This post is by Autism Speaks’ staffer, Leanne Chukoskie, Ph.D., Asst. Director for Science Communication and Special Projects.
Imagine having your child’s favorite therapist available to you at any time you need. With her uncanny way of drawing out your child, together they practice speech, social interactions and some physical gross and fine motor control drills to improve his coordination skills. Now imagine that all this doesn’t cost millions and you don’t even need to build on to your house for the therapist’s new abode. All she needs is a dose of 110V every few days to keep going.
Such fantastic scenarios aren’t necessary the realm of science fiction any longer. Social robots are making appearances in classrooms and even serving as teacher’s aides. An article in this week’s New York Times makes the case for the unique advantages that these machines bring to the learning environment. For example, RUBI, a robot developed at UCSD, has been deployed in preschools where it has been teaching children Finnish words. RUBI modifies her interactions on the basis of the children’s approach and retreat behaviors as well as a real-time analysis of children’s facial expressions. Is this child looking frustrated or engaged and content? Using this interactive technology, RUBI and robots like her can individualize interactions to optimize the engagement with each child.
It is precisely this ability to optimize that makes robotics so useful for working with individuals with autism. Katharina Boser, Ph.D. the co-chair of Autism Speaks’ Interactive Technology for Autism Initiative (ITA) underscores the utility of programmed real-time interactions for autism. People are unpredictable. For an individual with autism, this aspect of behavior can make interactions with people less desirable than interactions with a robot that matches expectations frequently. With the phenomenal advances in computer technology, a robotic social partner can work on lags in, for example, gestural communication by keeping to a very predictable set of gestures, while adding a bit more variety with speech and intonation if the learner is ready. Other ITA research underscores the point that the social partner does not need to be human-like. Researchers noticed that some young children with autism really engaged with Crush, the sea turtle made famous in Finding Nemo. and have built a therapeutic strategy around this character.
In addition to the robots, other sorts of technology are offering therapy support for autism. By integrating eye-tracking into a game interface, Felicia Hurewitz, Ph.D. (Drexel University) has developed an environment for children that requires gaze interaction with the characters to reveal clues to advance in the game. Nilanjan Sarkar, Ph.D. (Vanderbilt University) is using feedback from an individual’s autonomic responses (such as heart rate, sweating) to find the right level of difficulty for a game—not too frustrating and not too easy. The computer interface and simulated human interactions are put to work in service of practicing social skills in a safe environment from SIMmersion, a technology company in Maryland. The ability to analyze, review and evaluate interactions is a great strength of this technology.
These great new tools will undoubtedly change the landscape of autism therapy, but they will never replace a human therapist. Our models of the behaviors being simulated in silico are human. We also need humans to serve as final arbiters of the therapy—did it really work? Are the data being collected valid in other settings? After all, the point of this training is to interact with other people. As of this writing, we still need humans for that!