Skip to main content

Robots: the Answer for Treating Children with Autism Spectrum Disorder?

By Guest Contributor Irina Lucaciu, Emory University  

A smile appears on Jack’s face as the robot he is playing with congratulates him for accomplishing a task. Aiden seems captivated by the moving arms of Nao, a robot that has become his new playmate. Thousands of miles away, in London, a copy of Nao sits in the middle of a circle of five boys no more than 10 years old, encouraging them to imitate his movements, touch his hands, and try to identify the feelings he is describing.


When asked how the robot makes him feel and why, one of the boys replies that he is happy because the robot feels happy too.

However, Nao and the other robots are not simply toys, and neither are Jack, Aiden, and the five British boys simply children at play. They have autism spectrum disorder, and Nao is acting as a treatment tool for improving their life experience and helping them develop socially-relevant skills. Above are described the results of robot-assisted therapy [1, 3, 9, 13].

What is autism spectrum disorder, and why would the use of robots benefit those who have it?

Social interaction is embedded within human nature. We engage in it continuously, and, when deprived, we seek it to alleviate an immediate absence that provokes pain and anxiety. Regardless of whether it manifests in the form of dialogue, gestural interchange, or touching, social interaction is crucial for maintaining emotional and psychological well-being. The prevalence of communication is mirrored by its significance in human societies, and it justifies the persistent demand for socially-adaptable individuals as members of societal structures. [10]

However, individuals with autism spectrum disorder (ASD) have significant challenges with performing social functions, which makes it harder for them to be self-sufficient, follow a traditional education, or form families. Their difficulties primarily rest on an inability to relate to other people, understand the social information conveyed by gesticulations and facial expressions, and reciprocate touch or conversation initiated by an interlocutor [6].

Researchers are focusing on different strategies to encourage the development of social functions in autistic individuals. Because early intervention has been shown to be particularly effective in helping people with ASD develop skills for living autonomous lives as adults, children are currently the age group most targeted by autism researchers [10].

At present, the main therapies targeting children with ASD are applied behavioral analysis and non-human facilitation. With applied behavioral analysis, trained specialists interact with the autistic child, encouraging reciprocation, attention, and following of directions, and discouraging anti-social behaviors. Non-human facilitation can involve partnering with a trained pet that would subsequently ease the challenge of interacting with another human, engaging with an interactive computer program, or undergoing virtual training that offers gradual exposure to common social environments, such as a classroom or a playground. [6, 10]

Both human-facilitated and non-human facilitated therapies have helped children with ASD to develop better social skills over time. However, problems remain. For example, pet-assisted therapies include a risk of being bitten or hurt during the therapy session, and some children are allergic to the pets. Furthermore, emotional attachment to a pet can lead to a traumatic experience of the death or loss of the animal [8]. In a recent study done by Dr. Sandra Y. Okita at Columbia University, the risks of pet-facilitated therapy were removed with the introduction of Paro, a robotic seal and class II medical device. Equipped with various sensors that detect touch, recognize speech, and generate behavior based on the information collected, Paro was previously recognized for its therapeutic effects in the treatment of children and elderly with mental and cognitive disorders. In Dr. Okita’s study, at interest were Paro’s qualities as a social facilitator between parents and their children. Would the robotic seal increase social interaction and modelling of parental behavior, thereby reducing the pain of the patients or helping them cope better with the treatment? The results showed that the children who interacted with Paro together with their parents had lower levels of pain than children who interacted with the robot alone. In turn, parents proved a heightened ability to empathize with their children and react in ways that comforted them [8]. In the light of many studies showing that children often model parental behavior in interpreting social stimuli and forming a reaction, Paro seems to have succeeded in improving this transfer of information.


Considering the previous work using robots to both better understand human social behavior and to improve human experience in general, it is no surprise that the recent technological advancements led to the introduction of robots in a variety of fields, including autism research and treatment. But how can robots such as Paro facilitate the therapy of children with ASD? The answer lies in the many studies using social robots- or robots meant to stimulate social behaviors and increase the perception of social cues- which proved that children with ASD tend to engage actively with social robots [7]. In the following videos, both Jack and the five boys from Britain prove the above stated.   

The results of these studies are particularly uplifting considering that autistic children seem to show the more enthusiasm and interest for interacting with a person after they have interacted with a robot. A study by Elizabeth Kim and her collaborators at Yale University showed that children with ASD who interacted with a robot spoke more with an adult partner than children who interacted with a person alone [7]. In other words, it appears that interaction with a robot facilitates social interaction with an adult, at least in the short-term.

Another study showed that when a therapy session is led by a human facilitator, children with ASD spent significantly less time watching the instructor than children without the condition. However, when the session was led by a robot, both groups spent approximately the same amount of time directing their eye gaze towards the robot [9]. What is it that social robots can offer and human facilitators cannot? Dr. Kim attributes the amplified interest shown by children with ASD towards robots to a greater curiosity in them, as well as to the design of the protocol [7]. Dr. Brian Scassellati, Henny Admoni, and Dr. Maka Matarić suggested that perhaps the simplified appearance and behavior of the robot in comparison to that of a person helps with the overstimulation problem specific to many people with ASD [4, 10, 12]. Furthermore, robots might not be perceived as judgmental, helping the children to be more open to interacting with them [12].

However, no long-term studies on the benefits of robot-assisted therapy have been conducted yet. Despite the benefits that robots seem to bring, many questions remain about the ethical implications of introducing robots as large-scale therapy aids.

  • What should the robot look like? At present, the appearance of robots used in autism varies widely [10]. Nao, for instance, looks like the stereotypical machine-like robot. Keepon has the appearance of a toy duck, and it was even introduced as merchandise at Toys ‘R’ Us in 2011 [14]. Paro, on the other hand, is clearly more complex, and its appearance resembles closely that of a real seal. Kaspar shares many features with a human child, but it is still clearly a man-made object. However, FACE looks incredibly human-like, and has complex mechanisms that allow her to express human emotions with very accurate facial expressions.
FACE, image from   


Keepon, image from
Kaspar, image from



Considering that children with ASD are easily overstimulated [10], would it be wiser to introduce them to robots with a simplified appearance such as Keepon or Nao, rather than a robot like FACE? In that case, how are they to interpret such an interaction, and transfer what they learned from it to the interaction with a human? All these robots are programmed to reproduce the characteristics of either humans or animals. Nao, for instance, uses language. His speaking abilities, along with many other human-like behaviors he shows suggest to the child that Nao’s actions are governed by a brain very similar to that of a human. But Nao does not have a brain of its own, and despite its multiple sensors that make him able to adapt its behaviors according to the responses of a child, he does not possess the human ability to produce behavior and speech that are sensitive to context. Even robots such as Kismet, which received a social “brain” based upon the four modules identified by psychiatrist Simon Baron-Cohen as necessary for social interaction- Intentionality Detector, Eye Direction Detector, Shared Attention Mechanism, and Theory of Mind Mechanism [11]- can only simulate simplified human behaviors, such as following people with its gaze and seeking human company. The only way to increase a robot’s range of behaviors and make its reactions more particular to the emotions and responses of the child is to increase the amount of information that the robot receives from its interaction partner. In turn, that would require more physiological detectors, which could range from galvanic skin response detectors and pulse monitors to brain activity detectors. Such recording devices could become bothersome and invasive, and could have a negative effect on therapy. [10]

  • Take the case of FACE. Her appearance is, indeed, very human-like. However, modern-day technology has not been able to give her the ability to express more than six basic human emotions. Facial expressions are possible on account of sensitive nervous control of multiple muscles of the face. As expected, such control is difficult to reproduce using modern-day technology [10].
  • What if a malfunction occurs in the mechanisms of the robot during its interaction with the child [10]? It was discussed above that the death of a pet providing animal-assisted therapy could be traumatic for the child. However, death is at least embedded within human experience, and it is something that the child with ASD will have to face as a functional member of society. Robots cannot die, but they can be damaged, because they are currently not designed to resist bumps, spills, and falls.
  • Last but not least, what role should the robot take around the child? Should it be a friend, instructor, parental figure, or simply a toy [10]? Using a robot during therapy would not be as simple as using a pet. While a pet, although trained, remains essentially an animal with the characteristics of its own species, a robot borrows many essential human features, such as language and gesticulation. In the case of Nao, for example, the robot also looks drastically different from a human. The child holds the hard and cold hand of the robot, and then the warm and soft hand of a human, and yet, both the robot and the human play games, talk, and dance. How will the child react to such conflicting features?

A robot, no matter how advanced the technology invested in its creation is, does not think similarly to a human. Despite its many sensors, it cannot integrate information with the same efficiency, simply because its motors do not possess the level of connectivity that characterizes the human brain. While there are still so many things to be understood about our brain, it seems unlikely that a robot with the ability to mimic human emotions and behaviors could be created. Until neuroscience can decipher the secrets of the brain, the introduction of robots as a means to teach children with ASD about human social behaviors should be carefully thought-out. Only long-term projects investigating both the benefits and the downfalls of using robots for treating children with autism spectrum disorder could testify either in support or in opposition to robot-assisted therapy.


[1] Aldebaran Robotics. “Robots teach communication to children with autism.” Online video clip. YouTube. YouTube, 29 Apr. 2013. Web. 30 Apr. 2013. Retrieved from

[2] Associated Press. “Kaspar the Friendly Robot Helps Autistic Kids.” Online video clip. YouTube. YouTube, 8 Mar. 2011. Web. 30 Apr. 2013. Retrieved from

[3] Breen, Tom, and Bret Eckhardt. “How Robots Can Help Children with Autism Learn and Communicate.” UConn Today. 25 Apr. 2013. Web. 30 Apr. 2013. Retrieved from

[4] Foss-Feig, Jennifer H., Duje Tadin, Kimberly B. Schauder, and Carissa J. Cascio. “The Journal of Neuroscience.” A Substantial and Unexpected Enhancement of Motion Perception in Autism. Web. 5 May 2013. Retrieved from

[5] Goodrich, Michael A., Mark Colton, Martin Fujiki, Alan Atherton, Lee Robinson, Daniel Ricks, and Margaret H. Maxfield. “Incorporating a Robot into an Autism Therapy Team.” Incorporating a Robot into an Autism Therapy Team. IEEE Life Sciences, n.d. Web. 29 Apr. 2013. Retrieved from

[6] “How Is Autism Treated?” Autism Speaks. Web. 3 May 2013. Retrieved from

[7] Kim, Elizabeth S., Lauren D. Berkovits, Emily P. Bernier, Dan Leyzberg, Frederick Shic, Rhea Paul, and Brian Scassellati. “Social Robots as Embedded Reinforcers of Social Behaviors in Children with Autism.” National Center for Biotechnology Information. U.S. National Library of Medicine, 31 Oct. 2012. Web. 16 May 2013. Retrieved from

[8] Okita, Sandra Y. “Self-Other’s Perspective Taking: The Use of Therapeutic Robot Companions as Social Agents for Reducing Pain and Anxiety in Pediatric Pacients.” National Center for Biotechnology Information. U.S. National Library of Medicine, n.d. Web. 16 May 2013. Retrieved from

[9] Salisbury, David. “Humanoid Robot Helps Train Children with Autism | Research News @ Vanderbilt | Vanderbilt University.” Vanderbilt Research. N.p., 23 Mar. 2013. Web. 3 May 2013. Retrieved from

[10] Scassellati, Brian, Henny Admoni, and Maja Matarić. “Robots for Use in Autism Research.” – Annual Review of Biomedical Engineering, 14(1):275. N.p., 9 May 2012. Web. 16 May 2013. Retrieved from

[11] Smith, Jeremy Adam. “Can Robots Feel Your Pain?” Greater Good. University of California, Berkeley, Summer 2007. Web. 16 May 2013. Retrieved from

[12] “Talking Robots Play Part in Therapeutic Treatment for People with Special Needs.” PBS. 9 Jan. 2013. Web. 5 May 2013. Retrieved from

[13] UConn. “A Story of Robots and Autism.” Online video clip. YouTube. YouTube, 25 Apr. 2013. Web. 30 Apr. 2013. Retrieved from

[14] Vance, Ashley. “Toys ‘R’ Us Wants a Robot to Sell for Christmas. Bloomberg Business Week Magazine. (2011). Retrieved from

Want to cite this post?

Lucaciu, IM. (2103). Robots: the Answer for Treating Children with Autism Spectrum Disorder? The Neuroethics Blog. Retrieved on
, from


  1. This comment has been removed by the author.


  2. Compared with other technologies, robotics is considered as a prestigious one among people. Robotics training is one of the important technology taken by the freshers mostly.
    Robotics for Kids


Post a Comment

Emory Neuroethics on Facebook