If you look at drawings of people and their dogs, you'll notice they're often drawn alike: for instance, saggy-faced bulldogs with saggy-faced old men. But is the stereotype really true? A recent study looked into it.
The psychology of picking a pooch. I'm Bob Hirshon and this is Science Update.
We've all heard that dogs tend to look like their owners. Now, it's been scientifically confirmed—at least for purebred dogs.
Psychologist Nicholas Christenfeld of the University of California, San Diego took pictures of dogs and their owners. Then they showed the pictures to other people, and asked them to match them back up.
And the finding was with purebred dogs, they could do it above chance. Not perfectly, but significantly better than chance. Whereas with mutts they were just at chance, they couldn't tell better than flipping a coin which mutt went with which owner.
Why the difference? Well, Christenfeld says that while mutts are often adopted on impulse, or through friends, purebreds are carefully selected for predictable traits. And humans are naturally drawn to animals who look—and act—like they do.
And looking at pictures of the matches it really does suggest both sorts of things are happening. We have one picture of a guy with shaggy hair and a goofy grin, and his dog in fact is a retriever, and it's got the same shaggy hair and exactly the same goofy grin, and it's this combination of style and appearance.
But he says we shouldn't be too surprised. Studies show that we also tend to resemble our human companions. I'm Bob Hirshon for AAAS, the Science Society.
Making Sense of the Research
Just because everyone believes something doesn't make it true. That may seem obvious, but it's one of the most fundamental ideas of modern science. That's why every idea needs to be methodically tested before it's accepted.
In this case, the question was whether dogs looked like their owners. If you want to test this idea, you have to make a few decisions. First, you need to decide what your standard of a "dog looking like its owner" is going to be. How do you measure something like that? Will it be a simple yes-no answer? Will it be some kind of score? Who will make the judgment?
In Christenfeld's case, he decided that a dog looks like its owner if other people think so. And by "other people," he meant people who had never seen either the dog or the owner before. This makes sense, since owners might be biased in their opinion, as might their close friends. One owner might think he looks like his dog because he loves the dog so much; another might deny that he looks like his dog because he doesn't want to look like a dog. Either way, the owner probably isn't the most objective judge.
Next, he had to figure out how to tell if a stranger believed a dog looked like its owner. He could just show the stranger a picture of a dog and a person, and ask "Do they look alike to you?" But again, that might bias the results. People tend to be suggestible; just asking if a dog and a person look alike might make you more likely to think they do. Also, how would Christenfeld distinguish between an answer like "Oh yes, they definitely look alike" or "Well, I guess they kinda do?" Would they both count as a "yes?" Would they be assigned a numerical value, like a 1 to 10 scale? And how exactly would you do THAT?
So Christenfeld decided to test the subjects instead. He showed people a picture of an owner, the owner's dog, and someone else's dog. (The pictures were all taken separately against different backgrounds.) Then he asked the subject to match the owner to the dog. He tried this with a whole bunch of people. And he decided that if significantly more than half of his subjects correctly matched a particular owner to a particular dog, it meant that the dog looked like its owner. If only about half guessed correctly, that's what you would expect if they were doing it blindfolded, so the dog was considered not to look like its owner.
Interestingly, he got two different results: purebred dogs tended to look like their owners, but mutts didn't. He guesses that this is because purebred dog owners tend to choose their dogs based on appearance and personality, while mutt owners tend to be less finicky. Of course, if he wanted to prove that for certain, he'd have to come up with another experiment.
Now try and answer these questions:
- In this experiment, what does it mean to say that a dog looks like its owner?
- What are the advantages of Christenfeld's method?
- What does the study not answer?
- If you were asked to make the study even more reliable, what changes would you make? What if you were asked to provide more detailed information?
- Can you think of another old saying or "common sense" idea that lots of people believe? How would you test to see if the idea was really true?
The Skeptics Society is dedicated to the preservation of science and critical thinking. They have a special interest in examining widely held beliefs.
Bad Science is a site devoted to exposing common scientific misconceptions.