Search the Site

Episode Transcript

What breed is she?” Pause on a dog walk to let your dog sniff another dog, and it’s likely that this is the question you’ll get from the person at the end of the other leash. I always have lived with mixed-breed dogs, and just as long have played the guessing game of their breed ancestry.

When we’re asking about a dog’s breed, we’re asking who they are. What’s the genetic history of this pup? But how much do a dog’s genes really tell us about who they are?

Welcome to Off Leash. I’m Alexandra Horowitz. This is the podcast where we explore the relationship between humans and dogs. I’ll invite an interesting person — and their dog — to join me for a walk, and then we’ll see where the walk takes us. Today on the show, it takes us to the question of what our dog’s D.N.A. tells us about how they’ll look and act – what kind of dog they are. We’ll talk to an expert on the canine genome:

Elinor KARLSSON: We went ahead and sequenced their Border Collie for them and then had a long conversation because it turned out that we couldn’t find any Border Collie ancestry in this Border Collie.

And we’ll play a little breed-ancestry guessing game.

Steve LEVITT: I don’t see anything Beagle-y or Cocker Spaniel-y about him.

Morgan LEVEY: Beagle-y.

But first, we’re taking a walk with a mixed breed dog named Coco and her person, Soledad O’Brien — who has also looked into her own ancestry.

 Soledad O’BRIEN: I definitely thought, like, “Oh, I have a relative, who’s an entrepreneur, so maybe this is in me?”

Join us.

O’BRIEN: Come on, big girl. You ready to go?

*      *      *

O’BRIEN: This is Soledad O’Brien and I’m with my dog Coco today. I have two dogs, but Coco’s the better-behaved dog, so we’re going to stick with Coco for the purposes of this interview.

Soledad is an award-winning documentarian, broadcast journalist, and author. She and her family are based in New York City most of the year, but they’d migrated south to escape the winter when I walked with her.

O’BRIEN: I’m currently walking through my neighborhood, which is West Palm Beach, Florida. And I hate to say this for all my friends in the Northeast, but it is a perfect 70 degrees this late afternoon.

HOROWITZ: Wow. That hurts.

O’BRIEN: I know. I’m sorry.  

HOROWITZ: Well, I am in New York City in Central Park, which is, in its own way, beautiful in the winter. Let me ask you a little bit about Coco. What is she like? How old is she?

O’BRIEN: Gosh, she’s just about 2 years old. You know, during the pandemic, I think the kids were losing their minds, as we all were. And it just seemed like a really good idea. Like if you’re ever going to get a dog, when you’re stuck at home, that’s the time to get a dog. So we saw Coco at a rescue and we made the decision to bring her home. And it has been an absolutely brilliant decision. She’s yummy and snuggly and just very sweet. Everybody who sees her, loves her.

Soledad appeared on the P.B.S. show “Finding Your Roots with Henry Louis Gates Jr.”, which does an extraordinary deep dive into guests’ genealogies. In the case of Soledad, she already knew her ancestry was multicultural.

HENRY LOUIS GATES JR: Soledad credits her parents with giving her the strength to overcome discrimination. But her ambiguous physical features, and her light brown color routinely force her to explain her ethnic identity.

When people say “what are you?” in one word, what do you say?

O’ BRIEN: If I’m feeling snitty I say “I’m an American, what are you?”

*      *      *

HOROWITZ: I’d love it if you could tell us a little bit about your parents, what were they like? Where were they born?

O’BRIEN: My mom was born in 1930 in Havana, Cuba. She was Afro-Cuban. But they were interested in my dad’s side of the family. So my dad’s relatives were from Scotland and Ireland then through Australia. And my dad was white and he was born in Toowoomba, Australia in 1933.

HOROWITZ: I noticed in just looking over your career, you have worn a lot of hats, and you’ve also been recognized for a lot of awards and some of them had an ethnic component. So Irish American Magazine called you one of the top 100 Irish Americans and you were on Black Enterprise‘s “hot list.” And you were a groundbreaking Latina of the year in Catalina Magazine.

O’BRIEN: I love it. That’s the upside of being multicultural.

HOROWITZ: It’s sort of quintessentially American. And I, I wondered if you felt connected with each part of your heritage?

O’BRIEN: I do. And I love it. I think it’s great. I am Irish American. I am Afro-Latina. I am Cuban American. I am African-American. I — you know, that is, to me, the beauty and the interesting thing about being in America. So I absolutely love it. And I like being able to bring a perspective that’s different from everybody else’s perspective. You know, there’s tons of Afro-Latinas, and we all have different stories and all are from different parts of the country and all have different ways that our family members got here. And so I like being part of a big fabric where you can contribute pieces to make the whole thing make a lot of sense. And I think for people who don’t know anything about their past or don’t know anything about their relatives or their history, it’s very hard because you don’t know where you come from. So it is — it’s always very comforting, maybe is the right word, to know like, “oh, this is the stock I come from.” I love that.

For Soledad, the most interesting part of her genealogical findings was learning about the lives of individual ancestors — like a great-grandmother in Australia who was widowed with 10 children.

O’BRIEN: She was a business woman who had to figure out how to make her way when her husband died. They ran a flour mill, and they worked with all the farmers. Well, when her husband died, the banks wouldn’t loan her money for the business because she was a woman. And so she went around and asked all the farmers they were working with to help pitch in to get her business off the ground. And they did. And she would go on to become a very well-off woman. She ran a very successful business. And, you know, what you really realize through the kind of work that Dr. Gates does is how much you learn about yourself when you learn about your relatives. At the time when we did that show, I was just starting as an entrepreneur. And I definitely thought, “Oh, I have a relative, who’s an entrepreneur, so maybe this is in me?” Come on, Coco. It makes you feel connected to: “I come from these people.”

HOROWITZ: Absolutely.

O’BRIEN: Come on, my love. Do you need some water? Here you go. There you go. Oh, you’re thirsty. Good girl. So we get this dog. Coco looks exactly like a Rhodesian Ridgeback, but smaller, but exactly like one. Actually there was one in the park today and they look exactly alike, but Coco is about probably 40 pounds lighter than a Rhodesian Ridgeback. And so we had her D.N.A. done and there’s something nice about, like, knowing where somebody is from, even if that somebody is your rescue dog. Now, Coco’s was all over the map and I’m not sure it’s accurate, cause it’s literally everything. It’s crazy.

HOROWITZ: So including what? What does it include?

O’BRIEN: Oh, gosh. Well, Pit Bull, she’s like 30 percent Pit Bull. She has some Lab. She has some Akita. She has some Shih Tzu. She has some Beagle. She has some — oh gosh, there’s a whole — I mean, basically you name it, Coco is it. Come on, Coco. Come on. Come on, my love.

HOROWITZ: Except Rhodesian Ridgeback.

O’BRIEN: Except zero Rhodesian Ridgeback. I think people don’t let the Rhodesian Ridgebacks out to get knocked up by neighborhood Pit Bull, would be my guess, because those are fancy dogs. You know, but part of it is you want to understand what you think you know about a Lab or what you think you know about the temperament or the behavior of a Pit Bull or a Shih Tzu or a Boxer. She’s part Boxer, too. You know, I think that’s what you think about. Like, oh, she’s this percent Boxer. Where do I see that play out in her personality?

HOROWITZ: And have you seen any Box-ing behavior? Or Pit Bull or Lab-like behavior?

O’BRIEN: No, none. I mean — oh, Coco! Oh, my goodness. You are so strong. Uh-uh. She loves — one of the challenges we have in Florida — come on! Stop eating those little —.

HOROWITZ: What did she get?

O’BRIEN: What are those? Salamanders? Little lizards?

HOROWITZ: Oh, wow.

O’BRIEN: She loves the lizards. Coco!           

HOROWITZ: Are they the ones that lose their tail when you grab them?

O’BRIEN: I don’t think — well, we’re never that good at grabbing them. So I’m going to say, who knows, because she’s slow and I don’t think ever she’s going to get her own little lizard. Come on! But that doesn’t keep us from trying extremely hard to grab a lizard.

HOROWITZ: That’s probably a Pit Bull characteristic right there.

O’BRIEN: She’s a little bit of a hunter, for sure. And, she is so strong. She’s the most mellow sweet dog, but you know what? If she sees something she likes, that’s it. She’s gone. But I don’t — this is my first dog, which is why she’s a little bit spoiled and she’s dragging me down the sidewalk past all these beautiful palm trees. And so I don’t even really know — like, I couldn’t tell you what a Pit Bull is like. Getting the DNA test was fun, but it didn’t give me a ton of insight.

HOROWITZ: Then there’s still the question, you know, what’s the meaning of it in the end versus just being the parlor game of knowing my dog comes from a Shih Tzu.

O’BRIEN: Right? Like, so many things. What does it all mean at the end of the day?

What does it all mean? If anyone can tell us, it might be one of the pioneers in the field of dog D.N.A.:

KARLSSON: My name is Elinor Karlsson. I’m a professor at the UMass Chan Medical School and at the Broad Institute. And I study what we call genomics, which is basically trying to figure out how changes in D.N.A. end up changing the way an individual of any species looks or behaves or responds to something in their environment.

HOROWITZ: Is what you do different than what a geneticist does?

KARLSSON: You know, that’s a really difficult question. Rather than looking at things one gene at a time, you’re really looking at all of the D.N.A. that an individual carries and trying to figure out how it works as kind of a whole thing. And then also how it interacts with the environment. And so, it’s kind of like we do a lot of things with very large data sets in order to try and understand this very complicated system.

Elinor’s career began with a bang, as she was part of a history-making effort to sequence the entire genome of a dog for the first time.

KARLSSON: Yeah, she was just a very Boxer, Boxer.

HOROWITZ: A Boxer named Tasha, I believe.

KARLSSON: Yes, and I was working with somebody called Kerstin Lindblad-Toh. And so, she kind of came to me and she said, “Well, what would you think about working on a project where we figured out how to find genes for traits like coat color in dogs?” And, you know, I’d been a genetics fan ever since I first learned about Mendel and his peapods back in ninth grade. I was just totally fascinated by the fact that you could kind of figure out what an organism might look like based on gene frequencies and things. And so, that was the perfect project that she could have pitched to me. And so, I had an opportunity as a PhD student to start working on the dog genome because we first needed to kind of have the basic map of what a dog genome looked like.

HOROWITZ: Was there a specific reason that a Boxer was chosen? I mean, is there something extra dog-like about this kind of dog or about this dog in particular?

KARLSSON: So Tasha being a pure-bred dog, and it looked like she was kind of exceptionally, what we would call in genetics “inbred,” and what they would call in dog breeding language, “purebred.” And a genome is basically about 3 billion letters of D.N.A. And so, in order to sequence a genome, we have to sequence all 3 billion of those letters. And we do this in pieces of — at the time, about 600 letters each. And so, what you do is you get a blood sample from Tasha, you extract her D.N.A., you break it up into pieces, and you put it on these machines and it gives you little separate 600-letter chunks of her genome. And you have to put it back together like it’s a gigantic puzzle. And because her mother and her father looked very like each other in terms of their genetics, that made putting the jigsaw puzzle back together much easier than it would have if her mother and father had looked very different from one another.

HOROWITZ: Why are dogs a good model with which to study genetics?

KARLSSON: The biggest reason that we were really excited about using dogs to do things like find genes for traits, was because in the breeds there’s less genetic diversity, which, given the tools that we had at the time made it much easier to map the genes, which means find the gene that’s responsible for a trait. The other amazing thing about dogs probably where their biggest advantage lies, is that every single pet dog lives with a human that spends a huge amount of time watching them. And that is fantastically powerful for doing genetics, because it turns out the really hard part isn’t the part that we do — the genetics part. The really hard part is what we call phenotyping. It’s knowing about the individual. It’s knowing what their behavior is like, knowing what they look like, knowing how much they like to chase things. And that’s something that, if we had to go out and test every single dog, would take forever, and we still wouldn’t have that many dogs. But if we can just get the owners to tell us — because it turns out they kind of all know this stuff — then we can get lots and lots of dogs signed up pretty quickly.

HOROWITZ: So this leads me to your next project, which you called Darwin’s Ark. Can you tell me a little bit about what it is and how it began?

KARLSSON: Yes. So Darwin’s Ark is a community science project and basically you can go to our website at and answer a whole lot of questions about your dog. You can also choose to get your dog D.N.A. sequenced. And if people do that, then that genetic data kind of goes into our Darwin’s Ark data repository, and any researcher can access that.

HOROWITZ: Are people getting information about the breeds of their dogs?

KARLSSON: That is definitely one of the things we do. When we sequence your dog, we analyze it using something we call breed calling, which is basically we compare your dog to all the breeds that we have data from out there and say, “This is what percentage of your dog comes from this breed. This is what percentage of your dog’s D.N.A. comes from this other breed.” I was always interested by the fact that whenever anybody has a dog and they don’t know what breed it is, people will ask. And everybody tends to give one or two breeds. And what we found when we were looking at mutts — or dogs, basically, of complicated ancestry that nobody understands — is that more than 60 percent of them actually have substantial ancestry from at least four breeds and often more.

HOROWITZ: I wonder if people ever respond to you about the results by saying, “Gee, I’m sure you must be wrong, because my dog doesn’t look anything like the breeds that you suggested their genome is closest to.”

KARLSSON: I think there was somebody that we worked with who had a Border Collie that they had adopted from the Border Collie Rescue that they volunteered for. And we went ahead and sequenced their Border Collie for them and then had a long conversation because it turned out that we couldn’t find any Border Collie ancestry in this Border Collie, which was just — from a genetic standpoint, I just found fascinating because you looked at pictures of this dog and it did look like a Border Collie. — black and white and it had floppy ears but those kinds of traits can come in from other breeds as well. And so, you get the right combination of traits coming into one individual and they can definitely look like a breed that they might have no ancestry from.

HOROWITZ: I know you’ve also done some research where you’re asking people to take looks at photos of mutts, of mixed-breed dogs, and guess what dog breeds are responsible for that mix. Can you tell us about that and what you found?

KARLSSON: Yes. So, this was a project that we called Mutt Mix, and I think you can actually still check it out at And we wanted to know when people guess what breeds are in a mutt, how right are they? So we signed up 30 dogs. Basically got their owners to take videos of their dogs and photos of their dogs and tell us how big they were and things like that. And we made a little page for each dog and asked people to guess what they thought the top three breeds were that were contributing ancestry to this dog. You know, is it a mix of Labrador Retriever, Golden Retriever, and Dachshund, for example? And I think we got over 30,000 people to participate in the two months that the project was collecting data for research purposes. Everybody had a really hard time with guessing a third breed. It was just quite counter-intuitive to a lot of people. They were so used to thinking about things as like a mix of one breed and another breed —.

HOROWITZ: A mother and a father.

KARLSSON: Yeah. But we also figured out that people could only on average, I think they guessed about 20 percent of a dog’s ancestry correctly. So, we could actually see to a certain extent, which traits guided people’s answers.

HOROWITZ: Do you mean like, what part of the dog, physically? It’s their head, or people were looking at ear shape or snout size or something?

KARLSSON: Exactly. So, if a dog had longer fur, people were more likely to guess that it had Golden Retriever ancestry, and if it had really upright ears, they were more likely to guess Chihuahua and things like that, which I just thought was really cool.

After the break, we’ll do a test of our own, with some Freakonomics Radio Network dogs — and persons — Steve Levitt, host of People I (Mostly) Admire and his producer, Morgan Levey.

LEVEY: I clearly am working on the wrong Freakonomics Radio Network show.

*      *      *

Welcome back to Off Leash. I’m Alexandra Horowitz. Inspired by Muttmix, we cooked up a dog-breed identification game of our own. I invited Steve Levitt, the co-author of the Freakonomics books, and Morgan Levey, who produces Steve’s podcast People I (Mostly) Admire to play along. I showed Morgan and Steve photos, and gave them a little description of several mixed-breed dogs who’ve been D.N.A.-tested; and you’ll hear them try to guess the top (three) breeds identified in their ancestry. And then Morgan finds out the D.N.A.-test results for her dog, Diosa. The first dog up is one of the dogs I’ve lived with for 14 years, the lovely Finnegan.

HOROWITZ: So Finnegan is kind of ink black and has an almost smiling face and a wagging tail and friendly eyes. He’s about 65 pounds in this photo. So he’s a medium, medium-large dog. He loved running through puddles. He was very friendly with everybody who we’d meet on the sidewalk. And he would follow us from room to room and even sort of follow us with his eyes within the room. So, Steve, what do you think?

LEVITT: All right. Well, he’s very black, and he kind of looks like a Lab. So I might say that he would have some black Lab in him.

LEVEY: He basically looks like every Lab I’ve ever known. They have those floppy ears, the long tail, the long legs, and his big smiling face.

LEVITT: He has something very Doberman about him, too.

LEVEY: I would say he would have maybe some brown on him if he was part Doberman. But I see that he has a little bit of white on his chest. Maybe part Dalmatian? That I’m going out on a limb.

LEVITT: And the fact that you said he liked to follow around and — this is probably completely stupid, but maybe he has some Border Collie? That’s the best I can do.

LEVEY: I feel like English Pointers can be hunting dogs, so maybe — yeah, I’m going to say part English Pointer, part Dalmatian, and definitely Labrador Retriever.

HOROWITZ: So Finnegan is: Labrador Retriever, Beagle, and Cocker Spaniel.

LEVEY: Whoa! Beagle? Cocker Spaniel?

LEVITT: I don’t see anything Beagle-y or Cocker Spaniel-y about him.

LEVEY: Beagle-y.

HOROWITZ: You know, if you get — maybe if you go way, way back. I mean, what it partly indicates is that with the D.N.A. testing, what we’re getting is genetic relatedness. And it doesn’t necessarily mean that Finn had a Beagle as a great, great, great, great grandfather, but that something about his particular genome more resembles Beagles’ genomes than others. You know, we got the idea for this game from Elinor Karlsson, and what she said was that there was Pit Bull/Staffordshire Terrier, in about 25 percent of the dogs. There’s Staffordshire/Pit Bull-y type dogs in so many mixed breeds now, right. They’re a very common dog. But the guesses of whether there was Pit Bull-Terrier in there really varied based completely on whether the dog was a stocky looking dog with short hair, so we do definitely identify those dogs just by their appearance.

LEVEY: I’d like it just for the record to show that we are tied at this moment. Steve and I are tied.

HOROWITZ: Both have identified one breed of three.

LEVITT: I had no idea that you were going to be so competitive, Morgan. I thought that we were just here to have fun.

HOROWITZ: Okay. Shall we go on to Bruce?

LEVEY: Oh, Bruce.

HOROWITZ: Bruce is our producer Molly’s dog. And Bruce is about 14 pounds. Has this lovely tan golden fur, pretty short hair. Pointy erect ears and a curled tail that basically reaches his back. He’s a big shedder and very furry around his neck. He can sleep all day, but also likes chasing little creatures in the backyard. And he will also make prolonged eye contact with you and then apparently will crawl on people on the couch and stand on their chest.

LEVEY: I mean, pretty much all my instincts are off in this game, but I really think this dog is a Chihuahua.

LEVITT: He’s very little. And he has that kind of Chihuahua face to him?

LEVEY: I don’t mean to be dismissive but those, like, buggy kind of eyes that Chihuahuas have.

LEVITT: Okay, so I’m going to say Chihuahua — though given how things are going, I should do the opposite of what I think.

LEVEY: He’s just a little furrier than a regular Chihuahua. I think he’s got more scruff on him, which makes me think that there’s some long-haired dog in there. This might be a little crazy, but I think he could be part, maybe American Eskimo Dog? Oh, that’s a bold choice though.

LEVITT: It also — it doesn’t have the distorted body of a Corgi. But the tail looks kind of a little bit Corgi-like because it’s flipped over and touching him on the back.

LEVEY: So I don’t think a Corgi or a Dachshund because they have such short, stumpy legs and Bruce’s legs look like, in proportion to the rest of his body. I’m also going to say part Jack Russell Terrier, I think, cause of the tail?

LEVITT: How about just for fun, we say Pit Bull.

HOROWITZ: Excellent. Bruce is Pomeranian, Chihuahua, and American Eskimo Dog.

LEVEY: Oh, wow! I am so proud of myself.

HOROWITZ: That’s exciting.

LEVITT: Maybe Morgan should have a podcast on dogs

HOROWITZ: Morgan is definitely a dog identifier—.

LEVEY: Look at me go.

HOROWITZ: We’re getting at something here. Okay. Here we have— Morgan, how do you say your dog’s name?

LEVEY: Her name is Diosa. And if you speak Spanish, that means goddess. And yes, we are those people who have named their dog goddess.

HOROWITZ: Diosa is about 40 pounds. Again, a tri-color dog with this very white mane of a face. “Sheds like a waterfall,” I can imagine that. She has long hair. She’s smart. She makes a lot of direct eye contact. And she — her eye contact is with these brown eyes that are also distinctive for being rimmed with pink. She prefers to hang out with people over dogs although she’s not a cuddler. She likes swimming. And she plays fetch. And doesn’t bark that much, but does whine when she’s excited.

LEVITT: So I’ve never seen a weirder-looking dog than this one, Morgan.

LEVEY: I think you mean a more beautiful-looking dog, Steve.

LEVITT: Yeah, exactly. So I am going to follow my usual faulty path of looking at colors and say the Border Collie because of the white and the black. I’m going to also say a retriever, because of the swimming and the furriness and the shedding. So like a Golden Retriever? And then just for old time’s sake, how about I go with the Pit Bull as the third part?

LEVEY: So Diosa is from an accidental litter. And so when my husband got her, he was told that her father was a pure breed Australian Shepherd, and that her mom is an Australian Shepherd and Border Collie mix. So you might be right, Levitt. I think she’s a mix of the two.

HOROWITZ: She doesn’t seem to have a tail, or she has a short tail. Is that right?

LEVEY: Yeah, they clipped it.

HOROWITZ: Oh, they clipped it. Oh, okay. Interesting. Because sometimes Australian Shepherds do have a very short tail, naturally as well. So, I wondered.

LEVITT: Why would you clip a dog’s tail other than to subscribe to—?

LEVEY: They do it a lot.

LEVITT: Some image of what a dog should look like?

HOROWITZ: It is very much about what the dog should look like. Ostensibly, tail docking was for the purpose of them not getting their tails caught in things or tangled in brush if they were chasing after prey or game, for instance. And then it just becomes something that’s — people are accustomed to seeing a certain way a breed looks. And it’s completely unnecessary and probably what we would describe as highly barbaric and is outlawed in a number of countries actually. But it’s still widely practiced in the States. For a short time in the 19th century, when pure breeding was really getting going, there was a tax on tails. In fact, dogs who had tails were taxed. And so people would crop the tails to avoid the tax.

LEVEY: Why on Earth would you tax the tail?

HOROWITZ: You know, the whole Victorian pure breeding beginning, it’s hard to answer “why” questions about what they were doing. But there were ideas about separating, you know, the elite dogs, the best examples of dogs, from the kind of street dogs, you know, the dog that anybody could have. And so some features were picked on as being indicative of one or the other.

LEVEY: That’s crazy.

HOROWITZ: So Morgan, what are her results?

LEVEY: Okay. I have the email, so I am clicking on “click to reveal.” What? Diosa is 100 percent Australian Shepherd. She’s a pure breed!? Oh my God. She’s so fancy. We’ve been calling her a mutt her whole life. This is like Cinderella Story.

HOROWITZ: It turns out she is the princess.

LEVEY: Wow! That’s wild. We always thought that Border Collie was in there because she’s so black and white that, you know, like Steve said, the coloring — we assumed that was totally true. That’s so wild.

HOROWITZ: And her coloring is very unusual for a shepherd. You know, I haven’t ever seen this pure white, almost like albino-y face, especially with the pink rims.

LEVITT: Do you love her more now?

LEVEY: I feel like we have not given her the due that she deserves no, I mean, I don’t know if I love her more. I think she just is much fancier than we could have imagined, but that’s not a surprise. She knows she’s beautiful.

LEVITT: Let me ask you one question. So if you trace back a particular breed, how many ancestors are there? How many —

HOROWITZ: Like founder dogs?

LEVITT: Adam and Eve. Like founder dogs, yeah. Uh, how many are there for a given species? A very small number? Amazingly inbred?

HOROWITZ: Yes. You can date the moment, say, the German Shepherd started being bred. You know, there was a dog who looked something like what you think a German Shepherd would look like today. His name was Horand. And a gentleman decided he was like, the perfect looking dog. And he mated Horand with other dogs that look sort of like that. And then immediately their puppies would be mated with one of them so that all dogs in this newly formed breed are genetically related to each other. This is called a closed breeding line. And so, yeah, there are basically founder dogs. And if you had a German Shepherd, and your dog is registered and pedigreed, they are connected to Horand.

LEVEY: Wow. That’s amazing. I clearly am working on the wrong Freakonomics Radio Network show.

HOROWITZ: Well, you can always guest-visit us any time, Morgan.

LEVEY: Great, well thanks for having me. It’s been wonderful.

HOROWITZ: Thank you both. Thanks for playing!

One of the clues to identifying a dog’s ancestry is their behavior — something Elinor Karlsson is especially keen on:

KARLSSON: We were really interested in studying behavior in dogs because that’s one of the things that seems to differ a lot between different dogs and maybe between different breeds. And it’s one of those traits that’s really hard to understand the genetics for. What’s going on with behavior is that we have changes in probably thousands of different genes that contribute to a dog’s personality and behavior. And then on top of that, you have a huge environmental influence on behavior. So, genes are only telling part of the story. And because of that, it’s much more challenging to find changes in genes that change the behavior of a dog as compared to a change in a gene that changes the color or the size or the shape of a dog.

HOROWITZ: You know, if somebody is interested in having their dog’s D.N.A. tested, what they want to find out is something about who that dog is. So that might be partly through their breed ancestry, but maybe it’s about explaining behavior that they see or predicting behavior that you could expect the dog to show in certain circumstances?

KARLSSON: We can start now to tell you some connections between breeds and behaviors. So, if you have a dog that has ancestry from Beagles, your dog is much more likely to howl than if your dog has most of their ancestry from Labrador Retrievers. One day we might get closer to being able to predict what a dog’s behavior is like just from their D.N.A. sequence alone. Although the fact that environment has such a big influence on behavior means it’s always going to be a probability thing. You know, “your dog is more likely to be like this.” It’s never going to be, “Your dog’s definitely going to be like this.” And so, one of our challenges has been figuring out how to distinguish between things that we find are interesting about breeds as a population, but making sure that people understand that if they’re trying to understand their dog, that that’s still an individual dog. And just because the dog has ancestry from a particular breed, there’s never any certainties about what that dog’s behavior is going to be like.

HOROWITZ: You told me you got some really exciting information from looking at surveys. These are surveys that the owners filled out about their dogs, right? So what was the purpose of having them complete these surveys?

KARLSSON: So, one of the questions that I really wanted to address with the surveys, and it took us a little while to figure out how to do this, was the question of “does behavior differ between dog breeds?” And so, the first thing that we kind of did with the surveys is basically design a way to take all those, you know, tens of thousands of owners that had answered the surveys for us and ask the question of, if we look at, you know, 50 Golden Retrievers, does their behavior tend to be different from 50 dogs just randomly sampled out of our Darwin’s Ark population of dogs. We took all of the 110 questions that we asked about their dogs and kind of simplified them into what are called behavioral factors.


KARLSSON: And so, we had one factor that we called biddability, which is sort of like trainability. It’s kind of like, does the dog understand what it is you want them to do? And do they really care about actually doing it? And that we do see showing up more in breeds that you would expect to see it in, like Border Collies tend to be exceptionally biddable. And we had another one that we called toy-directed motor patterns, which is like how much your dog likes to play with toys. And so, toy-directed motor patterns tends to have German Shepherds and Border Collies scoring quite high on that. And so, we were finding those actually did tend to be different in breeds than in dogs overall. But there are also factors that you might’ve expected to see differences between the breeds where we — we didn’t see any differences. And probably one of the more intriguing ones in that line was one that we called agonistic threshold, which is basically how strongly a dog reacts to something surprising or scary in their world.

HOROWITZ: We sometimes call that reactivity.

KARLSSON: Yeah. And when we actually looked at that factor, we saw essentially no difference between the breeds. And we also couldn’t really see genetics having any influence on that trait really in the whole dog population. And so, it seems to be a very environmentally driven trait as opposed to one that’s controlled by genetics.

HOROWITZ: That’s such an interesting one, because of course, if there’s one thing everybody wants to have an opinion about, it’s about which breed is — and I guess another word for agonistic threshold would be, “is most likely to become aggressive.” So, you’re saying that in your database, looking at survey and the survey results and the genomes, it really comes out in the wash between breeds.


HOROWITZ: And how many dogs are in your dataset?

KARLSSON: We had about 20,000 dogs that had answered the surveys. A little annoying for —

HOROWITZ: Well, not the dogs.

KARLSSON: Yeah. Sorry.

HOROWITZ: I mean, I love the notion. Our dogs do a lot of amazing things, but —.

KARLSSON: Sorry. I really have to get better at who’s doing what here. Yeah, so, we have about 20,000 dogs that people have answered surveys for.

HOROWITZ: And how many dogs’ genomes have you sequenced?

KARLSSON: At this point in time, we have about 3,100 dogs that we have genome sequence data for, which is pretty fantastic. It’s not as many as we would like. In a perfect world, I’d love to have genetic data from tens of thousands of dogs. We can find things with 3,000 dogs, but we’re going to be able to find even more things when we get up to 30,000 dogs. Everybody always asks me whether the project has ended, and I was like, “This project is never going to end because we will always want more data.” And the more that we have, the more that we can learn.

HOROWITZ: One question I had is, you know, by domesticating dogs, and then breeding dogs, we’ve essentially changed their genome over time. So, does domestication work on identifiable specific genes in particular? And what are those genes?

KARLSSON: Really interesting question. So, dog breeds are a very recent phenomenon, as we think about dog breeds today. They’re only about less than 200 years old. So, that’s super modern. Dog domestication probably happened — oh, the estimates are all over the place, but it’s probably somewhere between 12,000 and 20,000 years ago. So, that is way older than anything to do with dog breeds. And we don’t know how that happened yet.

So when someone asks you what breed your dog is, maybe you can give a pithy answer. But every dog I’ve met, just like every person, defies such easy categorization. Deep down, we know, “what breed they are” is only the beginning of finding out who your dog really is. Thank you to Elinor Karlsson, Soledad O’Brien, her dog Coco, Steve Levitt and Morgan Levey — and to you for joining us on this walk today.

*      *      *

Off Leash is produced by Stitcher and Renbud Radio, and is part of the Freakonomics Radio Network — which also includes Freakonomics Radio, No Stupid Questions, People I (Mostly) Admire, and Freakonomics, M.D. This episode was produced by Molly Getman and Lyric Bowditch, with help from Jared Hohlt, Alison Craiglow, Gabriel Roth, and Stephen Dubner. It was mixed by Greg Rippin and Jeremy Johnston.

Our staff also includes Neal Carruth, Zack Lapinski, Ryan Kelley, Rebecca Lee Douglas, Morgan Levey, Julie Kanfer, Emma Tyrrell, Jasmin Klinger, Eleanor Osborne, Jacob Clemente, and Alina Kulman. Our music was composed by Luis Guerra.

To see behind-the-mic photos of all the dogs you met in this episode — or to share your thoughts on the show — follow Off Leash on Instagram and Twitter at @OffLeashShow. You can also shoot us an email at To find a transcript, links to research, and a newsletter sign-up, go to As always, thanks for listening.

*      *      *

Hi! It’s Alexandra again. This is the last episode in our first season of Off Leash. We loved making these episodes for you and would love to make more. The best way you can help us do that? Tell your friends, family, or even strangers you meet at the dog park about Off Leash. Thanks so much.

O’BRIEN: You’re such a good girl. Come on. Come on, we’re walking. Come on. You’re too heavy for me to carry you Coco. I can’t do it. I cannot do it.

Read full Transcript


  • Soledad O’Brien, documentarian, broadcast journalist, and author. 
  • Elinor Karlsson, director of the Vertebrate Genomics Group at the Broad Institute and professor in Bioinformatics and Integrative Biology at the University of Massachusetts Medical School.
  • Steve Levitt, host of People I (Mostly) Admire and economist at the University of Chicago. 
  • Morgan Levey, producer of People I (Mostly) Admire.



Episode Video