Whose Planet Is It Anyway?

Tuesday, January 30, 2007

Deconstructing Sally-Anne

Some unfounded myths about autism just don't want to die. The most notorious is the urban legend about children becoming autistic after vaccinations, which recently was described by Arthur Allen as a story with legs. Another myth that needs a wooden stake through its heart is the claim that autistics lack theory of mind, or the ability to understand that others have separate thoughts.

The theory of mind idea was a conclusion drawn from the Sally-Anne test, an experiment performed by Simon Baron-Cohen and other psychological researchers. The study included several groups of children, some of whom were autistic children. The researchers performed a skit that went like this:

Sally puts a marble in a basket and leaves the room. While she is gone, Anne removes the marble from the basket and puts it in a box. Sally comes back into the room, and the child is asked, "Where will Sally look for her marble?"

The response of most of the autistic children was to point to or name the box, while most of the non-autistic children identified the basket. The researchers concluded that the autistic children who chose the box were incapable of understanding that Sally did not know the marble was in the box.

Of course, this is not the only possible conclusion that could be drawn from the data, and it's not even the most logical interpretation. Because autistic children have significant differences in speech processing, it seems far more likely to me that they simply misunderstood the question.

The word "look" has several distinct alternative meanings in the English language. It can mean viewing a person or object, such as in the phrase "look at." It also can mean searching for a person or object outside one's field of view, when used in the phrase "look for." Distinguishing between these two meanings requires a child to have a fairly complex understanding of prepositions and idiomatic speech. (Edit: Dr. Morton Ann Gernsbacher has noted that the questions asked during such tests "are some of the most complex syntactic structures in the English language.")

Many autistic children learn the word "look" in the context of behavioral therapy with "look at me" prompts. Such children may not be aware that the word has an alternative meaning of searching for an item that is out of sight. They will interpret the question as "Where will Sally see her marble?"

Also, autistics are more likely to be visual-spatial thinkers, whereas the majority of the population consists of auditory-sequential thinkers. A visual-spatial thinker, upon hearing the question "Where will Sally look for her marble?" will translate that question into mental images of Sally, the marble, and the location of the marble. An auditory-sequential thinker, on the other hand, will focus on the action of looking as a sequential process: Sally first has to look for the marble before she can find it. There is an implied "first" in the question—where is the first place that Sally will look for the marble—but a child who processes language visually may not understand that the question has a sequential component. Instead, such a child may interpret the question as "Where will Sally have to look to find her marble?"

There are so many possible ways that an autistic child could misinterpret a question like that, it's frankly astounding to me that any researchers would think they had a basis for drawing any conclusions whatsoever from such a study.

Here's my belated Christmas gift to Simon Baron-Cohen: a few garlic cloves to put in Anne's box to keep the vampires away when it's buried.



  • What do testosterone levels have to do with the box-basket conundrum?

    People who design tests like this are either morons or are looking for data to support their belief.

    By Anonymous Testy Osterone, at 4:00 PM  

  • Well blame it all on Descartes, he started it.

    As for Testy Osterone I am glad you had the balls to say that, it is what I have been saying about some of this science for some time.

    By Blogger laurentius rex, at 6:11 PM  

  • Don't even get me started on the theory of mind, my personal pet hate. It seemed so obviously 'true' when my children were first diagnosed, but I knew from 'experience' that it wasn't really, couldn't really be true.
    As you point out so much of it has to do with language comprehension, even if speech happens to be preferred method of communication, which it often isn't.
    I could spout about his bug bear for pages, but suffice to say that my children demonstrate daily, frequently, how far more empathetic they are the average Joe Blow on the street.

    By Blogger mcewen, at 6:11 PM  

  • The other day I came across a description of the Sally-Ann test. I had to read it a couple times to figure out what in hell the testers were attempting to do. I'm sure the problem is one of language interpretation and comprehension.

    By Blogger Joseph, at 6:32 PM  

  • Of course that test has been much criticised in the literature too and alternatives have been devised to get round the cognitive demands of understanding the instructions themselves.

    You don't think that these scientists give up on it that easily do you?

    By Blogger laurentius rex, at 9:01 PM  

  • Actually, to his credit, Simon has moved on from the ponderings of theory of mind deficits (which actually means difficulty in holding mental representations of one's own mental states and those of other people) to explore other avenues. The problem here isn't Simon himself (with regard to theory and testing of theory), but rather the uptake of the theory and its testing protocol from scientific endeavour into clinical practice by people who have failed to read and grasp what the idea was in the first place (a phenomenon somewhat akin to the issues currently being discussed on ballastexistenz's blog). When I spoke last to Simon about this issue, he was clear about the nature of the inquiry and its not being intended to inform clinical practice as a theory of aetiology... merely a pondering.

    By Blogger David N. Andrews MEd (Distinction), at 4:31 AM  

  • David: I'm sure you're right that the experiment was performed in the spirit of scientific inquiry and without malicious intent. The fact remains, however, that it gave rise to a very nasty stereotype and a huge amount of prejudice. At the very least, Simon Baron-Cohen ought to issue a press release in which he clearly states that the stereotype is wrong and apologizes to the autistic community for his role in creating it.

    By Blogger abfh, at 12:03 PM  

  • "I'm sure you're right that the experiment was performed in the spirit of scientific inquiry and without malicious intent. The fact remains, however, that it gave rise to a very nasty stereotype and a huge amount of prejudice. At the very least, Simon Baron-Cohen ought to issue a press release in which he clearly states that the stereotype is wrong and apologizes to the autistic community for his role in creating it."

    I think he did mention somewhere that it wasn't aimed at clinical work: sadly, the responsibility for what happens when clinicians get something remains with the clinicians. Thing is, regardless of what he has or hasn't said yet, it's the clinicians who misuse the notion. In my mind, they are the ones who owe the apology. It's a bit like trying to sue the maker of a gun that is used to kill someone: the maker is not responsible for the effect... the person using it is, and that's how I see things with clinical practitioners who misapply the results of research.

    I'd like to see, for example, apologies from Wakefield for his misapplication of theory that drove the whole MMR debacle.

    By Blogger David N. Andrews MEd (Distinction), at 1:50 PM  

  • David wrote:

    the maker is not responsible for the effect... the person using it is

    The way I see it, everyone has a moral responsibility to consider the effects of their work on society. I argued this point in more detail in my recent post Luddites, Biomarkers, and Baby Lize, in which I mentioned the German atomic scientists who fled their laboratories to prevent Hitler from getting the bomb.

    Those scientists easily could have concluded that it was OK if their research gave the Nazis the atomic bomb because they weren't responsible for what Hitler did. It was fortunate for the world that they thought otherwise.

    By Blogger abfh, at 3:09 PM  

  • "The way I see it, everyone has a moral responsibility to consider the effects of their work on society."

    From what I gather, Simon does try to consider things (albeit from often compromised positions... CAN and NAAR sem to crop up in the ARC's research funding lists). The ones that concern me are those who - without thinking of the consequences - think only of their own pockets. Simon Baron-Cohen is a lot different from Andrew Wakefield. And from Andy Cutler. And from the Geiers. As I see it, they are the ones not thinking of the effects of their 'research' on society as a whole. At least Simon does try to.

    Sure, he may be in need of feedback. We all need that. But I don't see him as the sort of threat that the likes of Wakefield, Geier & Geier, and Cutler are just now.

    By Blogger David N. Andrews MEd (Distinction), at 12:39 AM  

  • Ironically all the experiments say is that autistic people develop theory of mind at a later stage in childhood than neurotypical children.

    The initial premise is that nobody is born with theory of mind and that it has to be learned.

    The theory says that autistics do have theory of mind eventually.

    Anyway I believe it is an artefact of the way the Sally Anne test is constructed. Tests only measure the ability to perform that test, you can't necessarily reify that into the wider world.

    Unfortunately few people bother to look at the theory in depth and just take the popular sound bytes from it assuming falsely that no-one can be both autistic and possess a theory of mind at the same time.

    By Blogger laurentius rex, at 3:53 AM  

  • This post explains with such admirable clarity an objection that I've been muddling over for weeks. There is simply no way I would even attempt to test out the Sally-Anne experiment on my three-year-old son because there's no way I could explain the "Where will she look?" part so he would understand it - it's just light years beyond his receptive language capabilities.

    Unfortunately, the part of the theory of mind that makes it so media-friendly is the very part that breaks down under any kind of closer analysis: the idea that an autistic person is wholly unaware that other people have minds. At best, what we can take from this experiment is that autistic children may have more difficulty than others in sorting out who knows what (which, as joseph points out, puts them in the same boat as anyone attempting to read a description of the test).

    By Blogger bubandpie, at 2:04 PM  

  • "Because autistic children have significant differences in speech processing, it seems far more likely to me that they simply misunderstood the question."

    This is something I run into a lot when I am teaching. I teach autistic children ages 5-7 years-old.

    Often I feel that my students may know the answer to a question I ask but misunderstand the question. Some of my students will repeat my question back to me (possibly for clarification) I try to rephrase my question but often this leads to further frustration.

    Basically I am frustrated with my inability to ask my students a question appropriately.

    By Anonymous Anonymous, at 8:36 PM  

  • This comment has been removed by a blog administrator.

    By Blogger Phil Schwarz, at 1:00 AM  

  • (Sorry to litter the blog with a removed post -- it looks like a URL in my first attempt didn't get pasted correctly.)

    For a different deconstruction of the Theory of Mind hypothesis, see my comment to this blog entry.

    I argue that most non-autistics in fact do not have particularly good higher-order theory-of-mind skills, and that they *appear* to have the competence they do because of a privilege-of-the-majority advantage.

    By Blogger Phil Schwarz, at 1:04 AM  

  • Phil, that's a good point, and I also think that one of the reasons why autistics sometimes have trouble with social situations is that we understand other people's minds all too well, while being unaware of the extent of our own differences. We might assume, for instance, that because our friends at college all enjoy a big noisy party with lots of beer, we should enjoy it too. Then we can't figure out what went wrong when we don't. That's probably the main reason why so many autistics are relieved to get the diagnosis -- because it explains why our observations of others' behavior may not apply to ourselves.

    By Blogger abfh, at 10:37 AM  

  • I'm responding to this weeks late, so I'm not sure that anyone will even see this, but here it goes...

    I don't have any personal experience with autism. I'm a psychologist who tries to understand the mind, and who is trying to understand autism.

    I've read the "Deconstructing Sally-Anne" post, and yes, it does provided another possible interpretation of the Sally-Anne experiment. I'm not convinced, though, that this new interpretation is the correct one. Most children, by around age 2, spend a lot of time trying to get others to look at things that interest them. (For example, a child may hop up and down shouting "Kitty!" if they see a cat.) But, as I understand it, autistic children don't do this. They don't seem to try to interest other people in what they are seeing or doing. If autistic children have a normal theory of mind, why don't they do this?

    Also, if you are correct that autistic people have a normal Theory of Mind, then I don't understand what autism is. Can you explain it to me?

    By Anonymous Judy S, at 6:21 PM  

  • Hi Judy. Considering how vague and subjective the diagnostic criteria are, I don't believe that anyone really understands what autism is. However, it is a fact that autism is not, and never was, clinically defined by a lack of "normal" theory of mind (whatever that might be). As David Andrews mentioned above, the Sally-Anne experiment was merely a "pondering" and has nothing to do with clinical diagnosis.

    In my experience, autistic children often want to show or tell others about things that interest them, sometimes obsessively so. They may not use language to communicate their interests in the same way as other children, though.

    By Blogger abfh, at 1:44 AM  

  • I wasn't really thinking of how autism might be diagnosed/defined clinically, but rather in terms of what the underlying difference might be between autistic people and non-austistic people. I suppose that perhaps no one has a good idea of what this underlying difference is, though.

    Thanks for the comments.

    By Anonymous Judy S, at 9:15 PM  

  • i skipped over a lot of the comments cos i am about to go offline but i just wanted to say THANK YOU to everyone who had the same reaction as me to this test how ever many times i have read it in various different ways... i couldn't process it at first try either.
    also i would like to add that without a mental image of sally and anne i couldn't remember which was which, so that confused things further...
    is that really messed up?! (~_^)

    By Blogger Natalia, at 11:54 AM  

Post a Comment

<< Home