As I wrote about in my last post, bees are capable of learning which flowers offer good nectar rewards based on floral features such as colour, smell, shape, texture, pattern, temperature and electric charge. They do this through associative learning: learning that a ‘conditioned stimulus’ (for example, the colour yellow) is associated with an ‘unconditioned stimulus’ (nectar). Learning simple associations like these is the basis of all learning – pretty much all animals do it, from humans to the sea slug which doesn’t even have a brain.

However, the world is rarely as simple as this and so animals need to be flexible. For example, as humans we might learn that if we put our bank card in a machine and enter a pin number we can obtain money. However, we might also have to learn that we can only access the bank machine inside the bank during particular hours, or that if we travel to another country their bank machines might operate differently. Therefore we need some behavioural flexibility around what we’ve learned. The same is true for bees. In a bee’s world, much of what she learns relates to getting food from flowers. However, it won’t always be as simple as ‘blue flowers have better nectar than yellow rewards’. Instead a bee might have to learn ‘blue flowers have better nectar than yellow flowers, but only in the morning’ or ‘this particular species of blue flower which also has a specific smell has better nectar than yellow flowers, but another species of blue flower has worse nectar’.

Honeybees can indeed learn more complex relationships like this. This has been shown in many different experiments using different protocols and in different contexts. For example, bees can be trained that an artificial flower which has a blue checkered pattern has good nectar rewards, and one with a yellow checkered pattern has good nectar rewards but a combination of the two (blue and yellow checkered) is not good. They can also be trained to the reverse (that the combination of the two stimuli is good, but that either by themselves is not good). Similarly, honeybees can be trained that only very particular combinations of stimuli are good; i.e. A and B together are good, and C and D together are good, but any other combination (e.g. A and C or B and D) are not good. The list of other complex relationships bees can learn is seemingly endless, but other impressive feats include honeybees’ ability to learn that rewards can be found in a specific location only at one particular time of day and that bumblebees can learn that the location of nectar alternates between two available options and solve physical problems.

However, honeybees’ and bumblebees’ cognitive abilities go beyond these examples of simply learning about their worlds, be it under a number of complex conditions. One excellent study showed that bees could actually form abstract concepts about their world. Having an abstract concept is the ability to understand a general fact about the way things are and to being able to generalise that fact to new situations you might encounter, as opposed to learning relationships that only hold in one particular situation. As humans, we form abstract concepts about the world all the time, generalising from one situation to another. For example, one concept we form about the world is the concept of ‘sameness’ and ‘difference’. If we were having dinner together and I asked you if you’d like ‘more of the same’, you would understand that if we had just been eating pasta that I was offering you more pasta. In another, totally different situation, say we’re operating on someone together and I ask you to pass me ‘the same instrument for stitching people closed that you just gave me a minute ago’ (I’m not sure why any doctor would ever phrase it this way; but let’s just suppose that they don’t have a great memory for medical instrument names), you would understand that you needed to pass me another needle. Therefore, you have the ability to take the concept of ‘sameness’ and use it in two totally different situations. But how would you go about asking a bee if she can do the same thing?

Researchers did this through a cleverly thought-out experiment. First they trained a bee that if she saw a particular colour (say, blue) then when she was later given a choice between blue and yellow, blue always had nectar whereas yellow did not (stages 1 and 2 on the diagram). Similarly, she was trained that if she saw yellow then when she was later given a choice, she had to choose yellow to get the reward (steps 3 and 4 on the diagram). Therefore, she always had to go to the same colour as the one she had previously seen to get the reward. The bees learned this without much difficulty. However, at this point it’s not clear whether the bee had actually learned the concept of ‘sameness’ or instead had just learned a rule for this one situation (e.g. ‘I go to yellow to get a reward when I see yellow and I go to blue to get a reward when I see blue’). To test whether the bees had actually learned the concept of ‘same’, the researchers then presented the bee with a new stimulus, one she had never seen before. This time it was a pattern: black and white horizontal stripes. The bee was then given a ‘transfer test’; a choice between a black and white striped horizontal pattern or a vertical pattern. If the bee had learned the rule ‘when I see a stimulus I then need to choose the same stimulus to get a reward’ (i.e. the concept of ‘same’) then she should fly to the horizontal stripes pattern (steps 5 and 6 on the diagram). This is indeed what the majority of bees did. Another group of bees were trained only to black and white horizontal patterns and then given transfer tests using blue and yellow colours; these bees also showed that they had learned the concept of ‘same’ by going to the correct colour. Now, the really cool part of this experiment was that the researchers then gave a new set of bees stimuli in a totally different modality: scent. Bees were trained that when they smelled a particular odour, they had to go to the same odour to get a reward. They were then given a transfer test in colour, and the bees transferred their knowledge to this new context, going to the ‘correct’ colour even though they had never been trained with colour before. In another set of bees, individuals were trained to go to the different stimulus to the one they had just seen before being given a transfer test, and their choices showed that they were also able to learn the concept of ‘difference’.

After I tell people about some of these impressive cognitive abilities that bees have, another question that I often get asked is, ‘OK, so if bees are so smart, then why do they always fly into windows?’. I hope from what you’ve read in these two posts you can appreciate that when you want to ask a question of a bee you have to frame it in a way that the bee ‘understands’. If we were to ask a human a question, we could use language, to ask a bee a question, you generally use stimuli that represent flowers and nectar. Like all animals, the cognitive abilities of bees have been selected by natural selection to make the bee as good as possible at learning about things that it needs to know about its environment. This includes many complex relationships about how to get the best food from flowers, but sadly, doesn’t include the ability of how to best navigate windows.

Photo Credits

Bumblebee drinking sugar water from artificial flower: Caroline Strang

Bumblebee on flower: jinterwas


Clarke, Dominic, Heather Whitney, Gregory Sutton, and Daniel Robert. 2013. “Detection and Learning of Floral Electric Fields by Bumblebees.” Science (New York, N.Y.) 340(6128): 66-69.

Dyer, Adrian G et al. 2006. “Behavioural Ecology: Bees Associate Warmth with Floral Colour.” Nature 442(7102): 525.

Von Frisch, K. 1956. Bees; their vision, chemical senses, and language. Ithaca, N.Y., Cornell University Press

Von Frisch, K. 1967. The Dance Language and Orientation of Bees. Cambridge, Massachusetts: Harvard University Press.

Giurfa, M., Zhang, S., Jenett, A., Menzel, R., & Srinivasan, M. V. (2001). The concepts of ‘sameness’ and ‘difference’in an insect. Nature, 410(6831), 930-933.

Pahl, M., Zhu, H., Pix, W., Tautz, J., & Zhang, S. (2007). Circadian timed episodic-like memory-a bee knows what to do when, and also where. The Journal of experimental biology, 210(20), 3559-3567.

Schubert, Marco, Harald Lachnit, Silvia Francucci, and Martin Giurfa. 2002. “Nonelemental Visual Learning in Honeybees.” Animal Behaviour 64(2): 175-84.

Strang, C. G., & Sherry, D. F. (2014). Serial reversal learning in bumblebees (Bombus impatiens). Animal cognition, 17(3), 723-734.