Bumblebees for two weeks in a row! This week's paper comes from the Chittka Lab at Queen Mary University, lead-authored by Olli Loukola, published in Science. As you can see in the video above (which was provided in their online supplement), they have trained bumblebees to roll little balls into target holes upon which they receive a bit of sugary nectar solution. This paper has two parts which I will detail in different sections here:
Part I: Learning the task
The first section of this study simply examined whether bumblebees could learn this ball-rolling task. First, bees were trained to drink sugar water from a yellow ball that was stationary within the goal area. Then the ball was displaced from the goal area, requiring the bees to roll it into the goal to get the reward. Bees that did not move the ball into the hole on their own in the first few attempts had the ball-rolling solution demonstrated for them by a model clay bee on a stick that pushed the ball along. The author's don't say how many bees were able to solve the ball-rolling task without a demonstration, I suspect not many, but I am curious. The nine bees that made it through the 30 training trials were all able to solve this task in all 10 of the test trials. Not only that, but over the course of the training trials the bees got more efficient at solving the task, getting the ball in the goal faster and over shorter travel paths.
Part II: Social Learning
The second part of this study examined how bees learn this task. Naïve bees were exposed to one of three scenarios: 1) social demonstration - in which a demonstrator bee which had already been trained to solve the task rolled the ball into the goal and then both bees got sugar rewards from the ball. 2) "ghost" demonstration - this was the same as (1) except that instead of another bee moving the ball the experimenter moved the ball into the goal using a magnet in the ball and a magnet held under the table. 3) control (no demonstration) - the naïve bee entered the arena to find the ball already in the goal with a sugar reward. For the social and ghost demonstrations there were three balls in the arena, and the demonstrator moved the ball that was furthest from the goal. In the 10 five-minute test trials that followed bees could move any of the three balls.
Bees performed the most correct trials when there was social demonstration, then in the ghost demonstration, and the least number of correct trials in the control. Although the demonstrator always moved the ball that was furthest from the goal, the experimental bee would usually move the ball that was closest to the goal. This was even the case in an additional experiment where they made the closest ball black instead of the yellow that they had seen demonstrated. The bees therefore are able to generalize this ball rolling task from the yellow ball they had seen demonstrated to a ball of a different color.
So why should we care that bees can learn to move balls into goals? Well, I think for a couple of reasons. In a general sense this is yet another demonstration of the extraordinary cognitive feats that insects are capable of. What is particularly interesting about this example is that is more distant from normal bee foraging tasks. Many studies have demonstrated bee learning of flower colors, shapes, scents, and even electrical fields. Additionally, studies have shown bee learning of fairly complicated flower handling skills where they have to manipulate flowers to access nectar or pollen. What makes this task different is that they have to go away from the award location (the goal area) to locate the ball and roll it into the goal. It is more similar, therefore, to tool use than other studies on bee cognition. The authors, however, never use the term tool-use. I suspect that this study does not qualify because they first trained the bees to associate the ball itself with nectar rewards. Generally the tool itself is not associated with rewards in studies of tool use. I think the bees' choice to move the closest ball, and the black balls when those were not what they had seen demonstrated, is the most interesting result in this study. This result shows the ability of bees to generalize solutions to problems and apply them in different ways. I think this type of innovation on demonstrated behavior has the most important implications for our understanding of bee problem-solving.