other fish remained constant.)

In the new experiment, the researchers challenged the fish by dramatically increasing the number of shapes beside the doors into the hundreds.

In the previous experiment, the number of shapes labeling each door was relatively small: single-digit numbers, such as four or eight. In the new experiment, the researchers challenged the fish by dramatically increasing the number of shapes beside the doors into the hundreds. After an initial adjustment period, the fish again learned to successfully distinguish the numbers and select the correct door.

“It was kind of funny, most of them appeared to be surprised when we switched from small numbers to hundreds,” Agrillo explained. “However, after a short while they started to solve the task as well.”

The experimenters also found that, unsurprisingly, the fish were more successful when the ratio between the number of shapes on the correct and incorrect doors was large—for example, 1:2 or 2:3. When the ratio was changed to 3:4 (and, hence, the numbers of shapes were closer together), the fish could no longer distinguish between doors.

Next, the team designed a similar experiment for undergraduate students. Participants were forced to determine the difference between large numbers in two seconds, too little time to allow conscious counting. Although the human participants were more successful overall, their success rate, like the mosquito fishes’, declined as the ratio between numbers increased to 3:4.

“You just don’t expect interesting results like this when dealing with animals like fish,” Agrillo said. “We thought this was really incredible.” Although the scientists argue that a common ancestor explains the number-processing abilities of fish, humans, and other vertebrates, the research offers just as much support for common design.