Can One Assign the Wrong Intentions to Triangles?

I’ve recently run across two studies in which an ability to impute mental states and empathize with others was measured by having the research participants look at inanimate objects moving across a computer screen. Needless to say, I find this particular method rather questionable.

Here’s the rundown: A 2000 study by Abell, Happe, and Frith attempted to measure theory of mind by asking the participants to describe two moving triangles in computer animations. The researchers showed the animations to a group of adults, a group of eight-year-old autistic children, and a group of eight-year-old typically developing children. The animations were constructed by the authors to show random behavior, goal-directed behavior, and deceptive behavior. Most of the adults used intentional and emotional terms to describe the actions of the animations. The autistic children ascribed mental and intentional states to the triangles less often than the non-autistic children, and when they did ascribe mental states, the researchers described their answers as “inappropriate.”

A related 2006 study by Knickmeyer et al. attempted to measure whether fetal testosterone is inversely associated with empathy. To do so, the researchers analyzed the levels of fetal testosterone in the amniotic fluid of 38 typically developing children who had reached the age of four and, as in the 2000 study, showed the children cartoons with two moving triangles. The result was that more girls than boys used terms reflective of relationships, emotion, intention, and mental states to describe the triangles, and that levels of fetal testosterone were directly correlated with a lack of intentional thinking and the use of emotion-neutral propositions. The researchers reached the conclusion that the result shows a correlation between fetal testosterone and social development. Because a previous study had shown that autistic children score more poorly than typically developing children on the same task, the researchers also concluded that their findings support the extreme-male-brain theory of autism — that is, the theory that autistic people have male-gendered brains.

Before I continue, let me summarize the logic of both studies:

a) Autistic children do not impute mental states to inanimate objects as often as non-autistic children and adults,


b) Typically developing children who had higher levels of testosterone in their amniotic fluid do not impute mental states to inanimate objects as often as children with lower levels,


c) Autistic people have extreme male brains.

You’ll note a few missing pieces in the logic here. This phenomenon arises from the fact that the researchers failed to pose a number of critical questions:

1) How does a failure to anthropomorphize inanimate objects indicate a problem with mentalizing, empathy, or pro-social behavior? An alternative explanation would be a bias in the autistic children toward seeing the world as it really is.

2) Given that triangles are inanimate objects and don’t have mental states, how could anyone possibly measure, scientifically or otherwise, whether the mental state one ascribes to a triangle is correct? Showing the participants a computer animation and telling they’ve gotten the answer wrong is like giving respondents a Rorschach test and telling them they’ve failed.

3) What, exactly, in a scientific paper, is the objective, quantitative definition of “inappropriate”? To my ears, the word translates as “You haven’t given the answers we had in mind when we set up the test.”

4) How exactly does a higher level of fetal testosterone make the culturally defined construct of “male” as “high systemizer/low empathizer” biologically determined in autistic brains?

Of course, the chief flaw in the study is the subjective nature of the ways in which the researchers view the cartoons. For instance, in the 2006 study, the researchers see the motions of two of the triangles as a mother coaxing her child to go outside, and they expect that their view will be shared by all of the participants. When the participants don’t see the shapes in the same way, the authors conclude that the participants are lacking in empathy and pro-social behavior. I can’t see any evidence that a failure to anthropomorphize inanimate objects indicates a problem with empathy or social relationships. An alternative explanation would be a bias toward simply calling a triangle a triangle, which is in no way opposed to empathic response.

Now, I know what you’re thinking, because the same thought occurred to me: “Autistic people tend to take things literally. Of course they just see triangles. Why does that have anything to do with empathy?” But you see, in the logic of autism research, the fact that autistic people take things literally is itself evidence of impaired empathy and theory of mind. Here’s the (very circular) logic:

a) Autistic people take things literally because they have impaired theory of mind


b) Autistic people don’t ascribe mental states to inanimate objects, but see them literally,


c) Autistic people have impaired theory of mind.

Sometimes, it just amazes me that scientific studies purporting to result in objective and quantitative measures are informed by so much subjective bias. But of course, given that such studies are constructed from inside the consciousness of one set of human beings in order to describe the consciousness of another set of human beings, they are, by definition, permeated by subjectivity. It’s not the subjectivity I mind; if the subjectivity of the researchers were fully factored into the research, as is the case in qualitative research, then the issues would be clear for all to see, and the questionable nature of the conclusions would be more readily apparent. It’s the pretense of objectivity that I find most objectionable, and that I consider one of the most serious issues in the research.


Abell, Frances, Frances Happe, and Uta Frith. “Do triangles play tricks? Attributions of mental states to animated shapes in normal and abnormal development.” Cognitive Development 15, no. 1 (January-March 2000): 1-16. doi: 10.1016/S0885-2014(00)00014-9.

Knickmeyer, Rebecca, Simon Baron-Cohen, Peter Raggatt, Kevin Taylor, and Gerald Hackett. “Fetal testosterone and empathy.” Hormones and Behavior 49, no. 3 (2006): 282-292. doi: 10.1016/j.yhbeh.2005.08.010.


Rachel Cohen-Rottenberg blogs at Journeys with Autism, and presides at Autism and Empathy.

Can One Assign the Wrong Intentions to Triangles? appears here by permission.

The most recent installment in Rachel Cohen-Rottenberg’s published memoirs is Blazing My Trail.

[image via Flickr/Creative Commons]


on 01/16/12 in Autism, featured | 6 Comments | Read More

Comments (6)


  1. ictus75 says:

    Theory of Mind is one of the most bogus theories out there. It’s as bogus as the ‘backwards masking’ theory about rock records back in the 1960s. It all comes down to, you believe what you want to believe, get the results you were wanting, because you can only see what you are looking for.

    This is typical NT research where they judge all Autistics by NT criteria, which is just plain wrong. If you judge an apple by banana standards, it will always fail as a banana. But if you judge an apple on its own, it will be a great apple.

    And what in the world do triangles have to do with anything? Who came up with that one? I think it’s really the other way around: there’s something wrong if people read too much into animated triangles.

    BTW, what is so wrong with being different?

  2. peter says:

    By coincidence I showed this to my autistic son over the weekend, if you ask leading questions he is ok but I knew what it was supposed to be about because I had read the article before watching the clip. I also dislike the term inappropriate. I wonder what would happen if they included a control, a clip that meant nothing to anyone (created by someone who didn’t know what the expectations of the experiment were) and see how NTs reacted compared to As – or is that a stupid idea. In my opinion empathy is defined by NTs as having the correct verbal response to a situation and sincerity is optional.

  3. BunBot says:

    Yep, it’s the “pretense of objectivity” that is the most insidious aspect of autism research. When it comes to studying mental states and internal experiences, there is no objective reality, like the NT researchers believe there is – the NT subjective reality. All other subjective realities need not apply.

  4. I have little patience for Theory of Mind theories about autism. However, just thought I’d provide a little more background about why these researchers used animated shapes in this way.

    One common method for studying our mental wiring is studying “brain bugs”–the sort of misperceptions people make. Vision scientists study visual illusions because they reveal something about how we see. “Brain bugs” are brain functions that are normally helpful, but in certain situations give us information that is untrue and counterproductive. In addition to visual illusions, these include certain common logical fallacies and reasoning errors. Attributing intentions to shapes, which don’t actually have them, is a “brain bug” that NTs have but autistics don’t. Given the tradition of using brain bugs to back-infer function in NTs, it’s natural for researchers to assume that NTs have the function revealed by the brain bug and autistics don’t.

    The researchers then jump to conclusions about what that function actually is. All we know is that NTs automatically associate certain patterns of movement with certain intentional behaviors, even when looking at things without intentions, like shapes. We don’t know how they do this, what sorts of intentional behavior they do and don’t infer, whether all NTs make the exact same inference as the researchers, or what if anything this has to do with real-world social behavior. An unbiased researcher would explore further, but these guys are just using it as an opening to justify the old SBC “extreme mail brain” story. Disappointing.

  5. Maria Taheny says:

    Laughed pretty hard when I read this. Brilliant!

    I am a PhD candidate in Clinical Psychology doing a qualitative study on autism and families in a faith community. I read the study you mentioned in an effort to find support or dispel the argument of autistics having a theory of mind. I was disappointed by this study, in that I came to the same conclusions you did. The study is flawed precisely BECAUSE it is established with parameters defined by the neurotypical. THEY lack the empathy needed to understand an autistic experience! (Thus the reason I am doing my research work with autism. Where better to get the information than from the horse’s mouth?)

    Anyway, I agree. The biases of the study simply confirm that the ones that are unable to understand the autistic experience and culture are the neurotypical. Maybe they are the defective ones?

  6. […] Traduction d’un article de Rachel Cohen-Rottenberg, publié ici. […]

Leave a Reply