fb-pixelMIT team suggests autism may be tied to ability to predict - The Boston Globe Skip to main content
Science in Mind

MIT team suggests autism may be tied to ability to predict

Autism can seem like a perplexing collection of disparate symptoms, ranging from repetitive behaviors to impaired social skills. In a new paper, Massachusetts Institute of Technology scientists propose a common thread that could explain many of the facets of the disorder: a failure to make good predictions.

After reviewing dozens of studies and personal accounts by people who have personal experience with autism, researchers suggest that behaviors seen in autism may be a way of responding to a world that seems random and uncertain because their brains are unable to predict what will happen next.

For example, people with autism spectrum disorder may perform repetitive behaviors because personal habits and rituals are a safe harbor in a world they find alarmingly out of control.


People with autism may have difficulty in understanding social situations because they are poor at auguring how a particular social situation will affect what a person does next. And this predictive deficit might explain why many people with autism remain so good at things that do not involve dealing with a dynamic situation — such as memorization, playing music, or math.

“If you look at the reports from parents and other caregivers and individuals with autism, there is frequent mention of the need for a structured environment,” said Pawan Sinha, a professor of vision and computational neuroscience at MIT who led the work published last Monday in the Proceedings of the National Academy of Sciences. “To the extent a child with autism is distressed from having any deviation from structure, it means they have a hard time dealing with the unpredictability that comes from those deviations.”

The paper is far more theoretical than a typical piece of neuroscience research and is sure to stir debate.

John Elder Robison , an Amherst resident who has written extensively about his own experiences with Asperger’s syndrome, said that although he respects any thoughtful study of autism, he disagrees with the researchers’ conclusions.


“A piece of work like this is an example of how, when researchers don’t sufficiently engage with the community they are researching, they make wrong assumptions and go down paths that are less than optimally productive,” said Robison, who is the neurodiversity scholar-in-residence at the College of William and Mary.

Robison said that he would describe the core problem he faces not as a deficit in prediction — which he thinks he does reasonably well, as an extremely logical person — but as difficulties in perception.

Robison said that in his experience, social situations are confusing because he starts with a different data set from which to make predictions than typical people. That’s because people with autism often do not observe or perceive the same things that other people do in a social situation.

“Our problem is that we have no clue what is being asked of us. That means we are not sensing the things that other people are sensing to carry on a conversation — the whole idea of my book being called ‘Look Me in the Eye’ is an expression of the reality that the automatic, instinctive thing of looking a person in the eye doesn’t happen” for people with autism, Robison said. “That has nothing whatsoever to do with my predictive ability.”

The MIT researchers plan to do an experiment to test whether their hypothesis holds any kernel of truth. They are designing a video game that requires people to make accurate predictions in order to win the most points. They will test how people with autism perform on it and compare them with people without autism.


The researchers believe that if people with autism turn out to be bad at prediction when making complex judgments about how to collect the most coins in a video game, it may indicate their predictive ability is impaired on tasks that occur quickly and automatically within the brain, too.

For example, they may also be bad at calculating the trajectory a ball will take when trying to make a catch or understanding why a person is getting angry or frustrated.

Robison, however, worries that their theory may be based on a misreading of how people with autism describe their experiences. Just as Robison has misread social situations due to the impairments caused by his disorder, he thinks the scientists may have misread their information.

Carolyn Y. Johnson can be reached at cjohnson@globe.com. Follow her on Twitter @carolynyjohnson.