THE CONNECTOMIC REVOLUTION
The main question I’m asking myself at the moment is about the nature of the animal mind, and how minds and conscious minds evolved. The perspective I’m taking on that is to try to examine the mind's mechanisms of behavior in organisms that are far simpler than ours.
I’ve got a particular focus on insects, specifically on the honey bee. For me, it remains a live question as to whether we can think of the honey bee as having any kind of mind, or if it's more appropriate to think of it as something more mechanistic, more robotic. I tend to lean towards thinking of the honey bee as being a conscious agent, certainly a cognitively effective agent. That’s the biggest question I’m exploring for myself.
There’s always been an interest in animals, natural history, and animal behavior. Insects have always had this particular point of tension because they are unusually inaccessible compared to so many other animals. When we look at things like mammals and dogs, we are so drawn to empathize with them that it tends to mask so much. When we’re looking at something like an insect, they’re doing so much, but their faces are completely expressionless and their bodies are completely alien to ours. They operate on a completely different scale. You cannot empathize or emote. It’s not immediately clear what they are, whether they’re an entity or whether they’re a mechanism.
That interest was always there. That’s what led to my PhD, which was looking at mechanisms of learning and memory in fruit flies. That was fascinating, but there were limits to fruit fly behavior. Then there was this other insect: the honey bee. As I was working on fruit flies, this army of papers were coming out making these astonishing claims about what bees were capable of. Bees were not just doing simple learning, they were doing concept learning and they had social behavior. One thing after another came up. An insect that can learn abstract concepts and has a concept of number completely amazed me.
We have the same tiny insect with a brain on a scale that is comprehensible and accessible, and it was also clear at this point that there would soon be a genome for this organism. I thought, okay, if we want to ask fundamental questions of the mechanisms of behavior, we have to work with something like a honey bee, which has this complexity of behavior and a brain that’s simple enough that we should be able to get a mechanism, and we will be able to pool genomic resources. And that’s why eighteen years ago I started working with honey bees.
The overarching theme of what I’m trying to do at the moment is to get a holistic systems-level understanding of how the insect brain works. That involves engaging with the question of whether we can think about that system as being conscious in any way. It’s an enormously collaborative project. We’re taking it apart in the ways that we can sensibly also deconstruct the insect brain. The insect brain has this beautiful modularity, which means that we can focus on either certain questions or certain regions, but we can still keep an eye on how the whole system works.
The collaborative team I’m working with involves a major group at Sheffield University in the UK. I’ve also got colleagues still in Illinois and at Queen Mary in London, who I’m working with on this project. Of the disciplines involved, we span across genomics, computational modeling, mathematical modeling, biorobotic modeling, behavior, neurochemistry, and neurogenomics. We’ve got to bring every tool that we have into this problem, and the modeling is a core component of what we’re doing.
We’re dealing with a complex system and highly complex questions. Compared to the way I used to do experiments, where you would frame a simple falsifiable hypothesis and move on, it’s hard to deconstruct a question like how a nervous system, how a neural system works, to deconstruct that down into simple, purely falsifiable hypotheses.
Another way to approach it is to try to imagine how the system could work, then use modeling, whether it’s biorobotic or computational, as tools to explore whether your imagined system could function, and then use that model to help frame a question that you can ask of the real system. That becomes an indirect test of whether your assumptions are right. But if you keep iterating that process, and if you keep the communication open, and if you don’t become overly attached to your model, you can still make progress towards how the real system works. I found that very constructive and informative for asking these questions about how so-called higher order cognitive systems could be operating in something like a bee brain.
I describe myself as a comparative neuroscientist. This discipline of comparative neuroscience is itself quite young, but even within my career it has transformed completely beyond recognition because of developments that are happening in all surrounding disciplines. Across my career we’ve seen the advent of genomics and proteomics, which have evolved from being exclusive tools that would have taken years to develop, to enabling you to ask informatic questions and get answers in the time scale of months, and at the kind of cost that any lab can afford. That’s completely changed our perspective on the relationship between genes and behavior.
An even more recent and exciting revolution happening now is this connectomic revolution, where we’re able to map in exquisite detail the connections of a part of the brain, and soon even an entire insect brain. It’s giving us absolute answers to questions that we would have debated even just a few years ago; for example, does the insect brain work as an integrated system? And because we now have a draft of a connectome for the full insect brain, we can absolutely answer that question. That completely changes not just the questions that we’re asking, but our capacity to answer questions. There’s a whole new generation of questions that become accessible.
When I say a connectome, what I mean is an absolute map of the neural connections in a brain. That’s not a trivial problem. It's okay at one level to, for example with a light microscope, get a sense of the structure of neurons, to reconstruct some neurons and see where they go, but knowing which neurons connect with other neurons requires another level of detail. You need electron microscopy to look at the synapses.
What we have now, thanks to advances in microscopy, image processing, and AI, is the ability to generate the stack of electron micrograph sections for an entire brain or brain region. With better image processing, better AI, we can work through that stack, reconstruct the neurons, spot the synapses, and make a statement about functional connections between neurons. We don't quite know the strength, we may not know whether they're excitatory or inhibitory, sometimes we can make a guess. Even with that connection information, we can say that we categorically know what flow of information is capable through this brain even if we would need further study to know what flow of information is happening through the brain. That connectome means we now have the detail of the map. With the detail of the map, we can work back to asking what the essence of the map is and what’s important in this map for how the system functions.
In terms of what the goal is, I’ll address that first of all in thinking about the contribution that a connectome can make. I talked earlier about how modeling is now such an important perspective to advance an understanding of neuroscience when you’re looking at a systems level. The problem with a model is making sure that you’re modeling a brain and not your concept of a brain. The problem with a model is making sure that you can relate it to the reality of the system. The benefit of a connectome is it helps you ground into the reality of the system that you’re trying to model. If you know the connections don’t exist between regions, they can’t be in your model at all. This is a huge benefit of connectomics. It helps you ground your modeling into the biology of the system that you’re trying to understand.
I would like to ultimately have a sense of how minds work, and given that I do see myself as an evolutionary biologist, how minds have evolved. I don’t believe that we could answer either of those questions by looking at the human mind alone in isolation, particularly for the evolutionary dimension. We need to be looking at a range of minds. That might help us understand the progression of how a mind could have evolved from something very simple to something of phenomenal complexity, like our own.
Looking at the minds of smaller and simpler organisms like an insect could contribute to our understanding of our own minds. We experience ourselves as conscious entities; we wouldn’t describe ourselves as being a neural mechanism. Certainly, our perception of ourselves involves something greater than a neural mechanism; it involves a mind. I’m not a dualist, so I see the mind as being a product of the neural mechanisms. I see the neural mechanisms as being a system that has evolved. And if I accept that the mind is a product of the neural mechanisms, then the mind itself evolved. I certainly don’t think that we’re the only mindful entity on this planet. It’s a live question as to what other organisms also possess a mind as well as possessing neural mechanisms. The tension in that question is why I’m studying something like a honey bee. It’s not because I’m wedded to believe that the honey bee is mindful, but I do think the honey bee could go either way. Is it more appropriate to describe that as a neural mechanism, or is there a level at which we could attribute something like a mind to something like a honey bee?
When I say a neural mechanism, I mean an integrated neural system, a nervous system. I’ve spoken so far about my work on comparative neuroscience and the pure fascination of studying bee behavior. There are two streams to the research in my lab. Physically more than half of my lab is also studying what we call bee health and welfare. Since 2007, there’s been enormous concern about the global populations of bees, of pollinators, of insects, in general. There’s been an enormous response from anybody who works on bees and pollinators to react to that challenge, and I’ve been part of that. Half of my lab is also invested in investigating bee health and how we can make more informed interventions to support that.
It’s easier to fund the more applied research. That’s obvious. There is an immediate imperative there, and I absolutely endorse that. It’s not as easy to fund the comparative neuroscience. But that’s inevitable. These are questions that we should be working harder to fund. If we think we have something to say here, we need to make our arguments very clearly and fight for the funding for it.
In terms of the comparative neuroscience side of my work, I’m working towards trying to comprehend how the honey bee is able to so effectively and autonomously organize all of its behavior and demonstrate such a high level and startling range of cognitive capacities that genuinely do rival those of mammals. They’re central-place foragers, they can solve tasks of metacognition, they can do all these things. For example, we can set the bee a challenge by giving it a task that is sometimes easy to solve, or a task where the information is ambiguous. If the bee is punished for getting a task wrong where the information was ambiguous and you give it the choice to opt out of the task, it is more likely to opt out when the information is of low quality or unavailable. That’s been interpreted as metacognition. That’s an awareness of how much you know and adjusting your behavior adaptively because of that.
We can still explain that capacity in terms of the nervous structures of the bee, but this is a kind of behavior they manifest. They manifest abstract concept learning. They can learn features that are not bound to a stimulus, things like bigger than, smaller than, above, below. They seem to have a basic concept of numerosity, certainly quantities of more than/less than. It’s been recently claimed they have a concept of zero. These are all this family of traits that at one time were considered to be the thing that separated humans from all other animals, and then was slowly recognized to appear in primates and then large-brained mammals. And then suddenly we’re recognizing that something like a honey bee, with less than a million neurons, is able to do all of these things.
Designing an experiment a bee can opt out of is remarkably easy because there’s a very robust part of bee behavior. Bees like to collect nectar and take it home, so you can train them in that way to collect a small amount of sugar reward. You can train them into various sorts of apparatus and they will very readily associatively learn a whole range of features that will be associated with a sugar reward. They learn that with incredible speed, and their memories are lifelong.
An example of an experiment could be a simple chamber that you train a bee to fly into, where there are two stimuli, one of which offers a small volume of sugar solution and the other offers a small volume of a quinine solution, which tastes bitter to a bee. The bee will fly in of its own volition once it’s learned to enter the chamber. If it can distinguish the two stimuli, which can be colors, odors, quite abstract symbols, it will learn quickly once it’s sampled the sugar which stimulus indicates the sugar. And then once it’s sampled, you can release it. It will take the sugar back to the hive and it will come back.
We identify the bees with a paint mark or we stick number tags on them. These days we might attach a radio frequency tag to them. It’s very easy for us to tell the bees apart. The nice thing about bees when they’re feeding at sugar is that they’re very clam, so you can easily come up behind them and put a dab of paint on their back. Do I ever form attachments to a bee? The scientist in me is obviously going to say, no of course not, and the experimental realist says, yes, of course I do. In the course of the metacognition experiments, we’ll work with the same individual over hundreds of trials, and what you learn in that process is that there is enormous variability between these now individually identifiable bees and how they approach the task that you’re giving them. You end up really rooting for them. This is a large part of the reason why I've been putting forward this conversation of whether we can talk about the bee as in any way experiential.
When you work as closely with bees as I’ve done for years, individually marking and tracking them over the course of days or sometimes a week—they don’t live every long—then you start to see beyond the hard exoskeleton and the expressionless face and the incredibly alien appearance, and you start to recognize just how much individualistic behavioral variation there is within this organism. You start to recognize the sophistication of the behavioral decisions that they are making moment by moment when they’re assessing not just the tasks that I set them, but doing what evolution designed them to do, which is to gather nectar and pollen from flowers. When you think about that, that is a phenomenally challenging task not just cognitively and spatially, but energetically. We have minute flowers scattered all across the environment of which only a minute proportion will contain any food at a given point in time. You’ve got to find them, you’ve got to identify which are the ones that are going to give you a reward, and you have got to make a profit and take it home because if you don’t, your colony will die.
A lot of the reasons why the honey bee is such an extraordinary animal to work with is because humans have had an association with the bee for at least 7,000 years. It’s an open question as to whether the bee has been domesticated. Up until very recently, the bee has always had complete contact with the wild population. It’s a wild organism. We have just got good at giving it what it needs in terms of what hive structure is perfect for it so that we can take off the surplus of honey. But because we’ve got this long association of working with bees, it’s easy to open up a hive and take out a frame of bees just before they’re about to emerge as adults. We often pull a frame at that point and then we can emerge them overnight in an incubator. Then we get these beautiful fluffy little day-old bees the next morning. Their cuticle is too soft to sting, it’s too soft to fly, so you’ve got a few hours where you can pick them up and you can put a paint mark or a number tag on them and then put them back in their hive. And they will be accepted into their hive totally naturally. They will go about their entire life in the hive, but we know who they are, we know how old they are, and we can follow them. We can study them in their natural environment as well as having them participate in whatever various cognitive tasks we wanted to challenge them with.
Typically for me, an experiment I would set up is an observation hive. I set it up under the glass, two or three or four frames of honeycomb covered in bees, and a queen. That’s big enough to be a completely functioning hive, but it’s under glass. I keep it in a warm room, which allows me to study exactly what’s going on and who’s who. That doesn't disturb the bees as long as I keep the light level in the room low. They’re under glass, but they’re fine. That doesn’t worry them.
There’s a tube through the wall of the room that allows them to freely come and go. And to a bee, that’s like having a hive in a tree hollow; it’s perfect for them. So, they’re flying in and out of their wild environment. Outside, I might put a sugar water feeder. Bees are good at finding things that give sugar. So, they will find that themselves and they’ll land. When a bee from my hive has landed, as she’s feeding I might pick her up gently and put a little mark on her back and put her back on the feeder, and she’ll be okay with that. Then I know who’s who. Once she’s marked at that feeder, I can train her with little drops of sugar or by moving the feeder either somewhere else or into a specific testing chamber. Then I’ve got a known bee: I know where she came from, I know how old she is, and she can then participate in whatever learning tasks we’re setting.
Because we use the word queen—the Egyptians use the word king—we have a misconception of the role of the queen in the society. The queen is usually the only reproductive in a honey bee colony. She’s specialized entirely to that reproductive role. It’s not that she’s any way directing the society; it’s more accurate to say that the behavior and activity of the queen is directed by the workers. The queen is essentially an egg-laying machine. She is fed unlimited high-protein, high-carbohydrate food by the nurse bees that tend to her. She is provided with an array of perfectly prepared cells to lay eggs in. She will lay as many eggs as she can, and the colony will raise as many of those eggs as they can in the course of the day. But the queen is not ruling the show. She only flies once in her life. She will leave the hive on a mating flight; she’ll be mated by up to twenty male bees, in the case of the honey bee, and then she stores that semen for the rest of her life. That is the role of the queen. She is the reproductive, but she is not the ruler of the colony.
Many societies have attached this sense of royalty, and I think that as much reflects that we see the order inside the honey bee society and we assume that there must be some sort of structure that maintains that order. We see this one individual who is bigger and we anthropomorphize that that somehow must be their leader. But no, there is no way that it’s appropriate to say that the queen has any leadership role in a honey bee society.
A honey bee queen would live these days two to three years, and it's getting shorter. It’s not that long ago that if you read the older books, they would report that queens would live up to seven years. We’re not seeing queens last that long now. It’s more common for queens to be replaced every two to three years. All the worker honey bees are female and the queen is female—it’s a matriarchal society.
A worker honey bee in summer will live a total of nine weeks, if you’re lucky. There are three weeks of development from an egg to an adult, and there are typically two to three weeks inside the hive, which is a very protected environment. In the hive, they’re rearing the next generation of workers to come through. And then they start to forage. Again, we would have said not that long ago that a forager could live two to three weeks of foraging before they wear themselves out. We’re increasingly finding now that the life of a forager is probably seven to ten days. It’s much shorter than that. Foraging for a bee is always very challenging, energetically and cognitively. We’re making it harder for bees at the moment.
There are a lot of reasons why it’s a harder life to be a forager bee. Most current environments are enormously disturbed to what a honey bee would be used to foraging in. Often, they’re agricultural or semi-agricultural. The distribution of wildflowers, if there are any, is enormously reduced. That means the availability of food and diversity of food is way down compared to what bees would be used to. They’re having to fly further for fewer resources, there's a higher risk of them running out of power before they’ve even found enough fuel to fuel themselves, let alone get back home.
As well as that, many of our environments are contaminated with pesticide. A wakeup call has been the realization of the very low levels of pesticide that can be damaging to a forager honey bee. We used to think that if the pesticide didn’t kill the bee, that would be okay. What we’re realizing now is that a bee doesn’t need to be killed. All it needs is to be damaged, either its immune system damaged or its brain damaged. If it’s damaged sufficiently poorly that it cannot make it back home, that bee is as good as dead. From the perspective of the colony, it is dead. From the perspective of the bee, it will be lucky to last a night if it can’t make it back home. It’s these sub-lethal effects of either diseases or of pesticides that are impacting bees’ longevity in the environment and on the amount they can contribute back to their colony.
In terms of what could be done, a shift to organic farming methods would eliminate pesticide use. That could only possibly be of enormous benefit to bees. It’s an open question whether we could ever transition to an entirely organic method of food production given the weight of human populations that depend on agriculture. Even if we couldn’t, we could transition to an agricultural method that is less dependent on pesticides. That would involve what we would call integrated pest management strategies, things that encourage the use of natural enemies of crop pests rather than purely using pesticides, things that replaced field boundaries and margins of fields, and that broke up the crop monocultures, thereby reducing the likelihood of a pest outbreak.
These kinds of strategies that we’ve known for twenty, thirty years reduce dependency on pesticide. If we could start to shift our agriculture model to roll these strategies out, clearly, we’re going to have a better performing pollinator population, which will improve crop yields. We can have alternative agricultural models that are far more friendly to our bees and the pollinators that we ultimately rely on as part of the ecosystem that is generating our food crops.
The two things that are standing in the way of this kind of progress are inertia and current financial models. We’ve become very good at mass producing food in large volumes cheaply, but there is an environmental cost to that. We’ve been willing to pay that environmental cost so far. We’re reaching a point where the environmental cost is about to bite us really badly in the ass. I would hope that we could transition to alternative models that I believe are viable before it really bites us. But we need to be willing to pay more for our food, and we will need to be smarter in how we’re using food and tolerate less food waste. If we do that, I hope—I tend to be an optimist—we could transition to a more viable sustainable model without a revolution. I don’t think we need brand new technologies or to throw out models. We’ve got a lot of things we could do.
The concern about pollinator populations, insect populations and bee populations specifically, is a global concern. The way in which regions have adapted is variable. There is enormous concern about pesticides and the new neonicotinoids pesticides in Europe to the extent that Europe has banned the use of some neonicotinoids because of their concern about declining bee populations. That’s a constructive move, to be honest.
It’s also fair to say that not all parts of the world are hit equally by this problem. The most intensively agricultural, intensively industrialized parts of the world are suffering most. That’s very telling. This almost proves a natural experiment. My home country of Australia, so far, has done very well in terms of our bee populations. We’re not without problems, but our bees are faring far better. There are two major reasons for that. One is that Australia is the only country in the world that doesn’t have the honey bee pest Varroa destructor. This is a parasitic mite that jumped species onto the European honey bee from its sister species 100 years ago. That was thanks to human movement of bee populations. And then thanks to human movement of bee populations, we spread this emergent pest across the entire world, apart from Australia. Australia is an island continent that, remarkably, hasn’t got it yet, but we’re terrified we’ll get it.
The other reason Australia is faring better than most parts of the world is that we have these beautiful tracts of unspoiled native bushland across huge parts of the country that surround our agricultural areas in a meaningful and significant way. These are areas of the world that have never been cleared, never been sprayed with pesticide. They tend to be dominated by eucalypt trees, and eucalypt trees are flowering trees. In Australia, our bees have what bees would ideally want, which is natural, diverse wildflowers on which they can forage. Whenever possible, Australian beekeepers put their bees close to bushland, which is why we get some of the best honeys in the world.
I don’t think there’s much of a tension between the different dimensions of my work, the dimensions that study the bee as a model for comparative neuroscience and the dimensions that study honey bee health and welfare and its operation in our pollination system. First of all, there are interesting points at which the questions I look at intersect. Effective pollination, effective foraging is itself a cognitive task. The bee has evolved to solve phenomenally well. From that perspective, the neural mechanisms I like to study keep coming into play in terms of understanding what’s needed to forage well and what can go wrong in the bee brain that can cause it to forage poorly. That intersection is very rich.
Another point where there could be tension is the fact that I’m a beekeeper and I'm here talking about whether bees are mindful. Is it right that I’m a beekeeper? Unlike many forms of management of organisms for food production, beekeeping is the ultimate humanist activity. You can rear bees for honey production in a way that no bee is harmed in the production of the honey. You can organize the colonies so that they will naturally accumulate a very significant surplus that you can remove without harming the bees.
In that process, you’re giving the bees the perfect nest box, the disease-free environment, the completely open natural foraging situation that they would need to carry out the lifestyle they have evolved to carry out. You’re not harming them by harvesting the honey. So, I see this wonderful synergy rather than a point of tension between the different sides of my work.
