Amalia Bastos is a biologist and photographer who is currently a Ph.D candidate at The University of Auckland. She is part of the Animal Minds lab and work with three different species: dogs, kea, and New Caledonian crows. Her main interest is on how evolutionary pressures have shaped the minds of different species. Her PhD thesis focuses on the signature-testing approach, which aims to identify which cognitive processes animals use to understand their environment.
+ Full Transcript
John Bailer: We've all heard the stories about sign language-speaking great apes and problem-solving dolphins, but which animal species exhibit statistical thinking abilities and probability-based inference? That is the focus of this episode of Stats and Stories, where we explore the statistics behind the stories and the stories behind the statistics. I'm John Bailer. Stats and Stories is a production of Miami University's Departments of Statistics and Media, Journalism and Film, as well as the American Statistical Association. Joining me is Richard Campbell, former chair of Miami's Media, Journalism and Film department. Rosemary Pennington is away today. Our guest on this episode is Amalia Bastos. Bastos is a Ph.D. candidate at the University of Auckland who has been working with animals to see how evolutionary pressures have shaped the minds of different species. She recently co-authored a paper describing experiments that demonstrate that a parrot, the New Zealand Kea, is able to use data to make decisions. Amalia thank you so much for being here today.
Amalia Bastos: Thank you.
Bailer: Amalia, what motivated you to study statistical inference in animals? I mean I think that I sort of study them in animals, too, but they tend to be higher primates.
Bastos: Well, that’s exactly why, right? So, we know that humans are very good at using probability and we also know that apes are good at using probability. What we didn't know is that birds might be able to do the same thing. But kea show behaviors that suggest that they are very intelligent, so you want to see if they might be able to do the same sort of statistical thinking, and so this is the first time that it was applied to birds in the same way that it’s been applied to humans.
Richard Campbell: How did you pick the kea? I only know kea as a scrabble word, so I’m learning about- it’s a good scrabble word with a k. So why not other parrots? I know parrots are smart generally, but why the kea?
Bastos: I think they are smart generally. The kea happens to be the only truly omnivorous parrot so there the only ones that will each mate and hunt in some sense so you probably know about them digging into the backs of sheep and eating the fat off the sheep’s back, so they do that. That’s what the kea is famous for in New Zealand. And until the ’70s they were mass shot for it. The farmers would kill them; they killed over 150,000 birds.
Campbell: No, I’ve never heard that before.
Bastos: Yeah, because they would attack the sheep. It’s actually a minor problem, but yeah. They don’t do it that much, but the farmers blew it out of proportion and killed a bunch of keas. But you know, there’s a hypothesis that there’s a theory that the more omnivorous species might be smarter than specialists, so there’s some reason to believe that kea are particularly smart. And another reason is that they live in large groups. They have [inaudible] so there might be 50 birds and they come and go as they please so they know all these other individuals as well, and yeah the social brain hypothesis done by our data suggests that intelligence is involved to keep up with the sort of complex social structure.
Bailer: So how do you know if someone knows anything about statistical inference? How do you even formulate the properties that you investigate?
Bastos: Yeah, I mean the fun thing is that with humans you can just ask them, right? But with animals, you can’t do that. So quite a bit of design comes into this and we actually based a lot of our studies on what they’ve done with infants. So, we know that pre-verbal infants can do probability to some extent and so we sort of adapted what they’ve done with infants to the birds, and what we do there is give them two options to pick from. So, with infants, they usually do a time looking paradigm, which is they [inaudible] surprised. So infants will look for longer events that surprise them, and once they get a little bit older they can sort of crawl towards things if you give them choices, so that’s what we do with the birds; we had two sampling events they could pick from.
Campbell: So, at what point did you know that you had something really significant and how did you know that?
Bastos: To be honest the probability itself wasn’t that shocking to me, I had an expectation that they could do it to some extent, we didn’t know that they would statistical inference to do it, we thought that maybe they would use simpler associative to do it was surprising when they passed both conditions that are controls for these studies I think the most surprising thing though was when they started to integrate information, and yeah the first time that a bird got the social experiment right then I rang up my supervisor and said Alex they just passed this and they’re not supposed to pass this. And Alex was like did you run everything correctly? I was like yes. So, we have 30 birds, not all of them are cooperative though. We have two more adults-
Campbell: How many birds were you working with?
Bastos: Don’t want to do any work because they’re too busy with their social relationships. And we have a juvenile that doesn’t know how to work yet. So, by the time we finished all the training, which is quite intensive, we were down to six birds that would come down regularly enough to be tested.
Bastos: Yeah I mean-
Campbell: And why is this significant with just six birds?
Bastos: So what we did a sort of individual-level analysis and so if that specific bird passes or not and the reason for that is that we’re more interested in capacity than average ability in our field, which is quite different from regular psychology. We’re interested in proofs of concepts like can an animal do this or not, and like we don’t really care as much as psychologists if it’s an average thing across all kea or across all birds; it's more about do they have that ability.
Bailer: So, could you just give a sketch of some of these experiments that you’ve conducted? Just to paint that verbal picture for some of the listeners.
Campbell: And explain token preference tests, I think that’s what you’ll talk about-
Bastos: [Overlapping talk] [inaudible] tokens which means they bring them back to us and they would get a food reward if they brought the correct thing back.
Bailer: So, what motivates them? What kind of food motivates them?
Bastos: Cat food actually, so if you yeah, [laughter] we have fed them [crosstalk] [dog barks] but yeah if you just have cat food and soak it overnight it becomes a starchy mess and they really like that.
Bailer: Okay.
Bastos: Yeah so, they’ll do quite a bit of work for it actually. So, they will bring back black tokens for it, but they know if they bring back an orange token they get nothing. So that’s the very first step of training. That was like four years ago I think that they were trained to exchange tokens, so it’s a long process, and then we spent the next two years trying to get them how to understand hands.
Campbell: Wow.
Bastos: Because to apes and to humans, hands are quite intuitive. You see someone moving something in their hand and you know that you know hands can grasp things and hide things with birds it’s not so obvious so because they use their beaks they don’t really use their feet to hide things, and so that took us I think three different attempts of three techniques to try to teach them how to understand that hands hold and carry things around. Once that was figured out, [laughter] yeah no, it’s quite a long process. But once they understood hands they were really quite good at it, and then we showed them a set up where we have two jars and a wooden board and there’s a plexiglass between us and the birds, and the only reason for the plexiglass is so they don’t start making choices before we’ve presented both options because they get very fidgety. They’re a bit like toddlers when you’re working with them. And so we sample from one jar with one hand, present a closed fist, and from the other jar with the other hand present another closed fist and then we could either present them parallel so just bring them up over the plexiglass or cross them over, so the birds could track whether the hand had crossed over and was presented parallel, and the reason for that it wasn’t just picking the same side of the jar they wanted. They weren’t picking the jar, they were picking the sample, so that’s the first step.
Campbell: We should- okay, and we should remind people video and where you can watch this.
Bailer: We’re definitely going to link this.
Bastos: Cool there’s also our video which explains it in more scientific terms step by step, so if you guys want to link that too, that’s a more- it’s just me narrating through the steps as well, so.
Bailer: Yeah I thought that was amazing in and of itself but then the physical divider in the second step and recognizing that some didn’t count, I mean essentially discounting the things that were divided. And then the last bit where someone was purposely sampling the black reward, that must have blown your mind when you saw that.
Bastos: Yes, it did. I mean when they started picking the biased sampler, that’s when I rang my supervisor.
Bailer: That is just so cool. I can’t imagine what would be- that was the most surprising aspect that you have done. You were saying that you sort of expected the other pieces to come in.
Bastos: Yes, I mean I expected the first experiment, which was just can they use probability first off, so I thought maybe they can do that. The physical constraints, we were like okay well yeah maybe you can, I mean you do figure out how to use complex things in our experiments and you’ve dealt with a bunch of different physical things and, you know, in the wild there’s some evidence that they might use tools, so that’s not mega surprising. And then when they were actually reading cues from humans on whether humans are looking into a jar or not and whether humans might be biased or not, that was very interesting.
Bailer: I think I read in one of the background pieces that we looked at that you were wearing mirrored sunglasses. When I first saw this, I was thinking about- was it the story of Clever Hans, the horse that would count, and the story was ultimately debunked because the horse was picking up their cues.
Bastos: Yeah so the horse was attending to breathing patterns and posture and stuff like that. So, what we had I actually couldn’t test for much of the experiment, we had the blind testers so basically volunteers who were doing their honors thesis or they were visiting students and they would conduct the test. I could train but I couldn’t test because I knew what to expect.
Bailer: Oh, so you were- that’s awesome, so you were definitely controlling for some of these other cues that you might be giving.
Bastos: I mean I would- I unblinded them at the end and they were very surprised by what the birds did, but they had no idea what was going on so they weren’t as surprised as I was, it was just quietly trying to keep my enthusiasm so they would pick up on the birds are doing something interesting.
Campbell: How did- I noticed- I read the piece in Forbes and in The Guardian, so what’s the process of getting those general publications interested in this study? How did that happen?
Bastos: They actually contacted me, so it’s a good question. The university put out a press release and put out a press release and people started emailing me and there was three or four days where most of what I did was talking to the media. So yeah there was a lot of trying to be with the kea because we were still working with the kea then. So I was going to the Aviary helping the volunteers, teaching the students and then going okay, I’m just going to dip to the back of the aviary for a second and I need to take a phone call and then I was just trying to juggle the two things, but they mostly contacted me.
Campbell: And what did you think of their coverage, their initial coverage? One of the things we ask guests a lot is what are journalists doing well and what can they do better? How did they represent your work?
Bastos: I think they did quite well, I mean The Guardian one is good. I think it’s quite a difficult one to explain because you know there are three sets and there’s quite a lot of intricacy involved, but I think most journalists actually did really well. Yeah no, I think some of them got slightly caught up on the detail, but they did a really good job.
Bailer: Yeah I was going to ask the very same question, Richard. From reading through it I thought they seemed to cover the scientific paper pretty well. It seems like they captured the essence of it, and I think they loved having those pictures to be able to include the two conditions that were being selected and how that was done.
Bastos: I had some trouble with the pictures originally because the nature communications editor came back to me and said do you have any pictures of the kea, and can you sign a release? I said I do have pictures, but I can't sign a release for every single one of them because there are over 50, so you can go through the folder and pick what you think is good. So, I just sent them a batch of photos and signed one release for all of them.
Bailer: Well that is awesome. So, you think that you would see this in other bird species?
Bastos: Yes, I mean I wouldn’t be surprised. Other birds have complex social structures as well, and I don’t think the omnivore diet is enough to really generate this level of intelligence. So, I mean, birds like [inaudible] cockatoos and African gray parrots those birds with large brains and lots of neurons probably would be the next ones we would see.
Campbell: We’ll have to reevaluate the term birdbrain after all your work, right? It changes everything.
Bastos: Yeah that’s a popular one.
Bailer: You’re listening to Stats and Stories and today we’re talking with Amalia Bastos, researcher in the animal minds lab at the University of Auckland. Amalia, you’ve worked with other species including dogs and new Caledonian crows, what are some of the other kinds of research questions you’re exploring with these other species?
Bastos: Yeah so with dogs we’re very interested in domestication and social cognition because we think that their brains have been shaped, to some extent, by the fact that we keep them as pets and have done for thousands of years. So, we have some interesting social experiments with the dogs. We had a paper come out recently showing that they yawn in response to people but it’s not emotional contagion; it’s not empathy in any way. So, we don’t know exactly why they yawn, but we’ve basically shown that yawning is not an empathetic response which a lot of psychologists had assumed that it was. And with crows, it’s mostly technical intelligence. They have, so far, outsmarted me [inaudible] yet. We’re interested in how they use tools and whether they can plan for the future and that sort of general kind of thing.
Campbell: I have a question about a term that I came across that I’d like you to talk a little bit more about, domain-general thought. What is that?
Bastos: So, basically what we think is that birds have very specific brains evolved for very specific purposes and they could do one very fast, specific thing at a time. But if intelligence has evolved in a more general way, so that’s what we call domain-general, it means the different units of the mind can work together and integrate to solve a problem at the same time. So, what we see is that the kea they can use different types of information to solve one problem, and they use those at the same time. You know, like they do in taking physical constraints into account in their probability or taking social biases into account as well. So that’s the first evidence we have of birds definitely using two types of intelligence, two types of knowledge at the same time, which suggests they have domain-general intelligence.
Bailer: So how old were the birds when you started training them?
Bastos: So, when I started training them for this experiment or when I started working with them?
Bailer: Yes.
Bastos: So, I think they would have been- when they started doing token exchange some of the birds were really young, so they would have been three or so years old and that’s when they’re effectively moving from being juveniles to becoming adults. And the fun thing with kea that you can always tell a juvenile because they have sort of yellow, fleshy bits around their eyes and their beaks, and so when that yellow starts to disappear they are going through their sort of teenage years and they are terrible. We got bitten a lot. But they do learn not to bite you after a while. So yeah most of them are between six and nine now, but we do have one 26-year-old in the aviary. They live quite a long time.
Bailer: [inaudible] Did you see any differences? Oh I’m sorry Richard.
Campbell: So, is that an old age 26, or do they love even older than that?
Bastos: Yes, they live up to 40 so that’s just past middle-aged birds.
Bailer: Did you say some tend to learn faster than others?
Bastos: Oh yeah, so two of our birds are, what I consider, the genius birds so Loki and Taz. They tend to perform very well in every study, so it seems like it applies to different types of questions as well, and other birds are more there for the participation points. So, my favorite bird- he’s very sweet we called Blofeld like the James Bond villain, and he’s lovely but he takes a long time to get things. So, he comes down to the platform, he’s very willing to work but has no idea what’s going on for a long time.
Bailer: I love the names that you’ve given these birds too, that’s brilliant. I was just curious- I was asking I mean I know that you have a few birds and in some sense, you’ve mentioned you’re doing proof of concept if this is even possible within this species, and it really does make me wonder about whether there were certain features that might be predictive of certain birds being more likely to learn. Whether in- sort of in the future if you were to place a bet as to- could you guess if this bird was going to be a – certain characteristics that might predict a certain bird is likely to learn quite well. Do you have any insights on that or any speculation?
Bastos: I mean that’s a good question; they all have very different personalities between the birds. So, what I’ve learned from who does best is just from experience. So, I’ve seen them perform well in other experiments, so I know that in the future they will do something similar. But yeah I think the two birds that are the best are also some of the boldest birds. So, Loki who’s one of the best birds is quite up there in the hierarchy and he’s quite energetic, so he’s always displacing other birds and such. Maybe there’s some suggestion that birds that are higher up in hierarchy might be- although Johnny, who is the top dog is- so yeah, I don’t know.
[laughter]
Bastos: Exactly it’s hard to say.
Bailer: Says the former department chair.
[laughter]
Bailer: Yeah so you have a former department chair and a current department chair, so the image of the top dog being useless, I don’t know.
Bastos: Yeah no he’s retired for sure he doesn’t work with us.
Campbell: So, what I want to know- how you got into this kind of work? And what was sort of- you’re very young still, but what was sort of the path to getting where you are today and then what do you want to do? What’s your next study? What’s the next thing you’re really excited about?
Bastos: Well, if we go back a long way, I grew up- I started off with a mini zoo because I’d adopt animals left, right and center, sometimes without my parents’ permission. So one day I just walked up with three massive tortoises to the house, who were delivered in a van just so you have an idea of how big these things were, and because my mom had agreed to some tortoises but she didn’t know the size she just had to accept it. So, I’ve always liked animals. I bred cockatiels, which are a little parrot, when I was growing up. So, I had quite close interaction with birds especially and I always knew they were quite smart. So, from an early age I wanted to study animal intelligence, so I sort of had that in mind then. So, I went to Oxford to study biology and there I was mentored by Alex Kacelnik, who is a big name in the field and worked with cockatoos at the Vienna lab. So yeah, I got a grant to go study in Austria for a couple of months and work with the birds so that’s sort of solidified the kind of thing I really enjoyed.
Campbell: So, what are you excited about next?
Bastos: Sorry yes, I forgot the second part of the question. So, we’re starting to integrate this with artificial intelligence next. We have some ideas about how we might collaborate with a couple of artificial intelligence scientists because so far the artificial general intelligence systems have been based on human neural mechanisms but because birds have simple brains this might also provide inspiration for systems that can do general intelligence. So that would be the next big step, but you know, that could be years and years and years. On a more short-term basis, we want to test if kea can integrate three types of information at the same time. So, if they can do physical, social and probability we’ll see how far
Campbell: There was another term prosocial behavior, how does an animal manifest that, or do they?
Bastos: So, prosocial behavior is like when an animal helps another one with no evidence of like self-interest in that sense. So, we have a paper on kea saying that they might be able to act prosocially, but our evidence isn’t quite there yet. There is a really good study that came out last year showing that African Gray parrots are prosocial, and what they do is they will sort of hand over a token to another bird because they can’t exchange it for food themselves. So, they will pass it through a plexiglass wall, like a hole in a wall to the other bird when they themselves can’t exchange.
Bailer: Didn’t you tell a story in part of one of the lower hierarchy birds kind of sneaking off to slip you a token for a reward? So, can you describe what happened there and how you interpret that?
Bastos: Yeah so what happened is we were teaching them in a group setting to exchange tokens and what that means is that we throw a bunch of black and orange tokens on the floor and we have the birds go get them and run back to us; it’s like playing fetch with a dog, really. But you know, some birds are more dominant than others so they will bully others out of their back tokens and exchange it with us. And one of the birds called Neo, he’s less than a dominant bird, he’s towards the end of the hierarchy, and he had a black token but instead of coming around to the front of me where I could see him and he could exchange tokens but other birds could see him as well, he perched on a rail behind me and poked me with the token on the back, and that was sort of the real first sign of intelligence that I saw in these birds because it was early on in my time working with them. And what it seems to me is that he was aware that I couldn’t see him and so not only was he avoiding the subordinate birds but indicating to me in another sort of way, like through touch rather than visual cues, that he was ready to exchange the tokens, so that was quite surprising.
Bailer: So, you’ve been involved in doing this really cool science and you’ve also been involved in communicating this really cool science in different venues. What has helped you to prepare to tell the story about the science that you do?
Bastos: Yeah I mean, to be honest, practice. The first interview I had was not that great, but as you- which is funny because that’s the one on the nature communications video, so that’s probably why they use two lines of it, but as you do it more and more and you understand the kinds of questions that people are asking you and the sorts of answers that they are looking for in the sense of how can you explain this in a way that’s understandable to a lay audience, it sort of builds that skill.
Bailer: Yeah you’ve done great work.
Bastos: Thank you.
Bailer: Well Amalia, I’m afraid that’s all the time we have today for Stats and Stories. Thank you so much joining us.
Bastos: Thank you.
Bailer: Stats and Stories is a production of Miami University’s departments of Statistics and Media, Journalism and Film, and the American Statistical Association. You can follow us on Twitter or Apple podcasts or other places you can find podcasts. To share your thoughts on the program send your email to statsandstories@miamioh.edu or check us out at statsandstories.net and be sure to listen for future editions of Stats and Stories, where we discuss the statistics behind the stories and the stories behind the statistics.