Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Study of pet dogs shows breed does not predict behaviour

nature.com

Post reply on HN
Print view View original

111–120 of 191 posts

Re: Study of pet dogs shows breed does not predict behaviour

#111

Post by barry-cotter »

TameAntelope wrote: post #107
barry-cotter wrote: post #98

Earlier quoted context omitted.

I didn’t say anything about race. The word is useless, imprecise. There’s no canonical Swede or Arab? You don’t say. As I mentioned no one who looks like a Swede existed 5,000 years ago. That there is overlap between population characteristics does not mean they are identical. Some populations don’t overlap though. Healthy Pygmy and Maasai adult men do not overlap in height distribution. If you restrict your sample o…

You definitely keep on talking about race. "Swede" is a race. "Arab" is a race, as you use it. There absolutely were ancient people who shared genetic traits (and therefore some expressed characteristics) with modern Swedish people. There literally must have been, that's how genetics work, and 5,000 years is barely a blink, genetically. What you are conflating (among many, many things) are environmental impacts on hu…

> There absolutely were ancient people who shared genetic traits (and therefore some expressed characteristics) with modern Swedish people. There literally must have been, that's how genetics work, and 5,000 years is barely a blink, genetically.

That’s kind of my point. 5,000 years ago there were no people who looked like Swedes. Now there are because of selection pressures. A new genotype came into existence by working on standing variation and some de novo mutations. If we agree that selection pressures can radically change the appearance of a people then there’s not much to disagree on. Evolution can work on many things apart from physical appears.

Different places have different selection pressures, with the obvious consequences.

Re: Study of pet dogs shows breed does not predict behaviour

#112

Post by User23 »

TameAntelope wrote: post #94
barry-cotter wrote: post #90

Earlier quoted context omitted.

Who said anything about subspecies? All human groups can breed with no issue. Humans are clearly differentiated by geography. Swedes don’t look like Arabs. Arabs don’t look like Sub Saharan Africans. None of these groups look like Indians or East Asians. Quite visible geographic differentiation. If the specific claims I made about the length of reproductive isolation are false you will find it easy to provide sources…

All of those people you've named can and do look like one another, depending on circumstance. Every trait you can think of that you'd list as a "characteristic" of one of the groups you've named can be found in abundance in numerous other places as well, or accounted for due to environmental factors. There is no "canonical" "Swede" or "Arab". Your reliance on stereotype is not only racist but it's factually wrong. Th…

> Race does not exist, genetically. Period.

And yet ancestry.com can comfortably tell anyone which continental scale human population or populations they are descended from and that happens to map pretty neatly to the US Census Bureau's notion of race.

Being aware that major geographic features of the Earth such as the Atlantic Ocean, the Himalayas, and the Sahara Desert were more or less impermeable barriers to gene flow between continental human populations until about five centuries ago isn't racist and you should stop calling people names.

Re: Study of pet dogs shows breed does not predict behaviour

#113

Post by TameAntelope »

barry-cotter wrote: post #111
TameAntelope wrote: post #107

Earlier quoted context omitted.

You definitely keep on talking about race. "Swede" is a race. "Arab" is a race, as you use it. There absolutely were ancient people who shared genetic traits (and therefore some expressed characteristics) with modern Swedish people. There literally must have been, that's how genetics work, and 5,000 years is barely a blink, genetically. What you are conflating (among many, many things) are environmental impacts on hu…

> There absolutely were ancient people who shared genetic traits (and therefore some expressed characteristics) with modern Swedish people. There literally must have been, that's how genetics work, and 5,000 years is barely a blink, genetically. That’s kind of my point. 5,000 years ago there were no people who looked like Swedes. Now there are because of selection pressures. A new genotype came into existence by work…

> 5,000 years ago there were no people who looked like Swedes. Now there are because of selection pressures.

100% false. You need to realize that. 5,000 years is not enough time to do anything significant, genetically.

Re: Study of pet dogs shows breed does not predict behaviour

#114

Post by defen »

randomblast wrote: post #93
defen wrote: post #28

Earlier quoted context omitted.

No, the issue is that the reporting on this study represents a general problem in science reporting. This study measured behaviors that for the most part have zero relevance to public policy. Whether a dog likes to eat grass, crosses its front paws, turns in circles before pooping, or enjoys life (Those are all real questions from the survey) are not things that the government or the general public has any conceivabl…

Did you read the paper? They derived 8 behavioural factors from the survey questions, and there are a few which are very relevant to dog-human aggression. #1: Human sociability #2: Arousal level #4: Biddability #5: Agonistic threshold (agonistic is the behavioural term encompassing aggression) Those are probably the most relevant. There is an important consideration which does seem to be missing from the study, but o…

I did; do you understand why I am saying "DOG leaves food or objects alone when told to do so" or "DOG is quick to sneak out through open doors, gates" or "DOG is slow to respond to corrections" (among many other questions) are utterly irrelevant when people consider why certain types of dogs are dangerous?

> Most, if not all of these factors can be altered with behaviour modification (training).

What percentage of dog owners know how to train a dog properly, or hire someone who can? It just seems like you're not thinking about how dogs exist in the real world.

Re: Study of pet dogs shows breed does not predict behaviour

#115

Post by User23 »

TameAntelope wrote: post #113
barry-cotter wrote: post #111

Earlier quoted context omitted.

> There absolutely were ancient people who shared genetic traits (and therefore some expressed characteristics) with modern Swedish people. There literally must have been, that's how genetics work, and 5,000 years is barely a blink, genetically. That’s kind of my point. 5,000 years ago there were no people who looked like Swedes. Now there are because of selection pressures. A new genotype came into existence by work…

> 5,000 years ago there were no people who looked like Swedes. Now there are because of selection pressures. 100% false. You need to realize that. 5,000 years is not enough time to do anything significant, genetically.

It absolutely is if there is a severe population bottleneck that selects for some trait in the survivors.

Re: Study of pet dogs shows breed does not predict behaviour

#116

Post by thephyber »

echelon wrote: post #34
ShockTohp wrote: post #7

From the source: "The team found that some traits were more common in certain breeds. For example, compared with a random dog, German shepherds were more easily directed; beagles, not so much. And the authors’ genetic studies revealed that mixed-breed dogs with a particular ancestry were more likely to act in specific ways. Mutts with St Bernard ancestry, for instance, were more affectionate, whereas mutts descended…

This study/paper is ridiculous. Anecdotal, but I've had lots of dogs growing up. We have five right now. This does not reflect the reality I know. Shiba inu act like cats and prefer to spend time away from their person. They frequently get kicked out of dog parks for not being friendly to other dogs. It's documented and well-studied. They're not at all like other dogs. Corgis are protective, commandeering, and can be…

I don’t find the study ridiculous and I have met dozens of each breed you mentioned.

I find that most people assume the entire population of dogs grew up in a house and have been reasonably well looked after. In practice, I have seen lots of dogs that have that kind of history and lots of dogs that have had traumas that cause them to resource guard or be reactive to people and/or other animals (these behaviors are not bred, but are specific to individuals / life experience).

I totally agree that some behaviors that have been selected for have very high correlation with the breed, but lots of other behaviors are far less correlated. My Aussie will nudge me with his nose, more like the Corgis you mentioned and less like the Aussies you mentioned. Breed implies correlation of some phenotypes, not causation.

In the end, language matters. The title of the Nature article is a Rorschach. In practice, headlines are usually written by someone other than the author and almost always lose significant detail because they are short.

Definitions are really important here. I love herding breeds, but they were bred to nip at ankles and butts of livestock. Is that considered an “attack”? Does the person defining what behaviors are an “attack” have an incentive to exclude those small harm incidents? Chihuahuas are notorious for being aggressive and giving nips, but they present as small, quick incidents, whereas large breeds are more likely to bite down and/or swing the neck like a shark does. Does that difference bias the stats reporting/legislation choices?

Also, “Pit Bull” isn’t a breed. It is a vague intuition of some of the “bully breeds”. I always have to wonder if people pointing at attack frequency lists which show that inaccurate term are accurate. Did those lists gather data that actually investigated pedigrees or did they eyeball/guess breed based on assumptions?

Re: Study of pet dogs shows breed does not predict behaviour

#117

Post by barry-cotter »

TameAntelope wrote: post #113
barry-cotter wrote: post #111

Earlier quoted context omitted.

> There absolutely were ancient people who shared genetic traits (and therefore some expressed characteristics) with modern Swedish people. There literally must have been, that's how genetics work, and 5,000 years is barely a blink, genetically. That’s kind of my point. 5,000 years ago there were no people who looked like Swedes. Now there are because of selection pressures. A new genotype came into existence by work…

> 5,000 years ago there were no people who looked like Swedes. Now there are because of selection pressures. 100% false. You need to realize that. 5,000 years is not enough time to do anything significant, genetically.

Are you saying that there have always been people who looked like Swedes in Sweden? Because from my reading of the literature 5,000 years ago there people living there had blue eyes but dark hair and a complexion we’d associate with Mediterranean peoples.

Most dog breeds were created in less than 150 years. Call it 150 generations. 5,000 years is 200 generations at 25 years per generation. Enough time to change quite dramatically.

Re: Study of pet dogs shows breed does not predict behaviour

#118

Post by TameAntelope »

User23 wrote: post #115
TameAntelope wrote: post #113

Earlier quoted context omitted.

> 5,000 years ago there were no people who looked like Swedes. Now there are because of selection pressures. 100% false. You need to realize that. 5,000 years is not enough time to do anything significant, genetically.

It absolutely is if there is a severe population bottleneck that selects for some trait in the survivors.

...which has never happened in human history. Bottlenecks? Yes. Bottlenecks that selected specifically for a specific gene? No.

Re: Study of pet dogs shows breed does not predict behaviour

#119

Post by astrange »

_aavaa_ wrote: post #86
worik wrote: post #63

Earlier quoted context omitted.

> 1) All data is in distributions, and so no predictions can be made about an individual based on distributions. Good luck with getting people to accept that. It is obvious to any statistician, but I have had arguments with experienced researches who cannot grasp this simple fact

Maybe that means something different to me than it does to you, but you definitely can make predictions about individuals. Is that not what a prior? If I hear that someone is 12 years old and I know nothing else about them, then assuming their height is the mode for that age group is as good of a prediction as you can make.

A more important point is that the instant you meet someone, you have far more information about them than knowing their genetics would give you. So it'd be strange to meet a Dutch person and go "hah, by your genetics I predict you're 190cm tall" if you're looking at them and they're short.

Re: Study of pet dogs shows breed does not predict behaviour

#120

Post by TameAntelope »

User23 wrote: post #112
TameAntelope wrote: post #94

Earlier quoted context omitted.

All of those people you've named can and do look like one another, depending on circumstance. Every trait you can think of that you'd list as a "characteristic" of one of the groups you've named can be found in abundance in numerous other places as well, or accounted for due to environmental factors. There is no "canonical" "Swede" or "Arab". Your reliance on stereotype is not only racist but it's factually wrong. Th…

> Race does not exist, genetically. Period. And yet ancestry.com can comfortably tell anyone which continental scale human population or populations they are descended from and that happens to map pretty neatly to the US Census Bureau's notion of race. Being aware that major geographic features of the Earth such as the Atlantic Ocean, the Himalayas, and the Sahara Desert were more or less impermeable barriers to gene…

Ancestry.com simply does not do what you're claiming here. The whole "fun" of ancestry.com is that each person who does it shows up in surprising places across the globe, regardless of their supposed "race", because genetics don't adhere to any concept of "race" as it exists in culture.

And the this Barrier Theory was proposed by eugenicists in the 20s and thoroughly debunked by the 30s.

Post reply on HN