Do bees know what they don’t know?

Thinking about thinking, or “metacognition,” is no easy feat, since it requires self-awareness and reflection. We humans are masters at monitoring our own thoughts, and we can even assess what—and how much—we don’t know about something and use this knowledge to inform our decisions. It’s called “uncertainty monitoring,” and it’s a calculation that our brains often do without us even being aware of it.

It’s unclear whether other animals also have this ability. Although there is some good evidence demonstrating metacognition and uncertainty awareness in primates, the extent to which smaller-brained animals can ponder their own cognitive processes is the subject of hot debate in the scientific world.

In this week’s issue of PNAS, Clint Perry and Andrew Barron, two Australian researchers, examine uncertainty monitoring in honeybees. Bees are a popular subject for cognitive research, since their brains are relatively simple, and we know quite a bit about how some of their cognitive processes unfold. These reliable little critters are also relatively easy to work with, even outside of the lab.

The researchers created a test chamber with two cylindrical “rooms,” each of which held a horizontal reference bar and two targets. Free-flying bees could enter a chamber and drink a solution from one of the two targets inside. However, the location of the targets was a vital signal to the bees; for one group, targets positioned above the reference bar dispensed a sugary solution, and targets below it doled out a bitter quinine solution. For a second group of bees, the targets were reversed. Bees had to choose a target before they were released from the chambers. Not surprisingly, the tiny test subjects quickly learned which targets to avoid and which to drink from.

But the tests got harder when the researchers introduced varying degrees of difficulty. In this phase, some targets were positioned well above or below the reference bar, making the decision obvious. Other targets overlapped the bar but were offset slightly so that the choice was a bit more difficult. Still others were perfectly lined up with the bar so that it was impossible to tell what type of solution would be inside each.

In some trials, bees could simply fly from one chamber into the next without choosing a target, thereby “opting out” of the choice presented to them. In other trials, the exit hole was blocked off, forcing the bees to make a decision about which target to drink from before they were released. If the bees were capable of monitoring their uncertainty, they should opt out more often when the trials were hard, since they lacked enough information to make informed decisions.

And they did: the bees opted out more often when faced with difficult and impossible trials—when the target overlapped the reference bar—than they did on easy trials. The scientists also looked at the differences in performance on forced and unforced trials and found that opting out actually improved the bees’ overall performance. The bees were “right” on difficult trials more often when they had the choice of opting out and flying into the other chamber, compared to when they were confined to the choice in the first chamber.

According to Perry and Barron, these results suggest that bees are opting out of trials and cutting their losses when they don’t have enough information to make an educated guess. If this is true, it’s a pretty significant finding: bees—and potentially other non primates—know when they don’t know enough and can react to varying degrees of uncertainty in order to optimize their performance.