Does a dog's breed tell us anything about it's behavior, particularly aggressive behavior?
The largest and most recent study, by Morrill et al sequenced DNA of >2000 dogs and looked at genetic correlates of various behaviors. Some behaviors were weakly inherited, but aggressive behaviors were not predicted by breed. (Kathleen Morrill et a…
Does a dog's breed tell us anything about it's behavior, particularly aggressive behavior?
The largest and most recent study, by Morrill et al sequenced DNA of >2000 dogs and looked at genetic correlates of various behaviors. Some behaviors were weakly inherited, but aggressive behaviors were not predicted by breed. (Kathleen Morrill et al. "Ancestry-inclusive dog genomics challenges popular breed stereotypes". Science376,eabk0639(2022).DOI:10.1126/science.abk0639)
The authors concluded: " For less heritable, less breed-differentiated traits, like agonistic threshold (factor 5), which measures how easily a dog is provoked by frightening, uncomfortable, or annoying stimuli, breed is almost uninformative.
Attempting to eliminate certain behaviors by dog breed bans may be based on pseudo-science, or popular prejudice.
It was a survey that asked owners of purebred and mixed breed dogs to plot their pups on a likert scale for statements like "DOG seeks companionship from people," "DOG behaves aggressively in response to perceived threats from people" and "DOG is a people person." Only 10 of the items were about the dog's disposition toward people, and none asked specific questions about aggressive behavior, like whether the dog had ever bitten a person.
The paper was an attempt to investigate genomic correlates of a range of canine behaviors. Owners answered, on average, >100 questions per dog. It's true that aggressive behavior was only one of eight behaviors investigated (with 10 questions for each of eight behaviors). The questions asked on behavior were sourced from "previously published and validated surveys" such as the Dog personality Questionnaire (45 questions) and the Dog Impulsivity Assessment Scale (18 questions), so that the results of this survey could be meaningfully compared with other published research-a standard methodologic study design.
This study was not specifically a study focused solely on aggressive behavior in dogs. Nevertheless, it provides the best look, so far, at the relationship between dog "breed" and behavior, and suggests that the main inherited differences between breeds are aesthetic, not behavioral.
I do think it's an interesting paper. The key finding (that individual behavior within breeds varies much more than aggregated behavior between breeds) is much stronger than most people would intuit. I just don't think it's suited for this question, for a few reasons:
1. Serious dog attacks on people are VERY rare, and multiple attacks are often attributed to the same dog. If you look at the stats, you should already assume a lot of variation with aggression-prone breeds — most of them never bite a person, or there would be a lot more ER visits!
2. The questions asked would not distinguish a dog that growled or snarled from one thatb bit, and it left it to owners to determine what behavior counts as "aggressive." It's very possible that a person who chooses a miniature poodle as a pet has a very different idea about what counts as aggression than a person who chooses a pitbull.
3. The cluster of questions labeled "socialization with humans" — which contains items about aggression —also asks about the dog's interest in people, enjoyment of human companionship, willingness to be hugged, etc. By all accounts, most pitbulls are very human-oriented and affectionate! Unlike dogs bred for solitary tasks like livestock guarding, they tend to like being close to people. Even pits that eventually attack a person are often described as great family pets prior to the attack. I suspect that the model of aggression inferred here is not a very good one.
That actual dog-on-human attacks are rare events, and therefore unlikely to be explained by a study such as this is a fair point. Perhaps the general design of this study, relating genomics to observed behavior in a relatively representative population of dogs, could work with a more focused study looking at DNA analysis of dogs whose owners self-report an attack on a human. (?) The statistical power would not be as great, obviously, as the number of participants would likely be much smaller, given the rarer end-points.
I also agree that surveys of dog owners on their pet's behavior are unfortunately, but necessarily, subjective. I don't know how else we can get at individual dog behavior in a canine population though, except through a questionnaire of some sort.
Your final point falls back on the anecdotal, conventional wisdom that PBs are "human oriented", "affectionate", "like being close to people", yet somehow prone (?) to transgressive and surprising aggression. That all may be true, but it is interesting that the Morrill study, using a less fine grained lens on dog behavior overall, didn't find much support for any of the conventional wisdom on breed behavior.
That's a fair criticism. To clarify, my argument was that if you're using a small number of subjective questions to model each of these traits before analyzing their heritability, it matters A LOT whether you're asking the right questions. Any question that isn't actually relevant to the trait you're concerned about can be expected to add a lot of noise. I don't think the "sociability to humans" questionnaire does a great job of modeling a trait that would be a proxy for aggression (and it wasn't necessarily meant to).
That said, the authors published a MASSIVE set of infographics about their data (which are very fun to look at) and I'm seeing now that this wasn't a situation where pits did very well on questions about liking people and very poorly on questions about threat response or aggression. They actually were solidly middle-of-the-road on all human socialization questions: https://www.science.org/action/downloadSupplement?doi=10.1126%2Fscience.abk0639&file=science.abk0639_sm.pdf
Even allowing that Morrill et al did not completely nail down the issue of super-aggression (human biting) behavior, if they're correct that dog behavior is only very weakly associated with breed DNA, it begs the question of the effect of dog raising (environment) on aggressive behavior.
One of the concerns I have is that Pit Bulls may tell us much more about their owners, which is to say their upbringing and environment, than about their genes. It's entirely possible that certain owners desire certain characteristics in their dogs (e.g. aggression) and raise them to act in that way. Theoretically, any dog could be taught to be very aggressive toward unknown humans and other dogs. If that is true, knowing a dog's phenotype (breed)-or DNA for that matter- when it is obtained from a breeder or the SPCA would tell us next to nothing about its propensity for violence in the future. To know whether a dog is at risk for biting a human, you really would want to know how the human owners raised the dog.
That might contradict the conventional story of "he was such a sweet dog until he bit that baby!"- which has become such a trope that it's hard to dismiss it entirely.
While I find the results interesting, I don't actually give them a whole lot of weight. The same handful of breeds tend to dominate obedience, agility and herding trials, purpose-bred working dogs have been used for various jobs for centuries because they do tend to do those jobs better, and pits aren't used in dog fighting because they look like they are able to win, but because they actually do. A survey in which dog owners give a wide variety of responses to subjective questions about their dogs' personalities regardless of their breed doesn't provide a lot of evidence that extreme behavior isn't heritable.
There are many, many pit owners who claim that their dog was a beloved family pet who showed no human aggression until it suddenly "snapped." There is no test that will prove or disprove this. There is no way to know whether a pit in a shelter was always treated kindly. I'm not sure the question should concern policy makers at all.
"... doesn't provide a lot of evidence that extreme behavior isn't heritable."
It's very hard to prove a negative in science, particularly when it comes to the cause of rare events. The lack of positive evidence that extreme behavior is inherited is about the best we're going to be able to demonstrate.
I agree that, from the POV of policy makers, it may not matter whether extreme behavior is inherited or acquired (taught). If PBs and Bullies are doggy angels at birth but taught to be hypervigilant and violent by their owners, if you're the UK government you may still want to ban them.
This survey doesn't produce any direct evidence for or against the heritability of a propensity to attack humans because it didn't ask any questions about that.
You're using it to argue that any evidence for significant variability of personality traits within breeds (relative to between breeds) is tantamount to evidence that breed doesn't predict that kind of violence.
I'm pointing out that an identical assumption could be made about any "extreme" behavior — herding, guarding, escaping/wandering, retrieving, and so on. Do collies and heelers only herd better because they're raised to do that? Would a properly trained husky do just as well? Do golden retrievers absolutely dominate obedience trials because of bias? Could a lab make a great personal protection dog if he were trained for that?
People who spend a lot of time working with dogs will point out that none of that makes any sense; the volume of evidence that certain breeds do certain jobs better is enormous. Further, you would *expect* to find variability within breeds regarding extreme behavior, because everyone who breeds and trains working dogs knows that regression to the mean is a significant issue and most dogs within a working breed aren't actually suitable for work.
It would also be silly to argue that all pitbulls are killing machines, but that isn't the argument.
Does a dog's breed tell us anything about it's behavior, particularly aggressive behavior?
The largest and most recent study, by Morrill et al sequenced DNA of >2000 dogs and looked at genetic correlates of various behaviors. Some behaviors were weakly inherited, but aggressive behaviors were not predicted by breed. (Kathleen Morrill et al. "Ancestry-inclusive dog genomics challenges popular breed stereotypes". Science376,eabk0639(2022).DOI:10.1126/science.abk0639)
The authors concluded: " For less heritable, less breed-differentiated traits, like agonistic threshold (factor 5), which measures how easily a dog is provoked by frightening, uncomfortable, or annoying stimuli, breed is almost uninformative.
Attempting to eliminate certain behaviors by dog breed bans may be based on pseudo-science, or popular prejudice.
Here is the paper: https://www.science.org/doi/10.1126/science.abk0639
It was a survey that asked owners of purebred and mixed breed dogs to plot their pups on a likert scale for statements like "DOG seeks companionship from people," "DOG behaves aggressively in response to perceived threats from people" and "DOG is a people person." Only 10 of the items were about the dog's disposition toward people, and none asked specific questions about aggressive behavior, like whether the dog had ever bitten a person.
The paper was an attempt to investigate genomic correlates of a range of canine behaviors. Owners answered, on average, >100 questions per dog. It's true that aggressive behavior was only one of eight behaviors investigated (with 10 questions for each of eight behaviors). The questions asked on behavior were sourced from "previously published and validated surveys" such as the Dog personality Questionnaire (45 questions) and the Dog Impulsivity Assessment Scale (18 questions), so that the results of this survey could be meaningfully compared with other published research-a standard methodologic study design.
This study was not specifically a study focused solely on aggressive behavior in dogs. Nevertheless, it provides the best look, so far, at the relationship between dog "breed" and behavior, and suggests that the main inherited differences between breeds are aesthetic, not behavioral.
I do think it's an interesting paper. The key finding (that individual behavior within breeds varies much more than aggregated behavior between breeds) is much stronger than most people would intuit. I just don't think it's suited for this question, for a few reasons:
1. Serious dog attacks on people are VERY rare, and multiple attacks are often attributed to the same dog. If you look at the stats, you should already assume a lot of variation with aggression-prone breeds — most of them never bite a person, or there would be a lot more ER visits!
2. The questions asked would not distinguish a dog that growled or snarled from one thatb bit, and it left it to owners to determine what behavior counts as "aggressive." It's very possible that a person who chooses a miniature poodle as a pet has a very different idea about what counts as aggression than a person who chooses a pitbull.
3. The cluster of questions labeled "socialization with humans" — which contains items about aggression —also asks about the dog's interest in people, enjoyment of human companionship, willingness to be hugged, etc. By all accounts, most pitbulls are very human-oriented and affectionate! Unlike dogs bred for solitary tasks like livestock guarding, they tend to like being close to people. Even pits that eventually attack a person are often described as great family pets prior to the attack. I suspect that the model of aggression inferred here is not a very good one.
That actual dog-on-human attacks are rare events, and therefore unlikely to be explained by a study such as this is a fair point. Perhaps the general design of this study, relating genomics to observed behavior in a relatively representative population of dogs, could work with a more focused study looking at DNA analysis of dogs whose owners self-report an attack on a human. (?) The statistical power would not be as great, obviously, as the number of participants would likely be much smaller, given the rarer end-points.
I also agree that surveys of dog owners on their pet's behavior are unfortunately, but necessarily, subjective. I don't know how else we can get at individual dog behavior in a canine population though, except through a questionnaire of some sort.
Your final point falls back on the anecdotal, conventional wisdom that PBs are "human oriented", "affectionate", "like being close to people", yet somehow prone (?) to transgressive and surprising aggression. That all may be true, but it is interesting that the Morrill study, using a less fine grained lens on dog behavior overall, didn't find much support for any of the conventional wisdom on breed behavior.
That's a fair criticism. To clarify, my argument was that if you're using a small number of subjective questions to model each of these traits before analyzing their heritability, it matters A LOT whether you're asking the right questions. Any question that isn't actually relevant to the trait you're concerned about can be expected to add a lot of noise. I don't think the "sociability to humans" questionnaire does a great job of modeling a trait that would be a proxy for aggression (and it wasn't necessarily meant to).
That said, the authors published a MASSIVE set of infographics about their data (which are very fun to look at) and I'm seeing now that this wasn't a situation where pits did very well on questions about liking people and very poorly on questions about threat response or aggression. They actually were solidly middle-of-the-road on all human socialization questions: https://www.science.org/action/downloadSupplement?doi=10.1126%2Fscience.abk0639&file=science.abk0639_sm.pdf
(pg 39)
Even allowing that Morrill et al did not completely nail down the issue of super-aggression (human biting) behavior, if they're correct that dog behavior is only very weakly associated with breed DNA, it begs the question of the effect of dog raising (environment) on aggressive behavior.
One of the concerns I have is that Pit Bulls may tell us much more about their owners, which is to say their upbringing and environment, than about their genes. It's entirely possible that certain owners desire certain characteristics in their dogs (e.g. aggression) and raise them to act in that way. Theoretically, any dog could be taught to be very aggressive toward unknown humans and other dogs. If that is true, knowing a dog's phenotype (breed)-or DNA for that matter- when it is obtained from a breeder or the SPCA would tell us next to nothing about its propensity for violence in the future. To know whether a dog is at risk for biting a human, you really would want to know how the human owners raised the dog.
That might contradict the conventional story of "he was such a sweet dog until he bit that baby!"- which has become such a trope that it's hard to dismiss it entirely.
While I find the results interesting, I don't actually give them a whole lot of weight. The same handful of breeds tend to dominate obedience, agility and herding trials, purpose-bred working dogs have been used for various jobs for centuries because they do tend to do those jobs better, and pits aren't used in dog fighting because they look like they are able to win, but because they actually do. A survey in which dog owners give a wide variety of responses to subjective questions about their dogs' personalities regardless of their breed doesn't provide a lot of evidence that extreme behavior isn't heritable.
There are many, many pit owners who claim that their dog was a beloved family pet who showed no human aggression until it suddenly "snapped." There is no test that will prove or disprove this. There is no way to know whether a pit in a shelter was always treated kindly. I'm not sure the question should concern policy makers at all.
"... doesn't provide a lot of evidence that extreme behavior isn't heritable."
It's very hard to prove a negative in science, particularly when it comes to the cause of rare events. The lack of positive evidence that extreme behavior is inherited is about the best we're going to be able to demonstrate.
I agree that, from the POV of policy makers, it may not matter whether extreme behavior is inherited or acquired (taught). If PBs and Bullies are doggy angels at birth but taught to be hypervigilant and violent by their owners, if you're the UK government you may still want to ban them.
This survey doesn't produce any direct evidence for or against the heritability of a propensity to attack humans because it didn't ask any questions about that.
You're using it to argue that any evidence for significant variability of personality traits within breeds (relative to between breeds) is tantamount to evidence that breed doesn't predict that kind of violence.
I'm pointing out that an identical assumption could be made about any "extreme" behavior — herding, guarding, escaping/wandering, retrieving, and so on. Do collies and heelers only herd better because they're raised to do that? Would a properly trained husky do just as well? Do golden retrievers absolutely dominate obedience trials because of bias? Could a lab make a great personal protection dog if he were trained for that?
People who spend a lot of time working with dogs will point out that none of that makes any sense; the volume of evidence that certain breeds do certain jobs better is enormous. Further, you would *expect* to find variability within breeds regarding extreme behavior, because everyone who breeds and trains working dogs knows that regression to the mean is a significant issue and most dogs within a working breed aren't actually suitable for work.
It would also be silly to argue that all pitbulls are killing machines, but that isn't the argument.