Colin Allen is a professor of philosophy at the University of California, Santa Barbara, where he specializes in cognitive science, particularly animal cognition and artificial intelligence. Among his many publications, Colin has authored books such as Moral Machines: Teaching Robots Right From Wrong. In his efforts to study diverse intelligences, Colin received Templeton funding for a project called “Joyful by Nature,” which explores the evolution and function of joy in various animals. Colin joins the podcast to discuss the research behind animal emotion and behavior.

When it comes to animal emotions, no person stands out more than the naturalist Jane Goodall; Jane dedicated her life to studying the lives of primates and advocating for their dignity. To learn more about how Jane saw the world and our place in it, we invite you to watch a short video in honor of her winning the Templeton Prize.

Transcripts of our episodes are made available as soon as possible. They are not fully edited for grammar or spelling.

---

Tom: Welcome to the show, Colin.

Colin: Thanks. Thanks for having me.

Tom: I like to ask all my guests a little bit of the formative experiences that led them to where they are today.

So, when did you begin to gravitate towards formal philosophy? Was that early on? You’re reading Plato’s Dialogues as a 10-year-old, or is it something you’re introduced to later?

Colin: It was something I was introduced to later in the school that I was at in the UK. I was getting a little bored with the track I had chosen, which was very mathematically heavy, and I had a teacher who sort of recognized that I might be more interested in philosophy, but sort of got me into it by getting me interested in logic. So that was sort of my first step into thinking about philosophy, but I was always interested in how the mind works. I even toyed with the idea that maybe I should actually go to medical school and be a brain surgeon at some point.

Decided that looked like way too much work, toyed with going into a grad program in comparative psychology or psychology. Thought I was underqualified for that, but philosophy provided the right track.

Tom: From my experience with philosophy, non-human forms of mine are very, a small minority in terms of what philosophers focus on, whether it’s mind and machine, mind and animals.

Was there a, a mentor or a certain set of circumstances that drew you down that road?

Colin: Yeah, I think I was drawn down that road primarily because of who I ended up being alongside when I was a PhD student. So Robert Seyfarth and Dorothy Cheney made huge news in 1980, front page of the New York Times because they had found that the Vervet monkeys in East Africa that they were studying had different kinds of responses to calls that were indicative of a type of predator, even when the predator wasn’t around.

So the calls alone seem to convey information about predator type, and this was touted as a sort of an evolutionary step towards human language or words at least.

Tom: One has to study animal minds to some degree indirectly. Can you tell me the major different kinds of approaches that people take to at least infer and study the minds that animals have?

Colin: Yeah, I mean, the history of science here is kind of interesting because I think that there is different tracks and a somewhat of a pendulum swing, but also some convergence between biological approaches and psychological approaches. So psychologists take them first, tend to look for general principles that apply across all kinds of different species.

You might think of the laws of learning. Thorn Dyke’s Law of effect, right? So they tend to frame things in terms of psychological laws that are general. Biologists, of course, the big thing is diversity, right? Without difference, you don’t get selection happening. And so the framing of the questions is quite different between: What are the general principles versus what are the differences?

And for a long time, I think there was a lot of tension between biological, which mainly means ethological approaches, and comparative psychology, the more psychological approach, which was that the biologists thought that the things being studied in the lab weren’t really how animals behave.

They weren’t even, in some cases, real animals. They were lab bred artifacts, and the psychologists were saying all that field work is really hard to interpret because the experiments are not very good. They’re not controlled. There’s too much going on. You can’t infer anything from them. So that was sort of a major fault line.

The last 15 years, there’s a lot more people who have hybrid training themselves, who actually also have hybrid research programs. They do some of their work in the field and some of their work in the lab. So that old distinction is really broken down in many ways.

Tom: In your experience of what you’ve observed over the last two decades more, where do you see some of the most fruitful questions or most fruitful methods that have revealed how animals think and feel?

Colin: I think it has come about because of this convergence that I was just talking about. So you could point, for instance, to the work of Nikki Clayton in the UK who studied memory and scrub Jays. Scrub Jays are seed caching birds, and then designed a whole bunch of really clever experiments to show just how specific their memories are for what they buried and when they buried it and where they buried it. So that what, when, where combination is a big component of what we think of as episodic memory, right? We remember kind of an ensemble that forms of an event. There’s been continuing discussion though about whether that is episodic memory or episodic like memory because it doesn’t show that the bird itself is imagining itself in that situation.

So is there this kind of me, I was there kind of feel, but then you’ve got this great work by John Crystal in Indiana University with rats showing, actually, they do seem to understand whether or not their information was due to something that they did versus something that was done to them. And so that puts them into the memory picture as well, which is really then Crystal’s much more willing to say episodic memory rather than just episodic-like memory.

So that’s just an example of how I think where we start off thinking about, well, what are these organisms actually doing in the wild, and how can we use that in an experimental way in the lab to tease apart and get more specific about the contents of their particular memories?

Tom: So I know people are very careful that we don’t want to bring in purely human concepts and assign them to animals. We don’t want to anthropomorphize, but I want to ask you a different question. We are animals that think and feel. What are the upsides to us having that quality when we try to study other animals that think and feel?

 Colin: Yeah, I mean it. Gives us a source of ideas about what might be going on in the animals. Of course, we have to be really careful that we’re not anthropomorphizing, as you’ve said, sort of overreading the animal behavior. But also Cameron Buckner, who is a former student of mine, talks about the problem of anthropofabulation.

Our introspection’s not that good, and sometimes we make up stories about why we did something, and we have to make sure that we’re not importing those kinds of made up stories about ourselves into our thinking about animals. So I’ve actually just recently sent off a short commentary on a brain and behavioral sciences paper that’s coming out about fish vision and the evolution of vision throughout the vertebrates, where the authors argue that the big transition is from sea to land, where all of a sudden you can see a lot further.

But it turns out that fish actually see further than we do underwater because many of them use polarization in the light to help resolve objects and distance, and we’re completely blind to that, right? So you can get a completely wrong impression of what the environment perceptual conditions are for fish.

If you just import your own experience from being underwater snorkeling, or scuba diving, or whatever you do there.

Tom: I don’t know how much direct experience you have with this, but I’m wondering. The study of human infants, to what degree does that help mediate between humans studying themselves, but adults studying something that’s very different from them and really hard to infer what is this baby thinking?

Colin: Yeah, no, that’s a great question. And I think there’s actually a lot of crossover and in some slightly different ways. So, one is certain experiments that were first thought of in the context of “could we do with this with animals?” were actually done with infants. Partly because it was hard to think about how you would do the experiment if you couldn’t talk to the research subjects and tell them what to do.

And even with infants, you can give them a little bit of verbal instruction and you have the parents around also to help sort of structure the experiment, right? Nevertheless, I think you can do comparative work where you see what develops when and what’s the sequence in which things develop and how much is it necessary that one thing develops before something else develops?

And you can make some inferences from that as to what the evolutionary order most likely would’ve had to have been as well. So where is there developmental plasticity where the ordering can flip, right? And where is there not? Gives you some clues about the evolution of these things, and then possibly how to look for them in non-human animals.

Tom: I want to ask you about that concept of instinct. I feel like many times, like from a lay person’s perspective, we think an animal is doing or thinking or feeling a certain way, and I can imagine a scientist saying they’re not reflecting on the way things you are, they’re operating by instinct, but it feels like it’s carrying a lot of weight to say that, oh, this complex behavior is being carried out by instinct.

To me, that doesn’t really. Explain what’s going on. It just makes the word instinct that much heavier. Can you help me with understand what instinct is doing?

Colin: Yeah, I mean, exactly right. It’s a long contested term. Going back even to Darwin and before is what do we mean by instinct and does it explain anything?

And there are certainly many people who completely agree with what you said, that it explains rather little, and I tend to agree with that. There are also different notions of instinct floating around. So a highly trained tennis player instinctively returns the ball a particular way. That’s not an innate instinct, that’s a learned instinct, right?

So that distinction turns out to be really important as well. But in the case of animals, the term instinct was generally used to indicate that it wasn’t something that involved any intelligence or learning to pick up, it was just innate. And that notion of innateness itself is very contested.

And there’s a great story about Henry Morgan who’s known these days as the founder of American Anthropology, but his first book was actually on the American Beaver. And the Beaver is very interesting because it builds these complex dams and you sort of wonder how much of this is instinctive and how much of it is learned and reasoned out by the beaver as it’s building the dam .

and Morgan surveyed all the different, uh, definitions that people have given it instinct. And at the very end of this investigation, he says really ironically, that he’s looked at all these definitions. None of them are good for his purposes, but he’s come to the conclusion that instinct is a term that you would use for something that an animal does that, had it been done by a human, you would’ve called it intelligent.

Yeah, right. So it just kind of plugs a hole as it were for something that we can’t really explain. And even saying it’s in innate or built in, it doesn’t really explain how it gets there.

Tom: For an animal to feel something to experience emotions, are there some biological prerequisites or some sort of threshold that needs to be in place?

Colin: Well, you probably have to have a reasonably sophisticated nervous system, but we don’t really know what that is. Beyond that, I think we are talking about quite a few different things when we talk about feeling things. So we might be talking about inner interceptive feelings, or we might be talking about experiences of seeing a certain colored object.

So I think we probably want to give different kinds of answers to questions about when those arose and what’s required for them. So there are some biologists or biologically inspired cognitive scientists who think that interceptive feelings actually go a long way back. It’s just basically part of any living organism’s ability to maintain itself within a viable range and being able to detect when it’s outside that range, so that it adjusts it’s behavior.

So that’s a very kind of low level form of internal feeling, which might be really hard for us to project ourselves into the state of mind where that’s all that you have, but you certainly know that feeling when something’s slightly off, right? It’s not thirst. It’s not hunger, it’s just, eh, don’t quite feel right today. And I think that’s what’s being imagined here as being a very primitive, early form of sentience or feeling.

Tom: The other literary reference I want to pull, and I read this amazing novel in which the protagonist was a honeybee, like a sanitation worker in a beehive. The more I’ve looked into animal behavior, the more curious it became to me of like, what does that sanitation be, think, feel, what decisions can and can’t they make?

Can you just tell me a little bit about, I think certain kinds of bees at least are a common subject of study among animal pathologists.

Colin: So honeybees have long been an interesting species because they have this somewhat sophisticated communication system and it turns out that they’re great learners as well, so you can train them to pay attention to certain colors and shapes. You can test them on how they estimate distance under different conditions.

They have, with just about a million neurons in their brain, some very surprising capacities for making decisions that are sensitive to information that’s available to them. So if your definition of intelligence is adaptive decision making, they seem to do this on the individual level.

And then of course collectively, they do all sorts of things that no individual bee can do too. So there’s sort of a interesting model here for group cognition where the individual components themselves are pretty interesting cognitive systems in themselves. What does a huge aggregate of those manage to do that they couldn’t do individually.

But what it’s like to be a bee? I mean, I think we can say some very outward directed things about their ability to discriminate amongst certain colors, perhaps even their motivation to do certain things. What is rewarding and what’s not. If you really want though an account of what the bee’s consciousness is like, I think we’d have to get a lot more information than we have.

Tom: What emotions seem to be the easiest to study and maybe what ones are particularly really tough to unpack?

Colin: The easiest ones to study are the ones that are easiest to produce, so things like fear, right? You can produce quite easily and reliably, and so they get studied a lot. On the other hand, I’ve got a project currently on joy.

That’s kind of hard to produce, right? I mean, we’ve all had the experience on a bad day. Something that would normally give us joy, won’t produce any joy whatsoever, right? So it becomes a much more difficult thing to study just because it’s much harder to generate.

Tom: I wonder if you can tell me an anecdote or two about clever ways of inducing an or at least observing joy in animals.

Colin: Well, one of the things that we’ve been developing in our project funded by the Templeton World Charity Foundation is a technique that we call the windfall experiment. So this is just one of several things that we’re doing.

So the idea is that the animals trucking along doing something where they get a pretty modest reward for doing something not too difficult. So the modest reward if you’re a chimpanzee, might be a piece of chow. If you’re a Kea Parrot, one of the species we’re working with, it might be just a sunflower seed, and then all of a sudden, without any warning, you get something much better than that.

So for the chimpanzee, it might be 10 really nice grapes instead of one grape. Right. And for the Kea, it might be a dollop of peanut butter rather than a sunflower seed. And so then we look to see, well, how immediately does their expression change? And this could be vocalizations they make, it could be other things that they do, but we’re also interested in how does that affect their subsequent behavior in other tests, in other conditions.

So does this sudden windfall produce more optimism about stimuli that would be neutral between good and bad, for instance. And of course we can do it the other way around too, where they’re getting pretty nice reward all the time, and then on unpredictably on a certain trial, they get nothing or something not so good.

In that case, we’ve seen the Kea literally knock over the apparatus, for instance. That’s an immediate expression. But then the, the question is not just do we get something that looks like annoyance in that case, or joy in the other case, but we’re really trying to tease apart what other effects does this have on the animals decision making, its interaction with other animals and so on as a result of those experiences.

So we want to characterize not a, just a single response, but a whole suite of knock on effects of that treatment.

Tom: I want to pivot to maybe the bridge or if there’s a relationship between animal intelligence and artificial intelligence. And what prompted me to think along these lines, I heard a presentation from David Gruber. The marine biologist studies whale communication and is using some AI software to try to decipher and detect patterns and try to decode, do whales possess a language and a grammar and syntax and whatnot.

But that got me to thinking. Is it going to be easier for us to understand how animals think or understanding how artificial intelligence works? Like where do you see in terms of the frontiers of us grasping this profoundly other, between animals, which you have to make inferences to, and then artificial intelligence, which I don’t have the foggiest idea of how to understand it.

Colin: So it’s a great question and there are people actively working on this. So Marta Helena at the University of Cambridge has also got a TWCF (Templeton World Charity Foundation) funded project in which she and others are trying to develop a suite of tests that you could put animals and robots or AI into and directly compare how the animals do and how the machines do.

We’ve seen, of course, a lot of that between humans and large language models as well. So people are just trying to directly compare the two so we can get some sort of measure of performance on tasks that’s applicable both to animals and machines. But then your question really is, can we say anything about what it’s like to be that AI?

My take on this at the moment is that the AI systems that we have, we know enough about how they work to know that they’re nothing like how animals work. So take your large language model. It does nothing until you prompt it. And then it feeds forward and out comes an answer, and then it sits there and does nothing.

There’s no active mind, there’s no needs that it has, that it’s trying to fulfill. None of that is there. It doesn’t need feelings because it’s got nothing to satisfy, right? It doesn’t have to worry about where it’s power is coming from. None of that is part of the architecture of the system, whereas even a simple bacterium has all of those kinds of concerns.

This is why some people think this kind of feeling of something being right or something being wrong might even go back as far as single cell organisms because it’s just a basic requirement of life. That it’s always having to fight to maintain its existence and get the resources that it needs. And of course, the bigger the system, the more complex those needs become.

And so you’ve got all sorts of feelings that are tied to particular needs that you have as that complex organism. So I think the, the kinds of machines that we’re talking about right now just don’t even get off the ground in terms of being the sorts of active systems that organisms are.

Tom: Yeah, what you said just makes me appreciate the importance of metabolism, keeping yourself alive, getting where you need to go.

Any computer system is not sweating about whether it’s plugged in or not, but the entire history of life could not go a single moment without neglecting metabolism, without neglecting, locomotion, all these different. Sets of activities that must be continuously of concern.

Colin: And that’s sort of what’s ironic I think about human intelligence is that we tend to focus on, can we play chess, can we do mathematics?

These are just like the tip of the tip of the iceberg in terms of being a living organism, of course, you know, being able to pick up things is essential for playing chess. But the reason that we pick up things is because that’s how we eat, right? If you’re a bird, you don’t eat by picking up things you peck at it, right?

So there’s all of these differences make differences into how animals experience the world also.

Tom: Yeah. So maybe think bird parlor games are going to look a lot different than ours.

Colin: Yeah. That’s where I think studying the Kea is really interesting. So, you know, they’re the world’s only alpine parrot from New Zealand and they seem to, again, anthropomorphizing, but this is what we’re trying to study in the Joy project: They seem to just really. Love playing in the snow, especially on bright, sunny days. So we’ve got evidence that they’re emitting a particular call that they use in all sorts of what we take to be high positive affect, high positive emotion situations that comes out more on sunny days, that all other things being equaled than on cloudy days.

So you can begin to say, actually they get some joy or pleasure out of the weather conditions, and that makes them seem a little bit more like us in some ways.

Tom: Yeah. What kind of project approaches, even imagine discoveries over the next, say, five to 10 years, do you get most excited about or eagerly anticipate?

Colin: You had mentioned using AI to basically find patterns in communication, and I think more generally the use of computers to really speed up certain kinds of research. So one of my collaborators on the Joy Project, Alex Taylor, done a lot of work on bird cognition, they looked at New Caledonian crows, which make tools and testing to see whether or not they are capable of making things from a kind of mental template or memory of what they’re supposed to be making, rather than having to have a model for it right in front of them that they’re copying in some way, right. The finding in that paper is, yeah, they can do that.

But the connection back to AI is that the task involved tearing strips off piece of paper. And so they had dozens and dozens and potentially in a much bigger experiment, one would have hundreds or thousands of these things to measure the size of. In the old day you had to measure it all by hand or find some other clever way to estimate it.

Now they just give it to a computer vision system and it does it for them. Right? And so I see this all over the place and another project that I had involvement with was actually funded by John Templeton Foundation was on the evolution of human cognition, but we were interested in humans 2 million years ago, or hominins 2 million years ago, and archeologists are sitting on huge piles of what they call debitage, which is what’s left over after you’ve made a stone tool from a piece of rock that literally thousands, in some cases I saw one lab where they had a pile that was about 15 feet long and about three feet high and about four feet wide of little bits of rock that they’d all individually numbered. They’d know where they came from. Imagine if you could actually scan all that and then have a computer figure out how to put it all back together again.

You would know what tool had been produced right. From the bits that were left behind. So there’s a research idea for somebody that would’ve been completely impractical without any kind of computational data analysis, call it AI, if you will, that I think is quite possible to come and would also then give us, uh, real insight into the cognitive capacities of hominins one and a half, 2 million years ago.

Tom: Among the students you work with, whether undergrads or grad students, where do you see the trends or areas that they are most excited about in terms of the next generation of research?

Colin: Well, I think that there is a lot of excitement on both the AI side and the animal cognition side. So I’ve got a couple of students here at Santa Barbara, one of whom is going to write a dissertation on the concept of grief and how it applies in animals. And this follows on by other philosophical work. So Suzanne Monso just published a really nice book on concepts of death in animals. And so it’s a natural sort of springboard for them thinking about, well, what’s the emotional aspect of having a concept of death? And can we study that in animals?

And then another student who’s very interested in what way can we talk about these AIs as participants in some kind of moral decision making? Because people are going to use them. They’re already using them for advice, sometimes quite poorly. But there’s a sort of a deeper question about what kind of thing should be allowed into the sorts of back and forth that you and I might have over a moral decision?

If I bring in something that I got off ChatGPT, how should you and I treat that towards our moral deliberation? And if we don’t know where it’s getting. Its ideas from so to speak, then that makes it also much harder to figure out whether or not we should trust it.

So I think these are really interesting issues very much driven by current developments in both the animal cognition side and the artificial intelligence side.

Tom: What do you see as kind of ultimate goals or implications? What do we do with this knowledge as do you look forward

Colin: Studying animal cognition is one of those areas that’s just intrinsically interesting.

Whether it has any practical implications is a secondary question. Some people would say they do it because they know that it has implications for the animal’s lives themselves or for conservation of those animals. Others would say they do it because it helps us understand certain things about our own emotion, cognition, consciousness, and so on.

And some of it’s very clinically oriented. So, before I said fear gets studied because it’s easier to generate, it also gets studied because it’s clinically significant, right? People with PTSD and so on, fear and anxiety are big issues. So I think there’s a just a huge variety of motivations for why we do this.

My own motivations though, are much more just curiosity driven, and I think science needs curiosity driven research. Not always just application driven research.

Tom: Well, Colin, I appreciate you taking time to talk to me today, and especially a field is richly interdisciplinary as animal cognition, and I look forward to seeing what comes.

Colin: My pleasure, and thanks very much for inviting me to do this.

Tom: Now that you’ve listened to this episode, I have a question for you. If you had the opportunity to feel what it’s like to be another creature, which one would you choose and why?