Science Watch

The brain of a honeybee contains only about a million neurons--a hundred thousand times fewer than the brain of a human. But those neurons enable the bees to perform a variety of complex behaviors, from building intricate hives to remembering and communicating the location of food sources.

Now researchers at the University of Hawaii have found evidence suggesting that bees can learn to use short-term memory to control their behavior--one of the few basic learning abilities that vertebrates have but that, until now, Apis mellifera appeared to lack. The study appears in this month's issue of the Journal of Comparative Psychology (Vol. 117, No. 1).

"What this study does is provide another example of a vertebrate learning phenomenon that we couldn't find early on [in bees], but which then appeared as a result of further work," says M.E. Bitterman, PhD, who co-authored the study with Patricia Couvillon, PhD, and Todd Patrick Ferreira, a former University of Hawaii undergraduate.

The findings provide new evidence that, despite hundreds of millions of years of evolutionary separation, bees and vertebrates have converged on a similar set of basic learning mechanisms, researchers say.

Initial findings

Previous studies of short-term memory in bees have produced mixed results.

In the mid-1990s, Villanova University psychologist Michael Brown, PhD, and Indiana University psychologist Gregory Demas, PhD, then at Villanova, conducted a series of experiments to find out whether bees could learn to avoid places where they had previously found food.

In their experiments, reported in the Journal of Comparative Psychology (Vol. 108, No. 4), Animal Behavior (Vol. 50, No. 4), and Animal Learning & Behavior (Vol. 25, No. 2), bees were released into a radial maze with six arms, each of them stocked with sugar-water.

As the bees explored the maze, feeding on the food they found, they showed a slight preference for arms they had not yet visited. Like rats, bees seemed to be able to use short-term memory to avoid revisiting food sources they had already exhausted.

But in subsequent experiments, Couvillon, Bitterman and their collaborators were unable to replicate Brown and Demas's findings. And attempts to show that honeybees could use short-term memory for colors, as they had appeared to use short-term memory for places in Brown and Demas's studies, were at first unsuccessful.

Refined techniques

Couvillon and Bitterman's experiments on short-term memory for color involved two conditions: a perseveration condition and an alternation condition.

In both conditions, bees were given a reward of sugar-water when they landed on a "sample" patch of color. They were then shown two more "target" color patches, one of which was the same color as the sample. In the perseveration condition, the bees were rewarded for flying to the target patch that matched the sample. In the alternation condition, they were rewarded for choosing the nonmatching target.

Bitterman and Couvillon found that bees tended to perseverate--to choose the color for which they had most recently been rewarded. That suggested that they might be able to use short-term memory for color.

But the conclusions Bitterman and Couvillon could draw from the study were limited by several confounding factors. One of them was that the original sample remained visible while the bees were choosing between the targets. That meant that it was possible that the bees were simply being guided by the presence of the sample--not by their memory of having been rewarded for choosing it in the past. There was also no evidence that bees could learn to alternate; no matter how they were trained, they always tended to chose the matching target.

Then, in 1998, Brown and his collaborators published a study in Animal Learning and Behavior (Vol. 25, No. 4) that addressed the first limitation. Using a horizontal computer screen to display the color patches, Brown and his collaborators were able to turn off the sample color as soon as the bees had finished feeding. They found that, even without the sample, bees still tended to choose the matching target.

A resolution

Only one thing was missing: evidence that bees could learn to alternate. But, as Bitterman explains, "it's the alternation that is critical" for demonstrating that bees share with vertebrates the ability to use short-term memory flexibly.

In their most recent study, Couvillon, Ferreira and Bitterman adapted the computer-display technique used in Brown's 1998 study, but changed the training procedure to make the amount of time between each trial more consistent.

With that change, they found that bees no longer demonstrated a preference for perseveration. In fact, they learned to alternate just as easily as they learned to perseverate. The results provide the first strong evidence that bees, like rats and other vertebrates, could learn to use short-term memory to control their behavior.

The big picture

When Bitterman and Couvillon first started studying honeybees more than 20 years ago, they thought they might find major differences between learning in honeybees and learning in vertebrates.

After all, notes Bitterman, "Invertebrates and vertebrates have been evolving independently for about half a billion years, and their common ancestor had hardly a brain at all."

So it was a great surprise, says Bitterman, when the early studies on bee cognition showed that all the basic mechanisms of conditioning in vertebrates--with only a few exceptions--could also be found in bees.

As research on honeybee cognition has progressed, even those few exceptions have mostly disappeared. The key to those advances, says Bitterman, has been the development of new research methods designed specifically for foraging bees.

Unlike rats, for example, bees can't be kept in cages for more than a few hours, so researchers had to devise new methods to lure bees into the experimental situation, to keep them there for as long as the experiment required, and to get them to return to the lab after each trip back to the hive.

However, there are still limitations on the kinds of experiments that can be done with bees, says Bitterman, and future studies may yet reveal fundamental differences between vertebrate and invertebrate conditioning. There are also a number of open questions about the neurobiology underlying the bee's cognitive abilities.

"There's a lot of work to be done," says Bitterman. "But there is no mammalian conditioning phenomenon that we've looked for in bees that we haven't yet found."