Autism and The Machine: Better Communication Through Circuitry

Young adults do arts in craft treatment
Dr. Drew Rubin

You don’t have to be autistic to know that communicating effectively is difficult. Human beings thrive on feeling understood. Without the ability to communicate clearly, life would often feel quite lonely for many people. We are social creatures. And yet, many people struggle to get their feelings across through conversation. For most autistics, this is a simple fact of life.

The definition of communication is: the imparting or exchanging of information. That exchange must necessarily involve 50% transmission and 50% reception. In the case of verbal communication, which is how most communication happens, that means one person talking and another person understanding. There is verifiable evidence that demonstrates that many people experience sensations of joy upon hearing the human voice. This is usually not the case for autistic children, according to a recent Stanford study. The neurotransmitter dopamine is often known as the pleasure hormone in the brain. Dopamine is triggered by the sound of the human voice in most people. For individuals with autism, there was a significant decrease in the amount of dopamine transmitted. The lower the dopamine measured, using functional magnetic resonance imagery (fMRI) brain scans, the more likely an autistic individual was to have significant trouble communicating.

So, are autistics simply not interested in communicating verbally? Of course not. Many people with autism just might not respond the same way to the same cues that many other people do. It’s important to be mindful of the fact that autism can cause certain behaviors that might be difficult to understand. Extreme patience is a virtue that parents of autistic children must learn.

But does it have to be that way?

Could a Machine Really Help All People, Not Just Autistics, Learn How To Converse Better?

Improving communication is one of the main goals for many of the people working with autistic individuals. M.I.T. scientists have developed a piece of software that may hold the key to helping improve communication skills, not just for autistic people, but for anyone who has difficulty with social interactions.

“My Automated Conversation CoacH”, or MACH for short, is a software program that was made to provide conversational coaching sessions with individuals. The software appears as either a man or a woman on the screen, and you are meant to look directly into his or her eyes and have a normal conversation. The software uses input from a webcam, scanning facial expressions, carefully listening to speech patterns and registering indicators from other physical behavior. The face on the screen responds precisely the way that a patient listener would as you speak. The avatar nods and mirrors your own feelings. If you smile, the software produces a smile on the avatar’s face. In almost every way, the software is a perfect conversational partner. The software is programmed to do what too few people know how to do: to be a good listener.

In addition to listening to and interpreting the content of your speech, the software also keeps track of all of the nonverbal utterances that make of much of speech: the umms, ahs, ohs and hmms. It uses this data to provide feedback about the type and quality of your conversational abilities. The software can register the intensity of your smile and determine the mood and tenor of the conversation, in addition to tracking other movements and gestures that articulate nonverbally. It combines all of this data to provide a picture of how well you have communicated. Efficacy of communication is not subjective, like say, the depth of communication is. Whether or not someone is communicating in a way that they can be intrinsically understood between two relative equals can be assessed using a thoughtful array of criteria. 

Because MACH is so effective at maintaining the appearance of a realistic conversational connection, it is an ideal environment in which to practice social interactions.

MACH was developed to respond to social hardships expressed by individuals attending a workshop at the Asperger’s Association of New England. But it was tested on undergraduate students participating in the development of the MACH program, or preparing for other job interviews. It’s a convenient fact that the highly technical arena in which much high level technological work takes place can mirror, in a small way, the difficulty of social interactions in the broader world of individuals with autism. Highly technical individuals often have difficulty in ‘normal’ social settings.

One of the appeals of the MACH system is that people can sometimes tend to feel judged by a human observer of social interactions. The fact that it’s a piece of software that is making totally objective, rigorous, neutral assessments of the interactions made them more valid, and importantly, less likely to offend. People with sensitivities about the judgement of others could look at their assessments and likely identify where they had gone wrong. Watching a video playback of their conversation, side-by-side with a chart indicating each of their communicative behaviors gave them an irrefutable indication of their communicative efficacy.

When this software is ready, it will provide an invaluable tool to help therapists, parents, schools and others to prepare autistic individuals with the communication skills necessary to thrive in a conventional school or workplace setting.