They found that although the autistic children did not differ from the younger, typically developing children in the amount of time spent looking at their own faces, but that they did spend a lot more time looking at objects in the mirror, and that their behavior toward their reflections differed from that of either control group. The autistic children did not generally try to relate socially to the person in the mirror; what they did instead varied according to whether they recognized their own mirror images or not.
BPS Research Digest spotlighted this study published in last month's issue of Autism, which claims to describe another type of subtle behavioral difference between autistic, otherwise developmentally disabled, and typically developing children.
Autistic children, they hypothesized, would act differently toward their own images in mirrors than either typically developing toddlers or developmentally delayed --- but not autistic --- children (in this case having Down syndrome) of similar ages to the autistic children. They would look in the mirror, and would experiment with it (i.e., doing things to see those things reflected back to them, using the mirror to look at things behind them, etc.), but would not try to engage the person in the mirror socially; would not talk to it, smile at it, or try to show it things.
(Reading this, I was kind of surprised to see that it was normal to treat the mirror image as if it were a real person! It doesn't act like one, and you'd think that if you recognized it as yourself, you'd know you couldn't actually talk to it or interact with it, because it would just do the same things you did. Of course, the children in this study were very young --- the typically-developing ones were about eighteen months old, and the Down syndrome and autistic kids three or four years old.)
Methodologically, the study's not that interesting: the researchers went into the homes of each of their subjects (38 in total: 12 with autism, 13 with Down syndrome, and 13 typically developing toddlers), produced a mirror, filmed the children interacting with it freely for two minutes, and then tested to see if the children recognized themselves in the mirror by having their parents put a sticker on their faces to see if they removed the sticker after seeing it reflected in the mirror. If they removed it, they were said to have passed the "Mirror Self-Recognition" (MSR) test*.
Later, the researchers watched the videos they'd made of the children interacting with the mirror, looking for group differences (between autistic, Down syndrome, and typically developing children, but also between passers and failers of the MSR) in such things as level of interest in one's own face (i.e., how long does the child look at hir own face, rather than at other things reflected in the mirror), kind of actions directed at one's reflection (i.e., does the child greet hir reflection, talk to it, perform in front of it, or show things to it), and affective response to one's reflection (does the child appear pleased to see hirself, do they appear self-conscious, coy or embarrassed, do they smile at themselves).
Compared with the other groups, the autistic children spent a lot more time looking at objects in the mirror. However, contrary to what you might expect given the common wisdom about autistic people avoiding eye contact/looking at faces, though, they did not differ much from the typically-developing ones in how long they spent looking at their own faces. (The children with Down syndrome spent the most time looking at their own faces).
Figure 1, in Reddy et al. (2010). The panel on the left shows the amount of time (expressed as a percentage) that the different groups of children (autism, Down syndrome and typically developing toddlers, each subdivided into passers and failers of the mirror self-recognition test) spent looking at their own faces. The panel on the right shows the amount of time they spent looking at other objects in the mirror. In both graphs, the placement of the bars (i.e., are they higher or lower) is what tells you how much time a given group spent doing a given thing. The size of the bars (i.e., are they long or short) tells you how much variation there is within that group.
The autistic children also differed from the other two groups in what kinds of things they did in front of the mirror; autistic children, whether they recognized themselves in the mirror or not, spent a lot less time trying to relate to their reflections socially. What they did instead varied with whether or not they understood that they were looking at themselves: autistic children who passed the MSR test spent most of their time experimenting with the mirror, tilting it to see things around the room, doing things with toys or with their faces while watching to see those actions reflected back to them, while autistic children who failed to recognize their own reflections spent more time simply watching the person in the mirror.
Children with Down syndrome also spent more time watching their reflections if they failed the MSR test; the typically developing children spent about the same amount of time watching themselves whether or not they seemed to know they were watching themselves. They also spent more time watching themselves, relative to other actions, than either of the atypical groups. The study authors hypothesize that "[a] watchful focus on the self could be due to imminent self-recognition (suggested by the finding of a short-term alignment between watching and self-recognition in typical development, Nielsen et al., 2003)."
The typically developing children also did not show any relationship between MSR and social-relating behavior toward the mirror: whether they recognized themselves or not, they were just as likely to act as if their reflection were a social partner or an audience, as opposed to just a reflection (which is how the autistic children tended to treat their reflections, if indeed they recognized them as such). The children with Down syndrome tended to do *more* social relating with the reflection if they passed the MSR test, which seems counterintuitive to me. You'd think someone would be more likely to try to establish a rapport with something they believed to be another person, rather than with what they knew was only an image of themselves. (Oh, well --- like I said before, I'm no developmental psychologist!)
So, what do this study's authors (the University of Portsmouth's Prof. Vasudevi Reddy and Cristina Costantini, the University of Surrey's Dr. Emma Williams, and Britta Lang) think their results mean?
Their study was published in a special issue of Autism pertaining to how a sense of "selfhood" develops in autism, and indeed they do manage to tie these findings to an alleged autistic impairment** in developing this sense:
Difficulties in interpersonal relatedness in autism appear to extend to difficulties in relatedness with the self, supporting arguments about a reciprocal relation between a sense of self and a sense of other (Hobson, 1990; Mclaren, 2008). In typical development the affordance of the face is almost unavoidably social, with direct gaze attraction attention from birth (Farroni et al., 2002) and acting as an ostensive signal (Senju & Csibra, 2008). In autism, perhaps particularly in mirrors where there is no one else to initiate engagement and no other social behaviour to highlight interpersonal cues, this affordance may be even less potent in inviting interaction. Engaging with the self can also provide opportunities for learning about expressions and interaction. Given the enjoyment that typically developing children and children with some other developmental disorders derive from such engagement, the loss of this opportunity in autism might potentially contribute to further impairments in the development of a sense of self.
This passage, along with a passage I will also excerpt from the Introduction ---
In developmental psychology the mirror has become synonymous with the identification of the self ... . But mirrors can also symbolise and allow a relation with the Other. They can reflect the self back as the Self, as an Other, as seen by an Other (Kernberg, 2006) or, indeed, as just another reflection. How one reacts to the self in a mirror allows us to study the extent to which the self is perceived and presented as a social being.
--- argues that there is a social dimension to the sense of self, which seems to start developing pretty early.
It also echoes the feedback-loop theory I've read about in connection to another early-childhood marker of autism that's been studied recently: eye gaze. Autistic infants have been found not to prefer looking at human faces, which, the theory goes, leaves them without motivation to pay special attention to them, while their typically developing peers, who do derive pleasure from looking at faces, learn to specialize in reading faces. Reddy and her colleagues' suspicion that the mirror does not "invit[e] interaction" for autistic children in the same way it does for their typically developing and Down syndrome peers extends this notion to the broader realm of reciprocal social behavior in general.
I have one more thing I'd like to bring up in connection with this study, and with mirror self-recognition tests in general: this earlier post on BPS Digest describing cross-cultural studies of mirror self-recognition tests. Apparently, children from different cultures perform differently on the mirror self-recognition tests (which involve placing a spot on the child's face and waiting for hir to remove it when ze sees it on hir reflection), with children in some cultures failing to remove the spot even at ages much older than the second year of life, which is when children in the U.S. (and similar cultures) start passing the test. This is not because these children don't understand that they are looking at themselves, but because their cultures have different standards for how children are supposed to behave in adult company. The children may not know if they're *supposed* to remove the spot, so they might choose to do nothing rather than act rudely and be punished for it.
I bring this up, even though I'm not sure it has any direct bearing on the results of this mirror self-recognition study, because I think it is important to remember that autistic children's social behavior, in this (U.S., white, middle-class) culture and others like it, is policed much more aggressively than typically developing children's behavior is. Autistic children learn quickly that their natural ways of speaking, acting and relating to people are wrong, so they might well adopt a more passive social posture until they've seen enough to know what's expected of them.
Again, I don't think that's necessarily at work in this particular study (the children are so young, I doubt they've yet gotten much negative feedback about how they act), but I want to mention it because I don't know that many autism researchers consider the effect of culture and socialization on autistic children, much less ways autistic children are socialized differently than non-autistic ones. Instead, the common wisdom seems to be that we are immune to culture and socialization.
*An explanation of this test, and what it supposedly says about sense of self, can be found here at Noah Gray's blog, Nothing's Shocking, hosted by Nature.com. Basically, if the subject (human or animal) knows to remove the sticker (or wipe off the spot of paint, or whatever) based on having seen it in the mirror, we can reasonably infer that they know the mirror is showing them an image of themselves, not another person or animal.
**I think it's funny, if only from an etymological standpoint, that AUTistic people are now thought to have an impaired sense of self.
Reddy, V., Williams, E., Costantini, C., & Lan, B. (2010). Engaging with the self: Mirror behaviour in autism, Down syndrome and typical development Autism, 14 (5), 531-546 DOI: 10.1177/1362361310370397