2014: TSAO


We all take for granted how easily we can recognize the faces of our friends and family members, but the brain has to perform complex computations to enable this crucial perceptual ability. By discovering and characterizing a network of six distinct face-selective regions in the temporal lobe and three regions in the frontal lobe using an innovative combination of approaches, Doris Tsao, assistant professor of biology at the California Institute of Technology, has revealed important insights into face recognition in both monkeys and humans.

“In the brain sciences, perhaps more than in other fields, it is clear that we need researchers to step back and look at the grand picture,” said Rudiger von der Heydt, professor of neuroscience at Johns Hopkins University. “Doris is able to do this, partly because of her broad knowledge of neuroscience, but also by a natural talent of what I would call cheerful independent thinking. At her young age, she has a remarkable track record in neuroscience.”

For her seminal contributions to the field of visual neuroscience, Doris Tsao has been named the recipient of the 2014 Golden Brain Award from the Berkeley, California-based Minerva Foundation. The award, now in its 30th year, recognizes outstanding contributions in vision and brain research. Tsao was honored in a private ceremony at the Society for Neurscience's 44th annual meeting in Washington, D.C.

“When I heard I got this award, I felt simply incredulous, since many of the people who had gotten it previously are my scientific heroes, whose work has deeply inspired and motivated me,” Tsao said. “I am just starting out in my career, and this award is an amazing piece of encouragement.”

As a graduate student in the lab of Margaret Livingstone, professor of neurobiology at Harvard University, Tsao published one of her first groundbreaking papers on face processing in Science. In that study, Tsao and her collaborators pioneered the use of two approaches that provide complementary information about visual processing. They used functional magnetic resonance imaging (fMRI) in macaque monkeys to search for the largest brain region that responds selectively to faces compared with objects, and single-unit electrophysiological recordings to characterize the responses of neurons within the brain region identified by fMRI. Nearly all of the visually responsive neurons in this region were strongly face selective, indicating that a dedicated cortical area supports face processing.

“This discovery was important because it opened the door to systematically understanding the neural machinery for face perception,” Tsao said. “For the first time, we could access on a daily basis a large, homogenous population of cells all involved in coding the same basic form, a face, and ask how they are doing it.” 

Two years later, Tsao and her collaborators reported in Science the discovery of a network of six discrete but strongly interconnected patches of face-selective cortex in the temporal lobe of each hemisphere in the macaque monkey brain. That same year, the researchers reported in Nature Neuroscience that monkeys also have three highly face-selective patches in prefrontal cortex, which is involved in higher-level functions and could contribute to remembering or paying attention to faces, or even social reasoning about faces. Another study revealed that face-selective regions in macaques are similar to those in the human brain, suggesting that core principles about face processing discovered in monkey studies may generalize to humans.

“Because of the forces of social evolution and the importance of faces in primate evolution in particular, primates have developed this specialized neural machinery for face processing,” said Leslie Ungerleider, senior investigator at the National Institute of Mental Health and a previous Golden Brain Award recipient. “I really think Doris’ contribution has been the discovery that it’s not just cells, it's a network of areas. But we don't yet know the specific contributions of the different components of the network. That’s a big question for the future.”

Tsao and her collaborators have begun to tackle this question. Focusing on the largest face-selective patch in the macaque brain, they found that these neurons detect distinct constellations of face parts. Moreover, the cells are tuned to the geometric shape of various face parts, such as nose width and the distance between the eyes, especially when the features are present on whole, upright faces. The findings suggest that these neurons use a combination of part-based and holistic strategies to detect faces.

By analyzing the activity of neurons across the face patch system, Tsao and her collaborators found that cells in some patches responded to specific viewpoints of faces, while cells in other patches were not sensitive to face rotation or showed only partial sensitivity. The findings suggest that each of these regions in the macaque brain plays a different role in face processing, such as detecting faces among non-face objects, or linking together different views of the same face. “Ultimately, I hope this research will reveal new principles about the steps by which a complex visual object is represented in the brain.”

This research interest has early roots for Tsao. “Sometime around the end of high school, I formulated the scientific problem I wanted to solve: How can a piece of tissue perform geometry and create a vivid percept of objects in 3D space?” she said. “Twenty years later, this remains the problem I am working on. My dream is to understand the visual system with mathematical clarity, the way physicists understand the universe.”

Having earned a double major in math and biology as an undergraduate at the California Institute of Technology, Tsao is now making use of her broad interests and educational background to tackle one of the most challenging problems in vision: how to recognize a face or object despite dramatic changes in appearance due to changes in perspective. To do so, she has teamed up with her mathematician father, Thomas Tsao, to develop a new mathematical theory of object vision, and she is currently testing some of the models’ surprising experimental predictions. “This mathematical work brings me back to my original dream: to understand the cortical geometric engine responsible for creating objects in space,” Tsao said. “I am very excited that I can now bring together my experimental and mathematical interests and begin exploiting this system to ask deeper questions about object representation.”

And all signs indicate that Tsao will find the answers she’s looking for. “Among her peers, there is no one who stands out like Doris or has had an impact as Doris has,” Ungerleider said. “Given all of her major contributions thus far at such an early stage in her career, we can all expect a bright and shining future from this rising star.”