Member preview

Meet the Robots Changing the Face of Autism Therapy

Kaspar the Robot. Photo: University of Hertfordshire

In Part 1 of a four-part series, we explore disability in the digital age: how tech is offering new solutions, and new questions, for the future of human interaction.

In a recorded therapy session, humanoid robot Kaspar blankly bats his large, round eyes at a young autistic boy. The boy timidly tickles Kaspar’s feet, then flashes an ecstatic grin when the robot says, “This is nice. It tickles me.”

Ben Robins, a Kaspar researcher at Hertfordshire University in southern England, hopes that such modest breakthroughs could help children on the autism spectrum learn to socialize.

“Every autistic child is so different, but most open up to robots,” says Robins. “Kaspar is predictable, meaning there are no surprises, so children feel that he creates a reliable environment.”

In clinical trials involving nearly 200 children over the course of more than a decade, Robins has seen Kaspar the Robot encourage autistic schoolchildren to eat at snack time, get comfortable with hugging, and, most important, find a safe space for practicing communication with peers — a daunting task for many on the spectrum.

With Kaspar’s slightly metallic voice and baseball hat fitted atop a shock of dark hair, parents of those in the clinical studies have likened him to Chucky, the serial-killer doll from the Child’s Play horror movies. But the “uncanny valley” effect — the idea expounded upon by Sigmund Freud that human replicas inspire unease and revulsion — doesn’t resonate for most autistic children.

Graph: University of Michigan

It is the human face, with its vast range of nuance and complexity, that can be stressful for autistic children. They often experience acute anxiety alongside remarkable sensitivities to detail and can be overwhelmed to the point of paralysis when faced with human interaction. Alternatively, a robot’s predictability, lack of emotional judgement, and bottomless patience can provide autistic kids with relief from the constant pressure to socialize and understand emotional cues.

Kaspar is a veteran of what is now a sizable cadre of friendly assistive robots for autistic children, spurred by a surge in research as the world counts a growing number of autism cases every year.

The autism uptick has been particularly severe in developed countries over the past decade and a half. Today, more than 70 million people worldwide — one percent of the global population — live with autism, according to the Centers for Disease Control and Prevention (CDC). In the United States, the CDC estimates that one in 68 children have autism, with a gender ratio of about five boys for every girl.

The cause, however, remains a mystery. Experts say higher numbers could be linked to higher levels of awareness, better diagnoses, or genetic or environmental biological factors.

Growing awareness has meant more inclusion in schools, and many robotics clubs today host autism departments. But in the face of progress, many experts worry the trend in computer-assisted aids for autistic people is being led by tech companies that are faster and more agile but also less concerned with regulations than clinical researchers.

Robins says that in racing to create a fully autonomous assistive robot, companies could violate Isaac Asimov’s first law of robotics: The robot must not injure the patient.

“The users are more important than the tech; the drive shouldn’t come from the tech,” says Robins, who adds that robots should only complement, never replace, professionals.

Even when a therapist is involved, the robot-human interaction requires delicate supervision so as to rigorously preserve the children’s welfare. Robots collect data on autistic children, who can be honest and literal-minded to a fault. An autistic child might not have the same qualms about sharing personal details of their lives and that of their families, all of which could provide a goldmine of data at a time when companies like Facebook and Twitter are desperately attempting to breach and monetize privacy.

A therapist is also needed to guide the child with an awareness of robotic ethics. For example, if the child learns to view the robot as a kind of slave, he could then transfer that model to human-human relationships. If hit, Kaspar is programmed to turn away and say “ouch.” But Kaspar doesn’t always sense a lighter slap, at which point the teacher needs to intervene and press the button to instigate the “ouch” response.

Laurent Mottron, a psychiatrist at the University of Montreal who has been working with children and colleagues on the autism spectrum for decades, says that all ethical problems are compounded by robots’ prohibitive pricing, at thousands of dollars per machine, which could be diverting precious research and funding resources.

“We have a saying in the Montreal group that anything that costs money for an autistic child is useless,” says Mottron. He adds that other alternative therapies have been shown to help autistic children improve their “joint attention,” the ability to share information and interests with others.

Mottron believes that, on a deeper level, many robotics companies are misguided because they reinforce neuronormative social stigmas about autism. He has long argued against the traditional belief that autistic people suffer from intellectual disabilities. It’s society, not the child, he says, that should learn to accept autistic children’s unique characteristics, in the same way that society has overcome the idea of homosexuality as a disease that needs “fixing.”

Autistic children may perform poorly on standardized tests or “are reluctant to be taught anything at a young age, but the autistic mind is oriented, and can flourish, when dealing with a kind of subsect of information that is more reliable, more structured where things are rigidly implicated,” says Mottron.

Studies show that the autistic cognitive inclinations toward patterns and puzzles may be linked to scientific ability. In 2011, Cambridge University researchers found that childhood autism was two to four times more prevalent in the Dutch technology hub of Eindhoven as in other Dutch cities of similar size. Other studies have found similar trends in Silicon Valley.

“I think that the role of the autistic in robotics could be in making robots themselves,” says Mottron, who says that autistic scientists are currently working alongside their neurotypical counterparts at MIT and in many of the world’s most prestigious robotics programs.

Japanese android. Photo:

At the MIT Media Lab, Ognjen Rudovic has been developing assistive robots equipped with AI for autistic kids to help realize Mottron’s idea that autistic people should be appreciated, rather than marginalized, for their cognitive differences.

Rudovic envisions a robot that will not only be controlled remotely but will also have cameras, microphones, and a small sensor watch to allow therapists and peers to understand the experiences that autistic children may be unable to express. The hope is that one day the “objective” medium would enable an autistic and a neurotypical child to meet spontaneously, using the robot to translate one another’s words and actions.

“Robots will inevitably alter the way we communicate today, and I am optimistic enough to say that it will happen for the better,” says Rudovic.

That future is still years away, as clinical studies remain small in scale. Rudovic has studied some 37 children, ranging in age from two to 14, but hopes that more families will be open to advanced robotic therapy as the fields of robotics and artificial intelligence gain momentum.

“We should keep in mind that human emotions are very perplexing and everyone is experiencing them in a different way, so expecting the robots to be able to read and access to our inner emotions is a far-fetched goal.”

A better goal, says Rudovic, could be a robot that “objectively” monitors interactions and “intervenes” in gaps in the conversation to help bridge the autistic-neurotypical divide, among others.

“Intervention should be discrete enough to let us know that maybe there is another way of saying or doing something that would be more pleasing for the other person,” says Rudovic. “But I think that doesn’t apply only to communication between autistic and neurotypical people, but all people, no matter the condition.”

One clap, two clap, three clap, forty?

By clapping more or less, you can signal to us which stories really stand out.