A pair of scientists from the University of Regensburg in Germany claim they've conducted a study that shows ants know when they don't know something, and that they react accordingly with doubt.

Typically, a colony of ants works together by leaving pheromone trails to help one another find food. But scientists Tomer Czaczkes and Jürgen Heinze found that when they created an unpredictable situation for test ants, they were less likely to leave such a path for their sisters to follow.

The study involved encouraging black garden ants to find food on a T-shaped maze. At first, the food was consistently placed in one arm of the maze. When the scientists switched the location of the food to the other arm, ants heading to the now-foodless location were less likely to leave a trail.

"It makes sense," says Czaczkes. "You don't want to give your sisters wrong information."

Once the ants had successfully found the food again, they left an even stronger trail of pheromones on their way back to the colony—presumably in the hopes that doing so would override the now-false information from before. But the most striking aspect of the study came when ants chose to check the newly-empty arm—this time, they didn't leave behind any pheromones at all.

"After finding the new food location once, they still usually think that the food is in the old location—they're just not sure," says Czaczkes.

This ability to recognize what is definite information and what is doubtful hints at a level of metacognition previously unseen in insects (except bees). Still, not everyone thinks that is what is happening.

"Ants might have other motives, such as exploring. And when exploring, they might not lay down as much pheromone," says Ken Cheng, from Macquarie University in Sydney, Australia. Joachim Zeil of the Australian National University posits that the behavior could be accounted for by thinking of it as a statistical anomaly.

The scientists behind the study maintain that they're on to something. "We don't have a smoking gun just yet for the definite proof of metacognition in ants, but we've got plans to find it," says Czaczkes.

[h/t New Scientist]