Most researchers believe the former, but the proposition has been hard to prove. So Nancy L Segal, a psychologist who directs the Twin Studies Centre at California State University, Fullerton, decided to test it and enlisted an unlikely ally.
He is Franois Brunelle, a photographer in Montreal who takes pictures of pairs of people who look alike but are not twins.
Segal was sent to Brunelles website by a graduate student who knew of her research with twins. When she saw the photographs, she realised that the unrelated look-alikes would be ideal study subjects.
I reasoned that if personality resides in the face, she said, then unrelated look-alikes should be as similar in behaviour as identical twins reared apart. Alternatively, if personality traits are influenced by genetic factors, then unrelated look-alikes should show negligible personality similarity.
For 14 years, Brunelle, 64, has been working on a project he calls Im Not a Look-Alike!: more than 200 black-and-white portraits of pairs who do, in fact, look startlingly alike.
I originally named the project Look-Alikes, but some of the subjects did not feel they looked alike, he said. The new name gives ownership to the people I photographed and allows viewers of my website to decide for themselves if they look alike or not.
Most come to him through social media links to his website. It has taken on a life of its own, he said. I have heard from people in China and even a man who has an uncle in Uzbekistan who is a dead ringer for former President George W Bush.
Two of his subjects, Roniel Tessler and Garrett Levenbrook, met three years ago at the University of Michigan. When the two got together, at a pizza joint in New York City, we ordered the same toppings, said Levenbrook, 25. But other than that, the two have little in common. Tessler, 27, describes himself as a free spirit; he called Levenbrook his exact opposite the most focused and organised person I know.
For Segals initial study, she asked Brunelle to send questionnaires to some of his subjects, and she received completed forms from 23 pairs of unrelated look-alikes. The questionnaires yield a score based on five personality measures: stability, openness, extroversion, agreeableness and conscientiousness.
As she expected, the unrelated look-alikes showed little similarity. By contrast, twins especially identical twins average high scores on both scales, suggesting that the similarities were largely because of genetics. Her results were published in the journal Personality and Individual Differences.
Their analysis was consistent with the findings of Segals first study: Personality traits do not appear to be influenced by the way people are treated because of appearance.