Feature Story Fall 2016 Issue 39
Peering into the Infant Mind
by Elizabeth Dougherty | December 12, 2016
Mothers often speak with a sense of wonder about seeing their new baby’s face for the first time. For McGovern Associate Investigator Rebecca Saxe, an even more wondrous moment came when she first saw her baby’s brain.
“I thought, wow, that will be my baby’s mind and my baby’s self,” she says.
Saxe, who is also a professor in MIT’s Department of Brain and Cognitive Sciences, had produced the image as part of an ambitious research program to study baby brains using functional magnetic resonance imaging (fMRI). The program builds on earlier studies of adult brains by McGovern Associate Investigator Nancy Kanwisher, Saxe’s former mentor who is also a collaborator on the new baby study. Kanwisher had found that the adult human brain has regions that respond specifically to images of faces, scenes, and other classes of visual stimuli.
Kanwisher’s findings posed a question: are these specialized brain regions formed by visual experience, or are they already present from birth? The question addresses a fundamental and longstanding debate about human nature: does the human brain and mind start out as a ‘blank slate,’ waiting to be shaped by experience, or are we pre-wired to perceive the world in ways that are shaped by our biology? Brain imaging could help provide the answer, but scanning the brains of young infants is far more difficult than studying adults, and the question remains unanswered two decades after Kanwisher’s original discovery.
Saxe has now taken up the challenge, exploiting new advances in fMRI and functional near-infrared spectroscopy (fNIRS), both of which measure changes in blood flow when particular brain regions become more active. Her findings are providing scientists with the most detailed images ever produced of brain activity in young babies. The results that have emerged so far suggest that baby brains are more specialized than anyone had realized. They also have important practical implications, raising the possibility of identifying developmental disorders early enough to make corrective changes before troubles amplify over time. “There’s a very strong push now to intervene early in developmental disorders,” says Saxe. “If we want to do that using brain scans, we first need to understand what typical development looks like.”
The Long Road to the First Scan
Saxe had been thinking for many years about using brain imaging to study awake babies. As a graduate student with Kanwisher, she had worked at the leading edge of fMRI studies of adults, and she had also done behavioral experiments with babies during her postdoctoral research at Harvard. “From the outside, baby behavior looks unsophisticated and random,” she says. “But behavioral research has revealed that there are richer,more elaborated conceptions of the world in their minds that are not obvious from their actions.”
After returning to MIT in 2007 to set up her own lab, Saxe and her then-graduate student Ben Deen began to plan a program to study young infants using fMRI. They first needed to obtain formal approval based on a safety evaluation. The scans themselves were deemed safe for babies, but there were concerns about noise, given that some scanning protocols can be as loud as a rock concert, around 125 decibels. So Deen worked with Atsushi Takahashi, the MRI physicist at the Martinos Imaging Center at MIT, to develop new scan sequences that would soften the potentially ear-damaging sounds inside the scanner. Deen also had to build a new head coil specifically designed for baby-sized heads. Solving these technical problems took several years of work, but finally, in 2013, the team was ready to do their first scan. They also had their first subject—Saxe’s first son, Arthur, who was then four months old.
They put him in the scanner and showed him movies of faces and scenes. No one knew if the images would reveal a functionally organized visual cortex or an unstructured blank slate. “I genuinely had no idea what to expect,” says Saxe.
The images of her son’s brain revealed an area of cortex that responded to images of natural scenes. That area was in the exact same place as in adults. “It was amazing to see,” says Saxe.
That first scan unleashed a wave of questions and ideas about how to answer them. Was part of the brain pre-wired to recognize scenes, or had her son’s scant visual experience already shaped this region? If experience had caused this region to develop, why didn’t the data show a similar region for faces? Her son had certainly seen more faces than forests and mountains.
“It was like the early days of fMRI,” Saxe says. “We’d get a day’s worth of data and by midnight we’d be emailing each other with new ideas for experiments.”
Burping, Sleeping, Crying, Scanning
After the initial excitement, though, progress was slow. fMRI scans, like photos, are very sensitive to motion, and head movements as small as a tenth of a millimeter can blur the data. Unlike earlier studies, in which babies were scanned while sleeping, Deen and Saxe needed them to lie still while remaining awake and watching a movie. Babies seldom remain still for long, and the researchers had to hope that their subjects would not burp or start crying or need a diaper change half way through the scan.
Eventually, by the end of the study, they had managed to scan 17 babies. All told, the team booked 126 hours in the scanner and obtained 23 hours of actual scan data. Much of it was corrupted by motion and had to be discarded; ultimately the researchers were left with just 4.3 hours of usable data, from 9 subjects.
As they analyzed the results, Saxe and her colleagues realized that the initial picture from her son’s first scan was incomplete. His brain showed responses to scenes but not faces, but with more subjects it became apparent that by 4 to 6 months, babies already have specialized visual areas that respond preferentially to faces or scenes. The responses are not as finely tuned as in adults, suggesting that later experience is also important, but the overall organization was remarkably similar to that of adults. “It’s lovely to see,” says Lindsey Powell, a postdoc in the Saxe Lab who is also scanning babies. “It’s not just a single region, but a whole network of brain regions seen in adults.”
Saxe now hopes to untangle how experience shapes these specialized regions. Lab manager Heather Kosakowski has started scanning individual babies repeatedly over time as they mature from three to six months. She is using a method known as multi-voxel pattern analysis to discover how the brain’s response to different types of images changes over the first few months of life.
She is also using images designed to test babies’ visual sophistication. For example, she presents babies with images of faces looking directly forward and faces in profile. For very young babies, sideways views may be categorized as objects, but a few months later, they are recognized as faces.
The biggest challenge will be working with the babies under time pressure to capture data during what is often a very short window of opportunity. If anyone has the patience and persistence to do this, it’s Kosakowski, who emerged from a childhood in foster homes, worked through 5 years of service in the US Marine Corps, put herself through Wellesley College as a single parent, and now is preparing for graduate study in neuroscience. “Allowing myself to believe I can go to graduate school is also allowing myself to believe that the questions I find interesting matter,” she says. “It matters that we understand how the infant brain develops.”
In addition to understanding how experience shapes the infant brain, Saxe and her colleagues also want to learn how the infant brain shapes its experiences. “Babies are active learners in their world,” says Saxe. “They choose what to pay attention to. But we don’t know what motivates these choices.”
Powell, who is leading this project, spent overlapping time with both Saxe and Kosakowski in the Harvard lab of Elizabeth Spelke, one of the world’s foremost experts on babies’ cognitive development. “In behavioral research, the main thing we can measure is where they’re looking and for how long,” says Powell. “We can’t ask babies what’s going on in their heads.”
But now, thanks to a new technology known as fNIRS, Powell can measure what is going on. Like fMRI, fNIRS detects changes in blood flow during brain activity, but instead of lying inside a MRI scanner, babies can move around freely during fNIRS scans. They simply wear a fitted cap with an array of light sources that shine light through the skull into the brain — “just like flashlights,” says Powell, the first researcher at MIT to use the technology. Sensitive detectors record the small amount of light that is scattered back through the skull, giving researchers a picture of the activity in the underlying brain regions.
Powell is focusing on two areas in particular: the medial prefrontal cortex, which is active when babies pay attention to a friendly face over an unfriendly face; and the dorsolateral prefrontal cortex, which becomes more active when babies pay attention to something informative, such as a person speaking a sentence over a person speaking gibberish. Babies tend to prefer friendly faces and informative speech, and the brain signals that Powell is recording may provide clues to how babies make these choices, which determine what they experience and perhaps shape the further development of their brains.
The work may also shed new light on the origins of autism. For instance, it could be that babies with ASD pay less attention to faces because they are less socially motivated and less interested in faces than typically developing babies. Alternatively, they could find faces uninformative or overwhelming.
“Our experiences shape our brains, but our brains also determine our experiences,” says Saxe. “It’s a back-and-forth that continues throughout development. My hope is that as we understand it better, we can start to ask how it goes wrong in developmental disorders, and how we might intervene to produce better outcomes.”
To learn more about volunteering as a research subject, please visit our website or email the Saxe Lab at firstname.lastname@example.org.