How Bumblebees Make Smart Decisions: Inside their flexible learning strategies (Ep. 82)

Do you make snap decisions, or do you like to pause and weigh your options? Today, we’re stepping into the decision‑making minds of bumblebees and the surprising logic behind their efficiency.

Dr. Anna Stöckl shares how she and her team used artificial flowers in controlled lab experiments to understand how bees choose which blooms to visit. They found that bumblebees don’t always learn the same way.

When the bees encountered flowers with very different colors—say, blue versus yellow—paired with shapes or patterns, they focused only on the color and learned it quickly. But when the colors were more similar—yellow versus orange—they slowed down and learned both the color and the pattern.

Anna also highlights why this matters beyond the lab. As our floral landscapes shift—through gardening choices, habitat loss, or climate change—bumblebees’ ability to adjust their learning strategies helps them keep up with new flower types and changing nectar resources. That’s hopeful news for conservation.

She also touches on the power of citizen science. Platforms like iNaturalist help researchers track bee diversity and behavior across regions. If you’re curious about how to get involved, we covered simple ways to contribute in episodes 77 and 81.

Dr. Anna Stöckl is a Professor at the University of Konstanz in Germany. She is the Principal Investigator at the Visual Neuroethology Lab which studies how insects process visual information in their environment. Connect with her on Bluesky and read the paper we discussed.

Good to know

Bumblebees are constantly weighing the cost of learning against the benefit of accuracy. When a cue (like color) is easy to distinguish, bees exploit that efficiency. When they need more information, they invest the extra time to learn it. It’s strategic.

Transcript

[00:00:00.190] – Jacy

Welcome to The Bee’s Knees, a podcast wild about native bees. Wild and native bees are under threat worldwide. In each episode, we look at actionable things we can do to support these adorable little guys whose pollination work is crucial for maintaining biodiversity. I’m Jacy Meyer, and I thank you for being here. Have you ever watched a bee land on a flower and wondered how it decided that that was the right one. Was it color? The shape? A pattern we barely notice? Maybe all of the above? Today we’re looking into the minds of bumblebees to explore how they make decisions when they’re faced with multiple cues at once. It turns out bees are a lot more strategic than we give them credit for. They’re constantly weighing the cost of learning new information against the payoff of making accurate choices, and they can switch strategies depending on how easy or hard a cue is to tell apart. Dr. Anna Stöckl is joining us to break down all these insights and more. She and her team tested how bees learn colors, shapes, and patterns during a foraging task. We talk about why bees sometimes rely on just one cue, why they sometimes learn several at once, and what this tells us about their learning dynamics.

[00:01:30.720] – Jacy

So can you give us a brief overview of how you conducted this study, and what was the inspiration behind using bumblebees?

[00:01:39.140] – Anna

Yeah, so we were really interested in decision-making questions. So we wanted to know how do pollinators make decisions about, in this case, particular flowers and the visual information on flowers. And bumblebees are great decision makers because they visit dozens, if not hundreds of flowers a day. So one single individual does. And every flower visit is a decision for do I approach this flower? Do I land on this flower? Does it have the right cues and the right visual information for me to land here? So a bumblebee in nature, and maybe I should say bumblebees. So we use Bombus terrestris, which there are different names for them floating around. We call them the buff-tailed bumblebees, but they’re also called large earth bumblebees. And these in particular make lots and make lots of decisions in a single day, a single forager. So they’re a great system for this study. And we conducted the study in the lab because also this particular species of bumblebee is also great for laboratory experiments. They very happily live in hives in the lab, and we can give them a big flight cage basically in which they can fly around and forage from artificial flowers.

[00:02:48.170] – Anna

And so this is How we conducted the study in that we gave them different artificial flowers, and what I mean by that is basically a paper cutout that could have different shapes and it could have different patterns printed on it. So for example, round shapes and star shapes and circular patterns and star patterns, and they came in different colors. And so these flat cutouts we stuck on little stools, if you want, so they’re not directly on the ground, and then we would provide sugar water on them and the bumblebees could fly from the hive into this big flight box and could forage from the flowers. And so this is the general setup, so to say, of the experiments. And then we would observe which flowers did a single bumblebee forager visit during their foraging trips into this box. And a bit more for the general questions of the study and how we set out to test it. So we wanted to know if we give the bumblebees different visual information. So for example, visual information they know, we know that they use, like colors, and shapes and patterns that natural flowers also have. We made our artificial flowers to always have two types of information.

[00:03:56.880] – Anna

They always had a color and then associated with the color could either be a shape or a pattern. And we always generated these cues in pairs. So there might have been a blue and yellow combination. So one flower was blue, one was yellow, and the blue one might have been a star shape and the yellow one might have been round. Or the blue one might have had a star pattern and the yellow one a circular pattern, so that the bees would always be presented with a color and a shape or a color and a pattern combination, and always a pair of these so that one would then be associated with sugar water. So we wanted them to make a positive association with it and visit it, and the other one with water, which the bees normally don’t appreciate very much, especially when we also present them with sugar water, so they will learn to avoid these flowers. And the whole idea of presenting two cues at the same time was then what we can do is if we give them, let’s say, blue and star and yellow and circle, is that we can train the bees to, for example, like blue and star and not visit yellow and circle because it had water and the blue and star had sugar.

[00:05:07.420] – Anna

We could then test what did they actually rely on with their decisions? Did they rely on the color? Did they rely on the shapes or the patterns by swapping the cues. So we could make the blue a circle and the yellow a star, and then ask, do they go for the color that they learned or the shape that they learned or the pattern that they learned? So that’s the general setup of the experiments, testing what cues do the bumblebees actually learn if we give them color shapes or patterns. And if we force them to make a decision, which ones do they rely on, the colors or the shapes or patterns?

[00:05:44.390] – Jacy

So let’s talk about that a little bit more, because you found that sometimes they relied only on the color, and other times they learned both the color and the shape or the pattern.

[00:05:53.380] – Anna

Yeah.

[00:05:53.760] – Jacy

So what does this tell us about how they weigh the cost of learning against the benefit of accuracy?

[00:06:02.500] – Anna

Yeah, so it was actually a little surprising for us. So what we found was that if we made them choose, and if we forced them to choose, they always chose color over the patterns or shapes. So their, their first instinct, so to say, is to go for color. But we had two different conditions. And as you said, in one, they actually learned both the color and the patterns or shapes, and in the other one, they didn’t. And these two conditions are basically conditions in which the colors of the two flowers we gave them were very different, or conditions in which they were very similar. And different means they were blue versus yellow or orange versus cyan shape of blue. And similar means they were yellow and orange or darker blue and cyan. And when the bees were presented with these very different colors paired with shapes or patterns, they didn’t even learn the shapes or patterns. They only learned the colors and they pretty much exclusively also chose the colors for their decisions. Whereas when they— when we use more similar colors like yellow and orange, bees still had a tendency to choose colors over shapes and patterns when we forced them to make a decision between them.

[00:07:14.550] – Anna

But they learned both cues, which we could then test in follow-up experiments on the same bee, where we presented only the shapes and patterns and asked them, do you actually know which one was the one that you got sugar water on? So with similar colors, the bees learned the shapes and patterns, and with very different colors, they didn’t. And we also noticed another thing, and that explains a bit the second part of your question, what does it tell us about the cost of learning? When we trained the bees with very different colors, like the yellow and blue, they learned very quickly. So after only a few flower visits, they already made very accurate decisions for the correct color, which we had rewarded with sugar water, and ignored the color that had water on it. But when we gave them more similar colors or patterns or shapes without a color, they took much longer to learn. So many more flower visits to learn, which suggests that really the cost associated in this case with the learning, at least the one that we could measure, is time. It took them much more time and more energy therefore to get as accurate in choosing the right flowers with similar colors, patterns, or shapes as it took them with very different colors.

[00:08:29.100] – Anna

So that was way faster. So it took way less time to learn. So that would suggest that— or our interpretation of our finding therefore is that when bees can use a cue with which they’re very efficient and very quick, like the very different colors like yellow and blue, they solely rely on them because it saves them a lot of time and energy. Because if they also then had to learn a pattern or shape, which takes them much longer to learn, they would invest much more time in visiting flowers until they’ve also learned that to a high degree of accuracy. So that’s the interpretation of our results, is that bees basically decide for a trade-off between how many cues do we learn and how long do we need to learn them. And in the case of the very different colors, they go, well, they can be efficient because they can make very accurate choices in a very short amount of time. And with the more similar colors, they can learn the patterns and shapes because they learn them about as fast as the similar colors, so they don’t need to invest any more time in learning them.

[00:09:28.740] – Jacy

So one thing that you talked about in the study about is that your results hint at a mechanism called blocking.

[00:09:37.020] – Anna

Mm-hmm.

[00:09:37.950] – Jacy

Can you tell us what it is, how it helps bees switch strategies, and what it reveals about their learning dynamics?

[00:09:46.700] – Anna

Yeah, so this is an interesting one. Blocking— it’s also Kamin’s blocking effect— is a psychological phenomenon that’s actually been described in many, many different animals. It’s been described in humans and mammals and fish, if I’m not completely wrong, but also many insects and other invertebrates. And I should say it at this point, it’s not entirely clear how it would be implemented in the brain, but it’s a very solid phenomenon, like it’s very robust to test and elicit in many species, even though— and when this is the scientist speaking, there are a few little asterisks, because there is an interesting scientific discussion going on how much this effect relies on the explicit testing conditions and how much it doesn’t. So this is just the asterisk to say I might be slightly simplifying a huge body of literature and discussion around it. But bottom line is it’s a phenomenon that can be observed in the behavior of many different types of animals. And what this phenomenon describes is that when you teach an animal, or when you train an animal to learn one cue, and we’ll stick with the example of a bee in this case, that a color is associated with a reward like sugar water, and they learn this one cue.

[00:11:01.640] – Anna

So let’s say blue is rewarded with sugar water, and they learn it. Then if you present them with a second cue afterwards, after they learned the first one, So let’s say a star shape is also associated with sugar water and is presented at the same time as the first cue. So blue, then many animals would have trouble learning this second cue. So I’ll just say it again because it’s so many things. But if you train an animal, let’s say that sugar water is associated with blue and it learns it, and then you provide it with two cues, the blue one that it already learned and the new one, the star shape. It will have trouble learning that the star shape is also associated with sugar water. And trouble meaning the learning ability might be impaired or it might be completely blocked so that they don’t learn it at all. And that’s why it’s also called blocking. So the idea is learning the second cue is basically blocked by learning the first cue that is presented together with it. Again, that’s something that’s very robustly observed in many animals and we thought that blocking might actually be a mechanism by which the effects that we observed could be implemented in the bees.

[00:12:09.750] – Anna

And I should make like one more explanation of why, why this could be, because in the typical blocking paradigm, you first train cue one, the blue, and then you train the second cue together with the blue, like the pattern or the shape and the star shape and the blue. But in our experiment, we always present the cues together. So blue was always associated with the star shape. But then we did see that the bees learned the colors, the distant colors especially, much, much faster than they learned shapes or patterns. So what we hypothesize is that when they learn these colors much faster, we essentially have the same paradigm. They learn cue 1, which is the color, and then they observe the learned cue, blue, together with the shape or pattern cue, and that learned color cue blocks learning the second cue. Again, we don’t know also from other experiments how this would be exactly implemented in the brain, but it could be a very robust mechanism for basically making this trade-off decision that if you can learn a cue very fast, like the very distant colors yellow and blue, then they would block learning any other cues.

[00:13:23.820] – Anna

And that would have the advantage that the bees wouldn’t basically need to spend time, so to say, learning the other cues, but relying on the learned first cue that they can learn really fast and thus capitalize on efficiency and saving time. Whereas when the cues have similar learning times, like the similar colors, like yellow and orange together with the shapes or the patterns, then there is no blocking because no cue is learned first and then comes the other, they learned together. So they’re not being blocked and they can learn both of them. That’s the idea. And I should say, this is entirely a hypothesis, but it’s a hypothesis that fits very well with what we know about this blocking phenomenon.

[00:14:03.870] – Jacy

Wow, that is so interesting. So thinking about everything that you learned, how might this information help us design better habitats or floral resources for bees in the wild?

[00:14:14.750] – Anna

Yeah, that’s— it’s a really good question and a really important question. And what I should say is I don’t think that from this experiment we can directly infer something about what would be important for the bees’ habitats in the wild. A, because we did do all the experiments in the lab in a very, if you want, sensorily deprived environment. Because yeah, we had these very artificial flowers, also the environment of the bees, there weren’t any other flowers or clutter or smells or wind or anything like that. But I think what we did see, and that could almost be, yeah, I think it’s a positive thing to take away. And that’s also been shown in many other bee studies is that especially the bumblebees that we did investigate are very, very flexible and they’re very good learners. So they are very good at exploiting new food sources that they come across and making new connections about what is a good nectar source and to exploit. And they’re not only relying, so to say, on innate preferences that might have evolved with the natural flowers that they would visit, So I think that’s a very positive thing to take away, that even in our changing floral landscape, so if we put new flowers, so to say, in the environments, or if we supplement flower diversity with seed mixtures, for example, my take on the bumblebees from our experiments would be that they’re actually very flexible in adjusting to new flowers.

[00:15:48.550] – Anna

Now, what we didn’t test, and I think that’s an important point, is the limitations. So we didn’t explicitly test how flexible are they in learning these cue combinations? Where’s the limits of it? How much do they rely on their innate preferences? We had controls for innate preferences in the experiments, but still. And I think, and now I’m taking a bit more of a tangent. I think if we really want to understand how we can design better habitats and floral resources in the wild, and especially how we can also preserve the ones that are there, we need much more data actually from the wild and from observations in the wild. And I can say maybe a tiny bit about what we’re doing, a little bit of that in our lab. I think there are many other labs in the world that do this much more extensively, where we are looking into, for these particular bumblebees, but also for other pollinators, what flowers they actually visit in natural observations. And we’re using citizen science data for that. And we’re working with a German platform that’s called Naturgucker, but an international platform would be iNaturalist, where similar data is collected, which I think is really, really valuable for providing worldwide observations on a scale that no single scientific study can provide on which plants different species of pollinators actually visit.

[00:17:14.900] – Anna

And I think that will give us a much better idea of the general preferences in their natural habitats. And might also give us better hints of which natural habitats are very species-rich and what’s the structure of them that we should preserve.

[00:17:29.660] – Jacy

Thank you, Anna, for giving us a clearer picture of what’s actually happening in bees’ tiny, extraordinary minds as they go about their day. While it sometimes seems random when watching bees forage, it’s fascinating to know that they aren’t just reacting to whatever is in front of them. They’re actively weighing options, switching strategies, and choosing the most efficient path to a good decision. Thanks so much for listening in today. More information about Anna and this study can be found on the website, thebeesknees.website. You can also sign up for our quarterly newsletter there. I’m looking forward to our next conversation. Until then, keep learning.