The Case of the Impartial Autistic

A study published in Proceedings of the National Academy of Sciences this week discovered some interesting findings about how people with high-functioning autism make decisions, but it says as much about the often problematic ways we interpret such findings as it does about autism.

Izuma and colleagues wanted to know how autistic people's decision making would be affected by being observed.  They tested this by:
asking people to make real money donations to UNICEF under two conditions: alone in a room or while being watched by a researcher.
This is a standard task for studying decision making in neurotypical people.  Researchers typically find that people donate more when someone else observes them, presumably in order to look like a good person.  Izuma's team wanted to know whether people with autism would show the same effect.

"What we found in control participants - people without autism - basically replicated prior work. People donated more when they were being watched by another person, presumably to improve their social reputation," explains Keise Izuma, a postdoctoral scholar at Caltech and first author on the study. "By contrast, participants with autism gave the same amount of money regardless of whether they were being watched or not. The effect was extremely clear." 

To make sure that autistic participants weren't simply ignoring the fact that there was another person, they used a control task that didn't involve moral reasoning, or any sort of decision-making at all.  Both the autistic and the control participants had to solve math problems, with or without an observer present.  This time autistic participants showed the same reaction to an observer as controls: both did better on the math problems when watched.

"This check was important," says Ralph Adolphs, Bren Professor of Psychology and Neuroscience and professor of biology at Caltech and the principal investigator on the paper, "because it showed us that in people with autism, the presence of another person is indeed registered, and can have general arousal effects."

The research team argues that being observed does not influence autistic people's decision making because they do not consider what others think of them at all.  Says Adolphs:

"what is missing is the specific step of thinking about what another person thinks about us. This is something most of us do all the time - sometimes obsessively so - but seems to be completely lacking in individuals with autism." 

This may be an intuitively plausible explanation, but if we look at the study a little more closely, we can find other possible explanations that these researchers do not seem to have considered.

First of all, note the strangeness of the decision making task--I'm not sure why it was even chosen in the first place.  While Izumi's team talks about this task as if it were a generic decision-making task like any other, moral decision-making is its own domain that involves its own complex set of concepts.  Furthermore, as one blogger has already pointed out, it is the autistic participants who make the normatively correct response, from a moral perspective.  We generally view giving to help someone else, or to act morally for its own sake, as better than giving to make oneself look good, and our holy figures, such as Jesus Christ, tend to exemplify impartiality.  It seems odd, then, to take a deficit perspective on the autistic participants' choices.

Some other possibilities:
  • What if autistic participants care what other people think of them, but also have a moral code dictating impartial behavior, and it is the latter that determines their moral decision-making?  

In this case, one should find that in a decision-making situation without moral content--say, an economic one--that autistic participants show the same pattern as controls.

One could also give them descriptions of hypothetical moral decisions, ask them what the correct decision would be, and then ask them to explain why they chose this answer.  If autistic participants rely on a moral code, this should be reflected in responses like "it's just the right thing to do" or "because we should follow x principle."  Alternatively, one could use a multiple choice approach, where the answers are code-based, utilitarian ("because this decision has a good effect"), or social status-based.

  • If anything, the fact that an autistic person often gets negative reactions from people could lead them to be more concerned with others' reactions rather than less.  That does not necessarily mean they can understand these reactions.  What if autistic participants care what other people think of them, but because of their difficulty reading other people and predicting how they will react, it does not influence their behavior in a systematic way?  In other words, it functions as a source of anxiety about an amorphous threat rather than as a guiding force towards appropriate behavior?

In this case, control tasks would need to be designed to isolate caring about others' reactions (maybe using some sort of autonomic measure of reactivity or approach/avoidance) from understanding such reactions. 

The study is currently behind a paywall, so I'm missing a lot of information that would help in interpreting the results.  For example, how generous are autistic participants, relative to controls?  If they follow a moral code, they would make high but unchanging donations; otherwise, they might make typical or low donations.  Unusually low donations might indicate a lack of focus on others' needs. 

I also don't know about relevant personality or motivational factors in either autistic or control participants.  I'm not very familiar with personality research, but people do differ in how much they seek others' approval.  If there are measures of this, it might be possible to tell more directly whether autistic people really do seek others' approval less.

Ultimately, what troubles me is that these researchers have taken up a theory of mind based deficit model so reflexively that they do not seem to consider other possible explanations for their results, or address the fact that the autistic participants actually make the more moral decisions.  This is a common trend in autism research that I think would be less so if researchers interacted more with the autism community.  Most professors oversee and write up their research, but do not actually interact with their participants.  In such a situation, it becomes all too easy to see autistic people as a collection of deficiencies rather than as human beings with strengths as well as weaknesses.


  1. Emily,

    I love this post. It's spot on. Would you give me permission to repost it on the Autism and Empathy site (www.autismandempathy.com)? If so, please send me an email at rachel (at) autismandempathy (dot) com.


  2. Idea for a follow-up study: if you primed some participants with autism to think about how others would judge their decision before they pledged money, how would that affect their decision? If people with autism don't typically think about others' reactions, as Izumi suggests, than explicitly telling them to consider this should make them behave more like neurotypicals. If they do consider others' reactions but don't see it as an appropriate basis for moral decision-making, it may not have this effect.