Those Lips, Those Eyes

Professor, Students Investigate the Secret to Recognition

by Mark Graham '97


"During my first year at Oberlin, I was looking for a winter-term project, and I just went and talked to Jim Tanaka, my advisor," says Jesse Lanz, a junior psychology and history major from Hubbard, Ohio. "That led to a winter-term project, which led to a really good job."

Lanz's "really good job" coordinating Tanaka's facial-recognition lab gives him more than a chance at experimental psychology. It gives him the chance to be published in a scholarly journal - an opportunity the vast majority of undergraduate students never get.

Lanz and Tanaka "It works out well in grad-school applications and career searches for our students to have their names on scholarly papers," says Tanaka, an associate professor of psychology. "You don't find that at most colleges or the larger research universities."

Oberlin can offer its students research opportunities because many faculty members blur the distinction between research and teaching by including students in their experiments and analysis, he says.

Tanaka's research investigates how people recognize faces. Using morphing technology that can match corresponding facial points on two faces, Tanaka and his research assistants can create a composite image that blends the two faces.

Filmmakers use similar technology for special effects in movies such as Forrest Gump and the upcoming Star Wars prequels. Conan O'Brien has turned less scientific morphs of celebrity faces into a running sketch on television's Late Night.

"When we morph the two parent images, the morph sometimes resembles one parent image more than the other," says Tanaka, noting that many people think the morphs better resemble the atypical faces - those with longer noses, smaller mouths, and the like.

"Since the morph is a mathematical average of the two, there must be some psychological reason for us to think that the morph looks like the atypical face," Tanaka says.

"When you look at a face, you look at the distinctive features that differentiate that face from other faces," he explains. By definition, an atypical face has distinctive features, so a morph resembles the atypical face in the mind's eye.

Tanaka and his lab group have also tested if we distinguish other objects, such as birds or cars, the same way we distinguish faces. Test subjects found that morphed images of both birds and cars resemble their atypical parent.

To run the experiment, the lab group needs faces. One of Lanz's first lab duties was to find them.

"We take college yearbooks from Oberlin and other schools and look for faces that are not smiling and don't have hair in the way, glasses, or facial jewelry," he says.

Once faces are chosen, Lanz and his labmates, freshmen Brendan Mislin and Diana Register, sophomores Jasmin Phua and Victoria Ravin, and juniors Dana Gruber and Ruchi Srivastava scan the faces and uniformly size them. In a voting process, the lab group chooses the best typical and atypical faces for morphing.

Test subjects - most are students enrolled in Introduction to Psychology - run the experiment on computers in the psychology department's computer lab. The computer shows a set of two faces followed by the morphed image. The subjects then pick which of the original faces the morphed image better resembles.

When they finish, Lanz and his labmates debrief the subjects on the project's psychological principles and experimental procedures. "For the subjects, this is the most important part of participating in the experiment," he says. By participating, they get a glimpse into the world of experimental psychology.

But, the lab assistants aren't limited to the "grunt work" of the research, Tanaka says. "They help design and run the experiments, work with the data, and analyze the results."

Lanz says he enjoys the lab work and plans to research facial recognition next year for his honors project. In addition to the chance to be published, Lanz says he's happy to "come out of this with research experience most college grads wouldn't get."