Skip to Content
MIT News magazine

Understanding Autism

MIT researchers are working to explain a baffling condition.
February 23, 2010

In the late 1980s, neuroscientist Mark Bear set out to study a cell receptor that appeared to play a role in making connections between neurons in the brain. He was interested in plasticity–the formation, strengthening, and weakening of brain connections that allows us to learn things and form new memories. Finding potential cures for Fragile X syndrome, a genetic disorder that can cause autism, was the furthest thing from his mind. “I had no idea what Fragile X was,” he says. “None.” But 20 years later, Bear’s research led to the extraordinary discovery that shutting off the receptor can reverse the symptoms of Fragile X in mice. Trials for drugs that block the receptor in humans are now under way.

Mark Bear

“It was a classic payoff of basic research,” says Bear, a professor of neuroscience at the Picower Institute of Learning and Memory.

Bear ended up studying autism serendipitously, but a growing number of other MIT scientists have set their sights on the condition in recent years as they’ve learned how serious and widespread it is. “Fifteen years ago, autism was thought to be a rare disorder,” says neuroscientist John Gabrieli, PhD ‘87. “People understand much more how common it is–and how difficult it is.”

About one in 110 American children has a disorder on the autism spectrum, which ranges from milder difficulties with communication and social interaction to much more severe deficits that may be accompanied by mental retardation. Children do not “outgrow” autism, but symptoms can improve with treatment, which usually includes a combination of behavioral, speech, and physical therapy. Doctors may also prescribe drugs to treat specific symptoms.

Under the umbrella of the Simons Initiative on Autism and the Brain at MIT, launched in 2009, researchers are tackling autism at all levels–from genes to brain anatomy to behavior. Their work so far has yielded some promising drugs for Fragile X and for a form of autism called Rett syndrome, as well as new devices that could help autistic children learn to cope better with social interaction.

Genetic influence

Children with autism appear to develop normally until about the age of two, when they start losing interest in other people, including their parents. They have difficulty making eye contact, reading social cues, and interacting with others, and they tend to exhibit delays in language ability, repetitive behaviors such as rocking, and an obsessive focus on order and routine.

Researchers have known for years that autism may have some genetic basis, and in the past decade they have identified dozens of genes that could be involved. Many of these genes play a role in the development of synapses, connections between neurons that allow them to exchange information. In autistic children, the brain’s ability to make those synapses appears to be impaired early in life–perhaps even before autism symptoms begin appearing. “The first two to three years of life are massively important for forming connections in the brain, and this is when autism strikes,” says Mriganka Sur, head of MIT’s Department of Brain and Cognitive Sciences.

Sur, Bear, and other molecular neuroscientists approach autism as a malfunction of plasticity–the phenomenon that allows the brain to change in response to experience or environment. “Disorders of development are disorders of how the brain is wired,” says Sur. “That’s why autism is so fascinating to me, because it maps so directly onto plasticity.”

About three years ago, Sur began studying Rett syndrome, a rare condition that was linked in 1999 to mutations in an X-­chromosome gene that codes for a protein called MeCP2. When neurons don’t have enough MeCP2, which is necessary for nerve-cell maturation, they can’t grow the tiny branchlike projections needed to form synapses. The result is heart abnormalities, seizures, severe speech impairments, reduced head size, and typical autism symptoms, including repetitive hand movements. Most patients are female; boys with the condition usually die before or shortly after birth.

These are fMRI brain activations in typical people (neurotypical, NT) and Asperger patients (AS) thinking about how words that describe traits (such as kind, smart, or lazy) apply to themselves, and whether the words are generally positive or negative. The control subjects engage a brain system known to be involved in self-reflection more powerfully than the AS people do. The bottom row of images quantifies this difference by directly comparing the activations of NT and AS participants. Warm colors in this image indicate areas of the brain activated more by the self-reflecting NT people than by self-reflecting AS patients.

Since Rett syndrome is caused by mutations to a single gene, researchers hope to cure it by overcoming the effects of that mutation. Sur appears to have found a promising tactic: in February 2009, he reported that a protein called IGF-1, an insulin-like growth factor that regulates development of nerve cells, spurs synapses to mature, reversing Rett symptoms in infant mice. Sur zeroed in on IGF-1 as a potential treatment after earlier work in his lab showed that it helps promote synapse growth. Clinical trials are expected to start this year in patients 2 to 12 years old. “Earlier is better,” he says, “but we believe the adolescent and even adult brain has potential for recovery of function in certain disorders of brain development.”

Those trials will join a handful already under way for the potential Fragile X treatment discovered by Bear. Fragile X, which causes learning disabilities as well as autism symptoms, is the most common inherited cause of mental impairment, affecting about one in 8,000 girls and one in 4,000 boys.

Children with Fragile X have a mutation in a gene that encodes FMRP (Fragile X mental retardation protein), a protein needed for normal brain development. An international team of researchers discovered the mutation in 1991, but they weren’t sure of the missing protein’s role in the brain. Later that decade, they discovered that it was often associated with a molecule Bear had been studying for years: the metabotropic glutamate receptor 5 (mGluR5).

Bear (then a professor at Brown University) knew that mGluR5 was involved in the neurophysiological phenomenon known as long-term depression (LTD), a suppression of synapses that’s part of the system the brain uses to fine-tune connections by weakening or strengthening them as needed. At first, he hypothesized that Fragile X protein was necessary for LTD: activating mGLuR5, he thought, would stimulate production of the protein and stifle unwanted synapse growth. He tested the theory by studying genetically engineered mice that cannot produce FMRP and thus display symptoms like those of Fragile X, including impaired learning ability, enlarged testicles, hypersensitivity to sensory stimuli, and increased susceptibility to seizures. To his astonishment, the mice showed exaggerated LTD–the opposite of what he expected.

The results were so strange that his team delayed publication to do experiments over and over, says Bear, who came to MIT in 2003. “We just had this orphan result which essentially got shelved for a while,” he recalls. Suddenly it occurred to him that maybe mGluR5 activates the process necessary for LTD and FMRP restrains it, preventing synapse development from being suppressed too much. That led Bear to wonder whether blocking mGluR5 could compensate for a shortage of FMRP, effectively restoring balance to synapse development and reversing Fragile X.

Bear admits to being “pretty nervous” the first time he went to a conference of Fragile X researchers to present the theory, in 2000. “It frankly seemed absurd to think you could correct a disorder as varied as Fragile X by this one mechanism,” he says. “I remember being relieved when I didn’t get laughed at.” But in 2007, Bear and his colleagues showed that halving the number of metabotropic glutamate receptors in mice with Fragile X symptoms does indeed reverse those symptoms.

Though research like Bear’s and Sur’s targets types of autism caused by single-gene defects, “the bottom line is that synaptic dysfunction is fairly common across multiple forms of autism,” Sur says. He hopes the drugs that researchers develop to target specific genetic problems with brain signaling can also treat other forms of autism that involve similar cellular malfunctions.

Seeing the big picture

While Bear and Sur tinker with the molecular underpinnings of autism, Earl Miller wants to know how large-scale brain networks produce autistic behavior.

One symptom of the disorder is a tendency to fixate on details. An autistic child may become used to brushing his teeth with a particular toothbrush, but “if the parent comes home one day with a new toothbrush and it’s blue instead of red, the kid falls apart,” says Miller, a professor of neuroscience at MIT’s Picower Institute.

That difficulty in seeing the big picture is rooted in an inability to categorize, says Miller, whose research focuses on high-level brain functions such as paying attention, recalling memories, and planning to achieve complex goals. For most people, categorization requires little conscious effort. For example, it may seem obvious that both a poodle and a pit bull are dogs. However, autistic patients have difficulty perceiving that dogs of two breeds, or toothbrushes of two colors, are different examples of the same thing.

Miller believes that categorization hinges on the relationship between the prefrontal cortex–the seat of many high-level brain functions–and the basal ganglia, a more primitive brain region associated with motor control, learning, and some vision processing. Visual information, he theorizes, flows from the basal ganglia to the prefrontal cortex, which pieces together the most important information and filters out unnecessary details; the whole process is controlled by influx of the brain chemical dopamine. In autism, however, the balance is thrown off. Elevated dopamine levels in the basal ganglia appear to be associated with unnaturally strong learning mechanisms there. The learning of details overwhelms the ability of the prefrontal cortex to piece them together into categories, with the result that the details dominate.

Miller is now testing this idea in monkeys, which can be taught to categorize objects. He hopes to show that dopamine levels are higher in the basal ganglia before categorization is learned and higher in the prefrontal cortex after it’s learned. Then he plans to investigate whether overstimulating the basal ganglia with dopamine impairs the monkeys’ ability to categorize. “Biology is all about balance,” says Miller. If it turns out that people with autism have an imbalance between learning in the basal ganglia and in the prefrontal cortex, for example, a drug could be developed to restore the correct balance and reverse some cognitive problems.

Communicating emotions

Research like Miller’s will probably take years to yield treatments, but a team in the MIT Media Lab is working on projects that could have a more immediate impact: helping people with autism manage the behavioral aspects of the disorder. Through the Autism and Communication Technology Initiative, for example, ­Matthew Goodwin, director of clinical research at the Media Lab, and ­Rosalind Picard, SM ‘86, ScD ‘91, a Media Lab professor, are finding a way around the difficulty that autistic children have in recognizing and communicating their emotions.

Goodwin, who spends two days a week at the Groden Center, a school for autistic children in Providence, RI, says that a child who appears calm in a classroom may actually be on the verge of an outburst. A teacher who tries to engage the child in an activity could unintentionally ignite aggressive behavior that appears to come from nowhere.

Those outbursts are often the result of stress caused by new people or situations, transitions between situations, or hypersensitivity to stimuli that others might not notice, such as flickering lights or low humming noises. Autistic children “have this constant barrage of sensory information they can’t understand,” says Goodwin. “They can’t tell you, ‘I am having a hard time making eye contact right now because I can see these flickering lights.’ “

Goodwin and Picard are building devices that measure reliable indicators of nervous-system arousal, such as temperature and sweating. The latest prototype is a sweatband with two electrodes that touch the inside of the wrist. A mild electric current, imperceptible to the wearer, runs between the electrodes, traveling across the skin. When skin becomes sweaty, it conducts electricity better, and the resulting voltage change can be measured and wirelessly transmitted to a laptop computer, PDA, or mobile phone.

Eventually, Goodwin hopes to build a wearable indicator such as a pin that would change color in response to those measurements, revealing whether a child is over- or understimulated. Teachers and parents would then know whether the child needed to be soothed (offering a cold drink of water is one approach) or stimulated by an energetic activity. The device could also help children learn to recognize their own emotions.

Understanding others

The Simons Initiative, funded by Jim Simons ‘58 and Marilyn Simons, supports about a dozen projects at the Institute. “There are very few basic science departments with as many people focused on autism as there are at MIT,” says John Gabrieli, whose work involves imaging the brains of autistic patients as they perform tasks. “Autism is such a pervasive development disorder. You can go at it in a lot of different ways.”

Gabrieli and MIT neuroscientist Rebecca Saxe, PhD ‘03, are studying a particular cognitive skill: the ability to make inferences about other people’s state of mind. In most children, “theory of mind” develops around the age of four, but autistic children usually have significant trouble interpreting other people’s mental states.

Gabrieli and Saxe have demonstrated this in a study of teenagers with and without Asperger syndrome, a disorder on the mild end of the autism spectrum. (Children with severe autism generally can’t handle the MRI process, which involves lying still in a large tube for the duration of the scan.) The researchers concocted moral scenarios and asked the subjects to judge the characters involved. In one, person A passes person B some sugar, and B puts it in his coffee. It turns out there is arsenic in the sugar bowl, and B dies.

Everyone confronted with this scenario agrees that if A knew about the arsenic in the sugar bowl, he has done something morally wrong. If A unknowingly passed the arsenic, however, control subjects do not say A acted immorally, but Asperger patients do.

“For most people, intentions are more important than the outcome,” says Gabrieli. But Asperger patients seem to have difficulty separating the two. He and Saxe are using functional magnetic resonance imaging to search for the neural basis of that impairment. (See for some sample images.) In previous studies, Saxe has shown that theory of mind appears to be seated in a brain region called the temporoparietal junction.

MIT neuroscientist Tomaso Poggio, meanwhile, is pursuing a new way to classify autism symptoms: he’s programming computer vision systems that can analyze behavior, helping to refine diagnoses and potentially allowing doctors to tailor future drug treatments to patients. Poggio, a computational neuroscientist, says better diagnostics are needed because autism takes a different form in each child. “With mental diseases, diagnoses are very qualitative and not very clear,” he says.

Poggio’s lab has developed a system that records and quantitatively analyzes behaviors in mice, but adapting it to human behavior is likely to take several years. With such a system, researchers could gather data from a large number of patients and correlate differences in speech patterns, motor coördination, eye movements, and reaction times with specific genetic mutations.

It may seem like a huge undertaking, but such efforts are essential to solving the puzzle of autism, says Poggio, who compares autism research to the “war on cancer” that President Richard Nixon declared in 1971.

“At first, people thought in a few years it would be solved. It turned out to be much more complicated,” he says. “But in the process, a lot has been learned, not only about cancer but also about molecular biology in general.” The attempt to unravel autism could be similar, he says: “I believe the problem can be solved at least to some extent, so I think we have to try.”

Keep Reading

Most Popular

DeepMind’s cofounder: Generative AI is just a phase. What’s next is interactive AI.

“This is a profound moment in the history of technology,” says Mustafa Suleyman.

What to know about this autumn’s covid vaccines

New variants will pose a challenge, but early signs suggest the shots will still boost antibody responses.

Human-plus-AI solutions mitigate security threats

With the right human oversight, emerging technologies like artificial intelligence can help keep business and customer data secure

Next slide, please: A brief history of the corporate presentation

From million-dollar slide shows to Steve Jobs’s introduction of the iPhone, a bit of show business never hurt plain old business.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.