The Scientist

As a child, Doris Tsao spent long hours musing on the mechanics and philosophy of vision with her father, who owns a company that designs artificial vision systems. "He made vision seem like the greatest scientific problem," she says. By the time Tsao was 11 or 12, she'd been hit by "the realization that your sense of vision is created by your brain" - and her fate as a brain researcher was sealed.

Tsao studied biology and math as an undergraduate at the California Institute of Technology, and in 1996 began her doctorate at Harvard Medical School, working on stereopsis (depth perception) with vision neuroscientist Margaret Livingstone. Researchers were just beginning to use functional MRI to study maps of human brain activation, and Livingstone sent Tsao to Massachusetts General Hospital to work with fMRI expert Roger Tootell and establish the technique in monkeys. Tsao thought she could also electrophysiologically probe activated areas of monkey's brains, thus combining fMRI's birds-eye view with the fine detail of single neuron recording.

"The big challenge was getting the monkey in the scanner," says Tsao, since they can't lie on their backs, as human scanners require, and are startled by the machine's loud clangs. Tsao had a friend taking a class at MIT's Media Lab build her a 'monkey chair,' and by 2001 she was running her first experiments.

Tsao began with stereopsis, but soon moved on to a topic that had also captured her imagination. In 1997, then Harvard University researcher Nancy Kanwisher used fMRI to identify a human cortical area specialized for face recognition. It wasn't clear, though, whether monkeys had such a dedicated module: Single-neuron studies had suggested that their face cells were scattered throughout the temporal cortex.

In 2003, Tsao and Winrich Freiwald, a monkey electrophysiologist and at the time a postdoc with Kanwisher (now at MIT), identified face-selective patches in monkeys using fMRI.1 And when they recorded from neurons in the patches, almost every cell responded to faces - surprising, since past studies suggested a maximum density of about 30% for responsive neurons.2 "It was one of the most exciting days ever," she recalls. The high concentration of responsive cells provided the strongest evidence so far that the brain essentially contains a face processing machine.

"It's lovely when someone takes a whole history of the field and says, 'No, it's not quite like you think it is,' and twists it back around" to present a new view, says Livingstone. In 2004, the Alexander von Humboldt Foundation awarded Tsao €900,000 to start a lab in Germany, and in 2006 she moved to the University of Bremen, where Freiwald was already working. The duo went on to identify six distinct cortical regions involved in face processing, each containing neurons with characteristically different firing patterns.3 They also examined connectivity between the different regions, and most recently, identified additional face-selective areas in the prefrontal cortex, the "executive" area thought to play a role in mood, emotions and behavior. "Now we can really study the whole system," she says.

Next year, Tsao will return to Caltech as an assistant professor in biology. She intends to return to stereopsis and other aspects of 3- dimensional vision, but there's no turning back with her work on face processing. Already, says Freiwald, "We are opening all these possibilities - almost too many for us to study them all."