Stepping on snakes for science, and crows that count out loud

This episode explores unique animal behaviors and unusual scientific studies, focusing on how researchers step on snakes to understand their biting responses and teaching crows to count.

1. Venomous snakes are more likely to bite under certain conditions, such as higher temperatures and younger age.
2. Female snakes are generally more aggressive and prone to biting.
3. Crows can learn to associate a number of vocalizations with visual and auditory stimuli, displaying a form of counting.
4. Intensive hunting practices by early humans may have contributed to the extinction of cave bears.
5. Research on crows and snakes can provide insights into animal behavior and improve human safety and knowledge.

1: Stepping on Snakes

Exploring a daring experiment where a researcher stepped on venomous snakes to study their biting behavior. Factors like snake gender and temperature were found to influence biting.

• João Miguel Alves Nunez: "Not only did I step on snakes, but I also discovered how their behavior changes under different conditions."
• David Grimm: "The protective boots used in the study prevented most snake bites, except one instance with a rattlesnake."

2: Crows Counting

Discussing how crows can be taught to associate vocalizations with numbers, suggesting a basic form of counting similar to human children's early counting skills.

• Diana Liao: "Crows demonstrate the ability to learn and vocalize counts based on visual and auditory cues, akin to how children use verbal tallies."
• Kevin McLean: "This shows not only the intelligence of crows but also their potential to inform cognitive science."

3: Extinct Cave Bears

Analyzing the potential human role in the extinction of cave bears through intensive hunting practices during colder periods when bears were hibernating.

• David Grimm: "Our ancestors may have driven cave bears to extinction by both hunting them and displacing them from their hibernation caves."

• Be mindful of wildlife and their habitats to prevent harmful interactions.
• Support and participate in wildlife conservation efforts to protect endangered species.
• Educate yourself and others about the behavior of local wildlife to reduce negative encounters.
• Promote and engage in ethical research practices that respect animal welfare.
• Advocate for the use of scientific findings in policy making to enhance wildlife conservation.

A roundup of online news stories featuring animals, and researchers get crows to “count” to four

This week’s show is all animals all the time. First, Online News Editor Dave Grimm joins host Sarah Crespi to discuss stepping on venomous snakes for science, hunting ice age cave bears, and demolishing lizardlike buildings.

Next, producer Kevin McLean talks with Diana Liao, a postdoctoral researcher at the University of Tübingen, about teaching crows to count out loud. They discuss the complexity of this behavior and how, like the famous band, these counting corvids have all the right vocal skills to do it.

People

João Miguel Alves Nunez, Diana Liao, Kevin McLean, Sarah Crespi, David Grimm

Companies

None

Books

None

Guest Name(s):

None

Content Warnings:

None

Sarah Crespi
This podcast is supported by the Icon School of Medicine at Mount Sinai, one of America's leading research medical schools. Icon Mount Sinai is the academic arm of the eight hospital Mount Sinai Health System in New York City. It's consistently among the top recipients of NIH funding. Researchers at Icon Mount Sinai have made breakthrough discoveries in many fields vital to advancing the health of patients, including cancer, Covid and long Covid, cardiology, neuroscience, and artificial intelligence. The Icahn School of Medicine at Mount Sinai, we find a way.

This is the science podcast for May 24, 2024. I'm Sarah Crespi. This week's show is all animals all the time. First, of course, online news editor David Grimm is here with stories about stepping on snakes for science, hunting ice age cave bears and demolishing lizard like buildings. Next, producer Kevin McLean talks with researcher Diana Liao about teaching crows to count, or at least a form of counting. They discuss the complexity of this behavior and how, like the famous band, these counting corvids have all the right vocal skills to do it.

Sarah Crespi
Now we have online news editor David Grimm. He's here with all the animal stories we could possibly want.

Hi, Dave. So glad to have you back on the show.

David Grimm
Hey, Sarah. Good to be back.

Sarah Crespi
Yeah, I'm really glad. We're just going to indulge ourselves and only talk about animals this week. That's what we really want.

Sarah Crespi
Right?

David Grimm
Well, those are the best stories.

Sarah Crespi
Exactly. So let's start with something close to my heart, which is a story on snakes.

David Grimm
I hope there's not a snake close to your heart, Sarah.

Sarah Crespi
No. Snake hearted. Yeah, no, but this person is stepping on snakes, which I would never do.

Why is this researcher stepping on snakes?

Joe Miguel Avez Nunez
And this is a researcher in Brazil named Jo Miguel Avez Nunez. And not only did he step on.

David Grimm
Snakes, he stepped on snakes more than 40,000 times, Sarah. And this is not just your garden variety snake. He stepped on a snake called a jawaraka, which is one of the most dangerous and venomous vipers in South America.

Joe Miguel Avez Nunez
So this is a researcher in Brazil named Joe Miguel Alves Nunez. And not only did he step on snakes.

David Grimm
Yeah.

Joe Miguel Avez Nunez
So this is a researcher in Brazil named Jao Miguel Alvis Nunez. And not only did he step on snakes there.

David Grimm
One more time.

Joe Miguel Avez Nunez
And this is a researcher in Brazil named Joe Miguel Avez Nunez. And not only did he step on.

Sarah Crespi
Snakes, he's getting his steps in. He's got poison.

David Grimm
He's getting his snake steps in. I don't know if there's an app that tracks that, but yes.

Sarah Crespi
Okay. But he's not hurting them.

David Grimm
No, he's not hurting them. He's actually stepping fairly lightly on various parts of their body.

Sarah Crespi
You step on snakes and they bite you, right?

David Grimm
Well, right, exactly. We know that snakes bite 5 million people a year. We know that thousands of people die and are permanently disabled. But we really don't have a handle on why some snakes bite and why some don't. Even within the same species, they're probably.

Sarah Crespi
Not trying to eat a person. But if you're just walking by and you're already past them, why are they biting you in the back of the leg?

David Grimm
Right, exactly. Exactly.

Sarah Crespi
That's why he decided to step on snakes, to see if something about where you step on them might have something to do with if they're going to.

David Grimm
Bite you, where you step on them, whether the sex of the snake matters, whether other factors like temperature make a difference, all these things, surprisingly, are relatively unstudied, and that may be because nobody's willing to get in the ring. Literally. They put a ring in their lab, and he stepped into this ring with a snake.

Sarah Crespi
Besides this researcher, did he get bitten in this process?

David Grimm
He wore these super thick, special protective boots. They were very heavily padded. He never got bitten by the jawa raqqas. But in a control study where he was also looking at rattlesnakes, actually a rattlesnake tooth got through the boot. He did get bit, and he actually had to get emergency care. Now, luckily, he was at a university that specializes in snakes and snake venom, so he was able to get antivenom pretty quickly. But he also discovered not only is he allergic to snake bites, but he's allergic to antivenom. Not a great situation to be in if you're going to be the guy that says stepping 40,000 times on venomous snakes.

Sarah Crespi
Oh, my goodness. Okay, so we got one result. He is allergic to venom.

Sarah Crespi
Okay.

David Grimm
Right.

Sarah Crespi
The other result, what did he learn about snake habits, and how does it relate to some of these environmental factors or other snake demographics?

David Grimm
Yeah, he learned a bunch of things. First of all, that females are more aggressive, more prone to bites, that the younger they are, they're also more prone to biting. Also that these snakes tend to be more aggressive in hotter temperatures.

Sarah Crespi
They have the energy to be feisty if it's hot.

What can you do with this information? You know, you can't avoid stepping on female snakes.

David Grimm
Well, yeah, antivenom is very expensive, very difficult to distribute. And so the idea is we should have more antivenom in these regions. These regions where there are younger female steaks, where the temperatures are hotter. And epidemiological data actually supports what this researcher found, which is that in these regions that are hotter, that have more of these female steaks, people are more likely to be bitten. So the idea is, if you've got more antivenom in these places, you might be able to save lives.

Sarah Crespi
I mean, snake researchers are kind of doing dangerous things when they work with the venomous animal. But is stepping on them or encouraging them to bite part of the normal laboratory practice for working with these animals?

David Grimm
No, I think this is pretty unusual. There are other examples with other species, like people poking their fingers at spiders and things like that to try to get a sense of their defensive capabilities. I don't think a lot of people have done this with snakes. He actually said some people thought he was a bit reckless for doing it, but he also feels like he couldn't have gotten the answers he got without doing the research in this way.

Sarah Crespi
All right, now, moving on from snakes, we're going to bears, actually, it's a story about cave bears, which are extinct, unfortunately, yes. I mean, they are really big and really tall.

David Grimm
They're big. 750 kilos. I mean, they weigh ten times more, or they weigh ten times more than a. Than a person did, more than 3 meters tall. So these are not cute, cuddly little cubs we're talking about?

Kevin McLean
Yeah. No, they're way bigger than grizzlies. And there's a possibility that we are actually to blame for their disappearance, that they were hunted to extinction by humans.

Sarah Crespi
We actually asked this about a lot.

Kevin McLean
Of animals, think mastodons or saber toothed tigers.

Sarah Crespi
They're gone.

Kevin McLean
Was it climate? Was it people? Was it a combination? And so we're now asking this question about cave bears, and the evidence is really interesting here. It comes from their bones, and it spans a lot of time. How long were we coexisting with cave bears?

David Grimm
We're talking about Europe now. Modern humans have been tens of thousands of years overlapping with cave bears. But what's really fascinating about the study is it actually doesn't just look at modern humans. It looks at neanderthals that were around in Europe longer than we were, but also species like Homo Heidelberg and cysts, which were the immediate predecessor of neanderthals in Europe. And so now we can actually go back hundreds of thousands of years.

Sarah Crespi
Wow. And so here we are where our close relatives or close ancestors, and we have these cave bears, which are, what, like, how many meters tall?

David Grimm
3 meters tall.

Sarah Crespi
All right, let's go way back, hundreds.

Sarah Crespi
Of thousands of years ago.

Kevin McLean
So now we're talking about our ancient close relatives.

What was their interaction like with cave bears?

David Grimm
When we go back 300,000 years, what the archaeological evidence suggests is that we probably weren't hunting these animals regularly. There's caves where we find these bones of hominins and we find cave bear bones. And it doesn't seem like there's much evidence of activity on these bones. We're not seeing, like, cut marks, scrape marks, things like that. So it's possible that we were around with cave bears, but we weren't doing a whole lot with them. Maybe occasionally hunting them or scavenging them. But things change. About 40,000 years ago, we start to see slice marks on the paw bones of these bears, where maybe their hides would have been cut free. Bones that are cracked open to extract the yummy marrow inside. Sarah. And scrapes on long bones, showing that they were maybe stripped of every last scrap of meat. So now we're seeing these signs of this intensive butchery of cave bears.

Sarah Crespi
And this is 40,000 years ago. What else was going on?

David Grimm
This is around a time, and especially when we're talking a little bit more recently, maybe around 27,000 years ago, what's called the last glacial maximum. This was a cold spell. Cave bears would have been especially tempting targets during this time because they were found hibernating in caves. And so humans, they're cold, they're hungry, they find these sleeping bears, which seems kind of like cheating to me a little bit, and then they're killing them. And we're seeing this happen. At least what the archaeological evidence suggests is starting to happen pretty regularly. Maybe cave bears were kind of seasonally on the menu before, you know, hundreds of thousands of years ago. Now this is becoming a much more common practice.

Sarah Crespi
Right? It's, like, strategic. It's like, oh, we're starving. We better just go see if we can find a bear in a cave.

David Grimm
Exactly.

Sarah Crespi
We know that that's where they are. We can just scrape together some dinner.

David Grimm
And we know they're sleeping, so maybe they're not going to fight back.

Sarah Crespi
Do we know if this was humans or neanderthals? Do we know who was doing this during the glacial maximum?

David Grimm
It's probably modern humans, especially when we're talking about this time. But modern humans are really starting to multiply, starting to dominate the continent in a way that they weren't before.

Sarah Crespi
Does any of this archaeological evidence point to whether or not humans might have been responsible for the extinction of the k bears.

David Grimm
Well, right, so, probably. And for two reasons. One, because we have this intensive hunting going on, but also because these caves were now very sought after places to be, especially as it's getting colder and you can't cohabitate with a bear in a cave, necessarily. And so we probably have humans not only killing these bears, but driving them out of the caves. It's kind of these one two punch where humans are killing the bears. Bears don't have their refuges anymore. And by 10,000 years ago, the bear seemed to have disappeared.

Sarah Crespi
I don't have a stinger.

Sarah Crespi
That's sad.

David Grimm
Yeah. What is sad?

Sarah Crespi
All right, let's go onto the last story that we're going to cover, and this. It's kind of an animal story. I'm really stretching it here. This is basically the question, when is a building like a lizard?

David Grimm
When is a building like a lizard? A building is like a lizard.

When.

Let me ask you, Sarah, what's one of the cool things lizards can do?

Sarah Crespi
Get things with their tongues.

David Grimm
If we say that it's a cool thing that they can do, what's a cool defense mechanism lizards have?

Sarah Crespi
Okay, I know this one. Okay, so a lizard will shed its tail when a predator basically grabs onto it. So this has happened to me. I, like, picked up a five line skink and, you know, its tail detached and kept wiggling in the grass. And I was like, oh, maybe I should go get the tail instead of keeping this lizard.

David Grimm
Yes.

Sarah Crespi
Which is not what I did, but that's what they wanted me to do.

David Grimm
Well, right. And that's. I mean, not only is it. I mean, who knows if it's very. The lizard thinks it's very cool. I'm sure it's a very sort of stressful, traumatic thing. But the lizards survive, right? You know, this is their crazy defense tactic, that they actually shed a part of their body, they get away, and then they can regrow their tail later. And so a group of building engineers said, well, what if we could do that for buildings? You know, we're all aware of these catastrophic building collapses where, you know, part of the building collapses, and then the whole thing collapses.

Sarah Crespi
Right? Like, if you ever watch demolition, you can kind of see where one part collapses and it just seems to take the rest of the building with it.

David Grimm
Takes the rest of the building with it.

Sarah Crespi
And that happens with earthquakes, flooding, all kinds of disasters. Once certain pieces of the structure are taken away, it's just.

Sarah Crespi
It's over.

David Grimm
And so the idea is, could we develop these lizard inspired buildings where if part of the building collapses, only that part collapses. It becomes isolated with the researchers in this state called a hierarchy base collapse isolation, which is a fancy way of saying lizard inspired building. But basically, if just part of the building collapses, the rest of the building can be spared. The building is not so interconnected that one part of the building collapsing takes the rest of the building with it.

Sarah Crespi
Right. So that's kind of a principle that we use now, which is, oh, there's some damage to one part of the building. It's designed to redistribute that weight and that stress to the rest of the building so that it doesn't collapse immediately. But that can end up inducing enough problems across the building that the whole thing collapses 100%. So that's the theory. But what's really, besides this being a lizard inspired building? They had a really interesting approach to testing this. You want to talk about that?

David Grimm
Yeah. So what the researchers did, you could see a video of this on the site, is they built this large two story building. It's roughly the size of a single family house, and they built it at a concrete. Then this must have been the fun part. They used a forklift to topple a critical load bearing column. But the way that they designed the building, and you can see this collapse in slow motion, is that only part of the building collapses and then the rest stays completely structurally sound, just standing there almost as if nothing had happened to it.

Sarah Crespi
Yeah. So I watched this video, too, and it looks like a parking garage. It does just take one of those pillars out. But what's cool about it, too, is that they said, okay, well, if we remove these two pillars on purpose, they're supposed to be removable. Like, so there is kind of an architectural standard, like, okay, well, engineering standard, where if you remove two pillars, it should be okay, that worked. But when they swept the leg, basically, and took out an extra one instead of the whole building collapsing, what they ended up with was just this section collapsing. So this could be something that could be expanded. Now, obviously, this is very controlled circumstances, and a lot of things would have to change for this to become building code. But it is definitely something we should be paying attention to.

David Grimm
Well, right, obviously you could be saving lives here. But the other thing I thought was interesting was these buildings often have to be, there's anything left standing completely demolished, or the whole building's already demolished. And we always hear about this is billions of dollars, millions of dollars in damage. As the idea here is, you're also potentially saving money, too, because if you only have to rebuild part of the building. The costs are not going to be.

Sarah Crespi
As great modular destruction.

David Grimm
There you go.

Sarah Crespi
What else should we look for on the site?

Sarah Crespi
Dave?

David Grimm
Well, Sarah, got a couple other animal stories for you. We got sacrificing medieval horses, why people did that, and when and how and where. A story about how cockroaches took over the world. And finally, a story about the potential dangers of the keto diet.

Sarah Crespi
Oh, interesting.

David Grimm
Be sure to check out all that on the site.

Sarah Crespi
Thanks, Dave.

David Grimm
Thanks, Sarah.

Sarah Crespi
David Grimm is the online news editor for Science. You can find a link to the story we discussed@science.org. podcast. You can also follow our news team on many social platforms, Facebook, Twitter, that was, and Instagram. Look for Science magazine or news from Science.

Sarah Crespi
Stay tuned for a chat with producer Kevin McLean and researcher Diana Liao about Crow's ability to count to four in their own special way.

You listen to us to hear about new discoveries in science. But did you know we're a part of the American association for the Advancement of Science? AAA's is a nonprofit publisher and a science society. When you join AAA's, you help support our mission to advance science for the benefit of all.

Become a AAA's member at the silver level or above. To receive a year's subscription to science and an exclusive gift, join today by visiting AAA's dot join. That's aaas.org join. This week's episode is brought to you in part by the Eppendorf and Science Prize for Neurobiology.

Are you or one of your colleagues doing great neuroscience?

If so, then we encourage you to apply for the prestigious Eppendorf and Science Prize for Neurobiology, an international prize which honors young scientists for outstanding neurobiological research based on methods of molecular cellular systems or organismic biology.

Submissions are due June 15. Visit science.org eppendorf to apply today as.

Kevin McLean
Humans, counting is built into our lives from a very early age. One of the first things young toddlers might learn when they start talking is how to count out loud, maybe count up toys in their room or anything else around them.

Learning this early on is amazing because counting is surprisingly complex.

You have to both understand numbers and have a lot of vocal control.

The combination of skills needed is so sophisticated, in fact, that it seems pretty uniquely human. But this week in science, Diana Liao et al described their work testing whether crows, chatty birds known for their numerical abilities, can learn to count, too.

Hi Diana. Welcome to the science Podcast.

Diana Liao
Hi. It's great to be here, and I'm always happy to chat about crows?

Kevin McLean
Absolutely. You know, I feel like the more time we spend learning about the inner lives of animals, the more that we find that things we thought were unique to us kind of aren't. But how human is counting so full.

Diana Liao
Symbolic counting, I think, is uniquely human. But, for example, processing numerosities with the approximate number system, for example, can be found relatively widely across different animal species.

These animal species are able to distinguish between, for example, one and two, or maybe one and eight. So large ratios and small numbers. But as these ratios get smaller, like, for example, six and eight, it becomes much harder for them. We humans also have this approximate number system when we're very young, and then we develop symbolic counting on top of that with mathematical schooling.

Kevin McLean
Yeah, you mentioned that some of what we know about how we learn to count when we're young influence some of this work. Can you talk about that a little bit?

Diana Liao
Yeah. So we were inspired by these studies of toddlers learning to count. So when they have not yet mastered the cardinality of number words or the number of elements in a set, they instead use number words as kind of a verbal tally. For example, when shown, like, five apples and asked, like, how many apples are there? The child might not answer five immediately, but produce a sequence like 12345, or even something like 1111, where they're producing one counting word for each object as a sort of verbal tally.

Kevin McLean
Okay, so, yeah, they're, like, tallying it up, but saying each tally mark, basically.

Diana Liao
Exactly. By using the number of their vocalizations.

Kevin McLean
I see. And what about animals? Is there anything out there about different species vocalizing numbers or quantities?

Diana Liao
One of my favorite papers involves chickadee alarm calls, where they found that the number of d notes in their alarm calls kind of scales with the wingspan of the predator. And they find this negative relationship between the number of notes and the size of the predator, which makes sense, because smaller, more agile predators are actually more dangerous to these tiny chickadees.

Kevin McLean
Okay, so there's sort of, like, a reason to be able to give a relationship between, like, a number and their survival? Basically, yeah.

Diana Liao
So they could be conveying the magnitude of the threat. And in these studies, there are two hypotheses. One is that the number of causes produced to convey information about the identity of the predator. And the other is that this behavior is kind of driven by involuntary mechanisms such as arousal. So, like, the more dangerous a predator is, the higher the arousal level of the chickadee, and thus, kind of the more calls the chickadee makes.

Kevin McLean
Great.

Kevin McLean
Then why were crows a good species for answering the kinds of questions that you wanted to get at.

Diana Liao
Crows are super smart. They can make these rule guided decisions. They have complex social structures, and some species even create and use tools. And theyre also quite good at numbers. And in addition, there are also songbirds. So we thought this combination of cognition and vocal abilities makes crows an excellent model system to study this in.

Kevin McLean
I guess I never really think of crows as songbirds, but, yeah, obviously they just are.

Diana Liao
No, I was also surprised by that, actually, before I started working in this lab, because they don't have very beautiful songs.

Kevin McLean
How did you go about designing a task to test whether crows can do this counting?

Diana Liao
We wanted to bring this behavior into the lab where we can control the stimuli. And so what we've done is we've presented the crows with visual stimuli, so these colored arabic numbers and also auditory stimuli, these 500 millisecond sounds. And then they're trained to associate, producing a number of vocalizations to each of these different stimuli.

Kevin McLean
Okay, so you're showing a stimulus and also they're hearing a stimulus, and they are trained to make a number of vocalizations associated with both of those.

Diana Liao
Yeah, exactly. So there's a cue period where they see the stimulus, there's a response period where they have to make the number of vocalizations. And then to report that they're done, they have to peck the screen, and then that ends the trial.

Kevin McLean
What was the process like to actually do this training for them? How long did it take and how many animals were you working with?

Diana Liao
Yeah, so I worked with three different crows, and this took quite a while, actually, because it's a complex task, almost a year of training. So we train them to produce one to four vocalizations to these different visual and auditory cues.

Kevin McLean
I see.

Kevin McLean
And then they get a reward or something for doing it correctly.

Diana Liao
Yeah. So they peck the screen, and then if the number of vocalizations matches the cued number, then they get a reward. So a little bird pellet or a meal worm. And if it doesn't match, then the screen kind of flashes and they don't get a reward.

Kevin McLean
So I guess let's get into the results a little bit. I noticed you were looking at both the results of the animals that did this correctly, but also you then dove into when animals got it wrong as well. But let's start with when they did it right. What did you find in terms of how these animals were learning this?

Diana Liao
We found that crows could do it, which was really cool. And they did it with kind of these signatures of the approximate number system. So one signature is the numerical distance effect where when an error is made, it tends to be around the target number. It's easier to confuse three and four as compared to one and four. And there's also, for example, the numerical size effect where discrimination becomes systematically less precise in proportion to increasing numbers. So one is easier than four to them.

Kevin McLean
I see. Was it just the number of vocalizations that they were giving? Was there also something about how they vocalized them that you were looking at?

Diana Liao
Yeah. So in the training, it was just a number that we were training on. And then after we found out that crows could do this, like, how were they doing it? We were interested in whether they were, like, maybe doing some sort of planning process or where they kind of prepare the sequence as a whole before they started vocalizing. Or were they kind of making these vocalizations one at a time on the fly, one might say.

And so when we first looked at the temporal information contained in the responses we found that this reaction time. So the time from the onset of the vocalization queue to the onset of the very first vocalization is much longer than any of the subsequent inter call intervals.

Kevin McLean
Okay, so they're waiting longer to start that call versus, like, the time in between each of the. The vocalizations.

Diana Liao
Yeah, it takes them longer to start the sequence. I should mention here. It was also really interesting that this reaction time scales with the number of vocalizations the crow makes. So the more vocalizations the crow has to make, the longer it takes before he starts vocalizing.

Kevin McLean
They're thinking about the number of vocalizations or something, or we don't quite know.

Diana Liao
But maybe this is the planning process there. They're thinking about the number of vocalizations or maybe preparing themselves to kind of make this number of vocalizations great.

Kevin McLean
And you also looked at how the vocalizations actually sounded to see if there was anything specific to the number and the way they called, I think, in the first call. So what did you learn from analyzing those calls?

Diana Liao
We looked at the acoustic features of the very first vocalization in the sequence to see if that contained information about the number of subsequent vocalizations in the sequence. So the idea is that if it's prepared, then there should be some signature of the number already in the first element. And so we took these acoustic features to see if the acoustic features of the first vocalizations could be used to predict the total number of subsequent vocalizations. And we found that, yes, it could. And we also found that data classifiers trained on one modality. So, for example, the visual cues could generalize to data in the other modalities. So the auditory cues as well.

Kevin McLean
Okay, so the signature of that first vocalization kind of indicated the number of vocalizations that they were going to make as well.

Diana Liao
Yeah.

Kevin McLean
Okay. So the signature of that first vocalization kind of indicated the number of vocalizations that they were going to make as well. And then for these classifiers, you're training a computer to recognize the different types of calls of each bird.

Diana Liao
Exactly.

Kevin McLean
Interesting.

Kevin McLean
And you did this with multiple individuals. Did they all do it the same way or how did it look across the different individuals you tested?

Diana Liao
That's a great question. What we did to address this question was we took the classifiers trained on each individual and then tested them on data from other individuals and found that basically these classifiers did terribly. Each of these crows, even though they could do the task, did it using their own vocalizations in their own way.

Kevin McLean
Okay. So they're not, like, teaching each other their own number systems or anything like that?

Diana Liao
No. That would be cool.

Kevin McLean
I thought it was so interesting that you looked into the times that they didn't get it right.

Kevin McLean
So if you play the queue for three and they don't make the right.

Kevin McLean
Number of vocalizations too many or not enough, there was something to learn from how they made those mistakes.

Diana Liao
Yeah. So the previous analyses I described were all on the acoustic features of the first vocalization. But what about the rest of the vocalizations in the sequence? So we also calculated the acoustic features of each of those vocalizations and fed those into a classifier to see if, in addition to the number, can we predict the position of each vocalization in the sequence? Can we tell in a sequence of four vocalizations? This vocalization is in the third position out of four.

Kevin McLean
I see. Okay.

Diana Liao
And the classifiers could do so. So these vocalizations are following some sort of trajectory in acoustic space.

And then from this analysis, we can use it to get a readout of where the crow thinks it is in the sequence and what types of errors it makes. So we have some things like skip errors where the crow skips one particular position. So instead of 1234, it might be going one, two, four.

Or a stutter error, for example, where they repeat a certain position. So they go like, maybe 123345.

Kevin McLean
Yeah, the stutter. Or maybe instances where they lost count or lost track. It's very relatable, actually.

Diana Liao
The crows aren't specifically making one vocalization for each number. It's just that they have to make a certain number of vocalizations. And the acoustic features of these vocalizations vary as they're making these sequences. So I'm not saying the curls are saying 12345.

Kevin McLean
So it's more that they're saying, like, 1111, but there are slight variations in those ones, I think.

Diana Liao
So, yeah.

Kevin McLean
When there's a difference in that first vocalization you've listened to a lot of these crows do this task. Can you hear it, or is it not until you do the computational analysis that you can tell that there's a difference?

Diana Liao
So, I've heard, like, tens of thousands of crow vocalizations at this point.

David Grimm
Yeah.

Diana Liao
So I can hear it, but I've shown them to other people, and they're like, what are you talking about?

Kevin McLean
Do you want to play some of the audio you have from the tests?

Diana Liao
Sure. So, here are trials that one of my crows did.

Kevin McLean
I'm hearing there's, like, the audio cue that it's given. Right. And then I hear, like, in the first one, it was, like, four different vocalizations. And then I'm assuming that's it, like, tapping the board at the end of it. Is that what's going on?

Diana Liao
Exactly. You hear the audio cue, other 500 millisecond sounds, and then the crow vocalizes, and then that thunk at the end, is it pecking at the touchscreen to report that it's done?

Kevin McLean
Okay.

Diana Liao
And the reason that's so loud is because the microphone is right in front of the touchscreen and the crowd. And then because these were all correct trials, there's a little bit of a ring at the end of the pec as well. And that indicates that the trial was correct. And then the crow gets a reward.

Kevin McLean
Ah, got it. Okay.

Diana Liao
All right, let me play it again.

Kevin McLean
Okay, so that's a cue.

Diana Liao
Four vocalizations.

Kevin McLean
There's four. And then a tap. Yeah. Okay.

One vocalization.

All right. And then two. And then the tap. So they were trained on the four was like that sort of like.

I don't know. I don't know how to describe it.

Diana Liao
Oh, it's a frequency sweep.

Kevin McLean
A frequency sweep. Got it.

Diana Liao
Yeah.

Kevin McLean
Okay, so these are all correct trials.

Diana Liao
Yeah.

Kevin McLean
Okay. And are you. Tell me, are you hearing the difference in that first vocalization?

Diana Liao
To me, they sound slightly different. But you might have also noticed it took longer for the crow to start vocalizing for the four than it did, for example, the one or the two.

Kevin McLean
Yeah. Yeah. Okay. What's next with this kind of work? What are you interested in finding out beyond what you've done so far?

Diana Liao
This capacity to control the number of vocalizations could be an evolutionary precursor of true counting, where numbers are part of a combinatorial symbolic system. And it's interesting to me that this precursor exists in crows, which have diverged from the human lineage over, like, 300 million years ago, and they have evolved a dramatically different brain architecture. So I'd be curious to see kind of what are the brain networks that would support such an ability.

Kevin McLean
Great.

Kevin McLean
Well, thank you so much, Dan. It was great to talk with you.

Diana Liao
Yeah, this is cool.

Kevin McLean
Diana Liao is a postdoctoral researcher at the University of Tubingen. You can read the paper we discussed@science.org.

Sarah Crespi
Podcasts and that concludes this edition of the Science podcast. If you have any comments or suggestions, write to us@sciencepodcast.org dot to find us on podcasting apps like Overcast, Apple Podcasts, or Spotify Search for Science magazine. Or you can listen on our website, science.org podcast. This show was edited by me, Sarah Crespi, and Kevin McLean. We also had production help from Megan Tuck at Prodigy. Jeffrey Cook composed the music on behalf of science and its publisher, AaAAS. Thanks for joining us.