Tuesday, October 12, 2010

Can a magician trick people with autism?


Are people with autism susceptible to magical illusions? There are a number of reasons to suspect that they might not be.

Firstly, magicians rely on misdirection. They'll use eye gaze and gesture to make sure the audience is looking one way, while they're secretly switching the cards or sneaking an elephant into a hat (or whatever it is they do). People with autism, it's argued, are less sensitive to these kinds of social cues, so perhaps they're not as easily misdirected.

Second, as mentioned in a previous post, there are a number of studies suggesting that people with autism are less susceptible to visual illusions. They see what's there in front of them rather than what they're supposed to see. Another reason to think that the magician might not be able to trick them as easily as a non-autistic person.

Gustav Kuhn and colleagues set off to test this prediction by getting participants to watch a short video clip of the vanishing ball illusion. In the clip, a magician threw a ball up in the air twice. He then pretended to throw the ball a third time but actually kept hold of it. Non-autistic people tend to report that they have actually seen the ball thrown a third time before disappearing into thin air. The illusion is particularly strong if the magician looks to where the ball is expected to be.

Keep your eye on the ball...

Surprisingly, Kuhn et al. found that adults with autism were, if anything, more susceptible to the illusion than non-autistic adults. They would report having last seen the ball at a much higher location. And when asked what had happened to the ball, they were more likely to give an explanation that involved the ball leaving the hand.

Kuhn et al also report some eyetracking data, which provides some clues. It turns out that the participants with autism spent less time fixating on the ball. They also tended not to follow the ball to the top of its trajectory, which fits in well with other studies showing problems on smooth pursuit tasks, in which participants have to follow the trajectory of a moving object.

The authors conclude that the adults with autism were misdirected by the magician's social cues, challenging the idea that they have social attention difficulties. This is quite a bold conclusion to reach based on a single 6-second video clip, particularly as there was no comparison between clips with and without social cues. And the eye-tracking data seems to suggest an alternative explanation - that the autistic adults were tricked because they simply were not following the ball closely enough.

Nevertheless, the study highlights the potential of using magic tricks as a means of investigating the social and non-social attention skills of people with autism. It would certainly be interesting to see how they fare on other magical illusions. At this stage, I'm not making any predictions.


Kuhn G, Kourkoulou A, & Leekam SR (2010). How Magic Changes Our Expectations About Autism. Psychological science : a journal of the American Psychological Society / APS PMID: 20855904


  1. In this study the questions, at least as reported, are too open to different interpretations. E.g., did they mean "where did you last see the ball in the air"?--that was my first thought.

    My wild guess is that with unclear and/or open-ended questions, autistics would regardless (good-naturedly...) try to answer, but would also consider more possibilities than nonautistics.

    So: try better questions, for one. This reminds me of Laura Klinger's prototype formation paper, where the question was unclear (and very hard for me to figure out, even after reading the paper several times).

  2. Thanks Michelle. You may well have a point there. On a related note, I worry a bit that the authors are assuming that if the person last 'saw' the ball in mid-air, they are referring to the illusory trajectory. But perhaps the people with autism didn't track it back into the magician's hand, and so may be reporting where they genuinely did see it on the second throw.

    More generally, I totally agree about the dangers of using questions or instructions that are open to different interpretations. The Klinger prototypes paper is a good example. From what I recall, the 'prototype-learning' task didn't actually require the person to attend to any of the features that defined the prototype. Cathy Molesworth addressed some of these concerns in a follow-up paper and failed to replicate Klinger's findings.

  3. You have touched on a critical point - the implied context behind instructions. I think experimenters are too ready to say "oh sorry, our fault, we should be more precise. Questions are always open to different interpretations. The wonder is that NTs are not usually bothered by this and even surprised that there are other interpretations than the ones they intended. Perhaps autistic individuals are not tuning out the unintended interpretations. How do NTs tune them out? I expect their brains do some computations to zoom out from the question itself to take in the implied context behind the instruction. Remember the anecdote about the bus taking autistic children to the swimming pool. The teacher had to say: we are going swimming...and we are coming back. NT children would think such a remark not so much disambiguating as weird.

    As to autistic children being taken in by the subtle social cues of the magician just as much as NT children, I admit I would not have expected this. But then, the social impairment in autism is NOT global. I believe we will discover more and more social assets. One of these was revealed in our Social Stereotype study in Current Biology, 2007.

  4. Hi Uta. Thanks for the comment - I'm honoured!

    While I agree that it's fascinating how people with autism often 'misinterpret' instructions or statements, I do think it's a real challenge for us as researchers. If different participants are understanding the instructions as meaning different things then you've lost experimental control. If Michelle's correct (and I'm not saying she is!) then it's difficult to know how to interpret these findings.

  5. The question then is precisely: Why do people with autism often misinterpret instructions or statements - when other people don't?

  6. So I think there are two separate points here:

    1. Uta's question: Why do people with autism often misinterpret instructions or statements - when other people don't?

    2. Researchers need to try and reduce the scope for misinterpretation of instructions (unless they are specifically aiming to address Uta's question).

    Re Uta's question, and since we're dealing in anecdotes, I particularly liked Pat Howlin's story of the autistic boy who wouldn't go on an escalator because there was a sign saying "Dogs must be carried" and he didn't have a dog to carry. In one sense, he had parsed it appropriately (think "Seatbelts must be worn"). But on the other hand, he seemed to discount various clues suggesting that he had just misinterpreted the sign: the fact that making everyone carry a dog would be a very strange rule to force people to abide by; the fact that most people didn't have dogs and were still happily getting on the escalator; the fact that none of them were being arrested. Even us neurotypicals get these kinds of ambiguous statements wrong a lot of the time, but we're usually pretty quick to correct ourselves. My hunch is that difficulties in this kind of correction process might contribute to problems understanding ambiguous language. But of course there are many other alternatives!

  7. Lovely anecdote and well worth pondering about.

    I just wanted to say that your Autism Research Blog is terrific. Much of it was new to me and I have already learned a lot. The "of interest" column certainly is of interest. And helpful too.

    Thanks, Uta

  8. Thanks Uta. You've made my week - and it's only Monday morning!

  9. Press release and video here:


  10. I do wonder whether part of this is due to how well autistic people understand deception. As a (mostly) neurotypical person, when I watch a magic trick I am very aware that the magician is trying to deceive me, and that that deception will involve some kind of misdirection.

    If the researchers told participants that they were going to see a video of a magic trick, that might prime the NT participants to be on alert for misdirection while having less of an affect on the autistic participants. Even if the subjects weren't told in advance, the NT participants might be better able to recognize when and how the misdirection happened, and re-evaluate their memories accordingly. They might also be more reluctant to admit it if the trick fooled them- maybe autistic people are just less embarrassed at having been taken in by the magician, or less likely to lie when embarrassed.

  11. This may be of interest, theory of mind use in magic/illusion.