Police dogs can distinguish identical twins

Being an identical twin might seem like a great way to fool a DNA test and get away with the perfect crime. But furry forensic experts can make sure justice is served. In a new study, researchers instructed a group of children, including two sets of identical twins and two sets of fraternal twins, to swab the insides of their cheeks and place the swabs in glass jars. Working with ten police German shepherds and their handlers from the Czech Republic police, the researchers then ran a mock crime scene investigation. The handler presented one twin’s scent to the dog and then told it to go find the matching scent in a lineup of seven jars, which included the other twin’s scent. In twelve trials per dog, none of them ever identified the wrong twin as a match, the researchers report online this week in PLoS ONE, even though the children lived in the same home, ate the same food, and had identical DNA. No word yet on whether these dogs will be getting their own CSI spinoff.

The story is here and the paper is here, for the pointer I thank Michelle Dawson.


This sounds too good to be true.

Ordinary people can also tell identical twins apart with 100% accuracy. For one, they look different. And Dogs are better with scent than we are with vision.

Interesting study. I wish there were more than 2 sets of identical twins though. I don't think the real sample set is 10 dogs * 2 sets of identical twins = 20 observations. Those two twins sets could be particularly easy to distinguish just by random chance, making it more of a sample size of just 2 observations.

"ate the same food"

Well, that's actually impossible. To be more precise, unless someone was regurgitating they probably ate from the same dishes but did not eat the exact same food.

Is there a potential problem with using swabs from the inside of the mouth? It would seem like this could allow environmental factors could be introduced into the study, thereby neutralizing the claim that dogs can distinguish between identical twins. This study may only prove that dogs can distinguish between odors from inside people's mouths.

Was everyone suitably "blind"? Could the dogs have reacted to the observers' emotions as they approached the jars?

My thought exactly. Dogs are very sensitive to their owners/handlers' body language and other non-verbal cues.

With different environmental exposures, even twins will have different bacterial cultures. Given the extent to which bacterial flora contribute to human odor, this result is not so surprising.
And yes, small sample size.

My little brothers are identical twins, and I can tell them apart with 100% accuracy too. Why is it surprising that dogs can?

It appears there is a lot of extraneous procedures and data that doesn't test the hypothesis: do police dogs distinguish identical twins? The trials with the same twin, the "distractors", the rearrangements of the trials for the different twins. I do not disagree that dogs can pick someone's scent out of a line-up. There didn't appear to be any trials with both twins in the line up! So at best this is really a test of the probability of a false alarm on the wrong twin.

By my read there are actually only 2 trials per dog, the 1Ax1B and 2Ax2B in the table, each with 1 correct result, for 20 observations as Brian points out above. Doing a simple estimate, we'd expect between 1 and 19 correct guesses 95% of the time assuming a hit/miss probability of 50%.

I also want to second Raj; I am not adverse to this being the "correct" result -- everyone's bacterial cultures are different, as well as our epigenomes ...


Pardon the pun, but this doesn't pass my smell test. Only studying two pairs of identical twins is insufficient and, more so, both were sets of children. I immediately reject the study because they used control trials of non-twins as evidence of distinguishing twins (this is the setup in the top of Figure 2).

As other noticed, there are some obvious flaws with the experiment design. But even assuming the main result is true (which is IMO very likely, just not with such a perfect rate), it only speaks of the awesome power of canine olfaction. MZ twins are clearly different and almost any human can tell their looks apart. Even with regard to their DNA, the differences between them can be detected on at least two levels: DNA methylation and somatic mutations. So they ought to smell differently, too.

As a reward, the dogs were given Doublemint Gum.


How often does a police dog "distinguish" an identical twin from himself?


As police dogs are not much above 50% in accuracy in identifying marijuana, I find this report incredible.

Do any of the prior commenters have dogs? I'm no scientist and certainly agree that bigger trials are always better. But having worked with hunting dogs in the field, I've seen some amazing feats of scent. A bloodhound--I know, I know, these were GSDs--can follow a cold trail that's several days old. Why would distinguishing twins--who may eat different things, wear different clothing, wash with different soap etc.--be so hard?

I'm not surprised at all, as it's been shown they can smell cancer, which probably even more like you than your twin.

Ah, the vast array of olfactory information we gave up access to when we started walking upright. Who knows how much more advanced a society we could be with such discerning noses.

Like many others, I'm not too impressed with the study, but at the same time can easily believe the result is true.

Smell is more powerful than we think, even for humans - a mother can distinguish the t-shirt worn by her newborn in a nursery from those wornby other newborns. And a dog's sense of smell is several orders of magnitude more powerful than a human's.

Comments for this post are closed