Photograph by Glen Mitchell

In a recent article over at Slate, I reviewed an astonishing new set of findings from Japan showing that subjects can correctly match people to their pets when given only a paucity of physical cues. That research by Sadahiko Nakajima revealed that its clearly the eyes that give it away: when you cover the eyes of either the dog or the dogs owner in the images participants choose from, their ability to match the right interspecies pair falls to chance. Mindboggling? You bet. But now Ive got something to really get your chassis rattling. In a study published earlier this year in Swiss Journal of Psychology, University of Vienna psychologists Stefan Stiegar and Martin Voracek found that not only do we look like our dogs in this vague but empirically demonstrable way, our faces also bear an uncanny resemblance to the frontend views of our automobiles. To top it off, our dogs mug shots apparently bear an objective similarity to our cars faces as well.

Previous research by the anthropologist Sonja Windhager and others had shown that when we look at cars from the front, we implicitly perceive faces staring back at us. Windshields evoke foreheads, headlights remind us of eyes, side-view mirrors are stand-ins for ears, grills do the trick for noses, and the additional air intake conveys a mouth. Of course, those of us who are saneand even most of us who arentdont actually believe that cars have faces. But that doesnt stop us from processing them this way. In fact, Windhager found that, just as we do for human faces, we attribute different traits and personality characteristics to different models of cars due to their unique frontend designs. The VW Beetle is seen as being happy, for example, the Honda Civic has a neurotic appearance, and the BMW 645ci gives off a dominant and angry scowl.

Anthropomorphizing cars due to their intrinsic physical affordances is one thing. Anyone whos seen one of the Disney Cars movies knows how easy it is. But choosing our cars because they physically resemble us is another thing entirely. And yet, thats precisely what Stiegar and Voracek have found that we do. Their initial study was quite simple. First, the authors took black-and-white facial photographs (neutral expressions) of 30 random people, along with separate front-end images of these peoples cars transformed into basic grayscale. The key here was that the owners had to have personally selected their cars for themselves, not received them as, say, a gift from a spouse or a hand-me-down from a relative. Next, the authors developed 40 different sets of images, each set containing one car photo at the top and the faces of 6 possible owners below, including three men and three women. (Each owner was shown only once in the entire collection of images.) The actual car owner was among these half-dozen photos, but randomly placed among the targets.

The task for the 160 participants who were then presented with these image sets was straightforward. On a scale of 1 to 6 (most to least likely the owner) they were asked to rank the possible owners of the given car.

Just as we human beings unknowingly tend to choose romantic partners who look like usa rather striking fact falling under the principle of assortative matingStiegar and Voracek predicted this egocentric bias would apply to the cars that we drive as well. When you combine this with Windhagers earlier findings about our attribution of certain humanlike traits to different car models on the basis of unique front-end styles, the authors suspected that the judges in their study would be able to match cars with their correct owners above chance levels. And thats what they found. The real owner was in fact assigned rank 1 most frequently, they write, and rank 6 least frequently. This proved true regardless of the subjects sex and age. There were an equal number of male and female judges, and they ranged widely in agefrom 16 to 78 years. In case the sheer bizarreness of these data hasnt quite registered, let me put it to you more bluntly: The average person can detect a physical similarity in the faces of cars and their owners.

To rule out alternative, simpler explanations for these effects, such as cultural or gender stereotypes about the types of people who drive certain kinds of cars, Stieger and Voracek ran a second study in which they included images of the automobile from the side and rear, in addition to the front. After all, most car models are easily distinguishable from any of these three angles (think of a Corvette, for instance), so if the subjects were using mere stereotypes to inform their rankings, wed expect them to be equally good at matching cars with their correct owners no matter the angle of the car that theyre shown. Again, this new crop of judges (male and female volunteers aged 14 to 79 years) could connect the cars with their correct owners at a level greater than chance, but only for the frontends of the automobiles. In other words, its indeed the physical likeness between the owners faces and those of their cars thats driving this preternatural matching skill.

Implied in these results is the startling fact that most car owners are unwittingly purchasing cars that look like them. If thats the case, figured Stiegar and Voracek, then is it possible that judges can even take it one step further, matching dogs to their masters cars? After all, we know now that its not a myth: dogs really do look like their owners. And since we choose both cars and dogs that physically resemble us, shouldnt our dogs and our cars look alike too? Here, frankly, the data just get weird. Nevertheless, theyre genuine. In their third and final study, the authors added 36 portraits of dogs into the mix. Half of these were of purebreds, and the others were mutts. In a twist to the previous studies, a new group of judges saw an image of a car (again, either the front, side, or rear view) and beneath that, six individual dogs. Subjects ranked each dog on the likelihood of its master being the owner of the car shown. Amazingly, the participants were able to pull this feat off as well.

Well, sort of. Just as before, the judges could only match the car with the correct dog when they saw the front ends of the vehicle. That tells us that the underlying physical similarity between people and their petsa resemblance in the faceextends to pets and their peoples cars as well. Yet the judges could only do this for the purebreds, not for the mixed breeds. To explain this discrepancy, Stiegar and Voracek speculate that although dog owners who acquire their best friends as puppies dont know exactly what the animal will grow up to look like, the adult faces of purebreds are far more predictable than those of mutts.

What to make of all this, Im not entirely sure. Personally, I drive a Fiat 500 and own two border terriers. Ive never noticed any obvious similarity between my car and my dogs, nor any of the above with what I see reflected in the mirror each morning. Were all a little annoying in our own ways, but not without our redeemable qualities. And where financial constraints enter the picture isnt clear either. I dont know about you, but my true doppelgangerobviouslyis a Ferrari. I just cant afford my own inflated ego, er, I mean, my vehicular visage.