The Curious Incident of Research into Autism and Mirror Neurons in the Night-Time.

Any study that seeks to look into how something works in your everyday normal adult, is well served by examining how that thing works in people who have some sort of 'abnormal' condition. Abnormal, I should point out, is one of these words that has pretty rough connotations outside of the academic world. When I use the term, I mean "having some condition that the majority of humans don't have". Not "Having a condition which makes them sub-human". The other way round works just as well. When studying panic disorder (there is a faint chance someone might recall) I learned a lot about the condition by comparing how afflicated and unafflicted individuals dealt with their anxiety. It works for almost anything, and is a very good starting point.

In the last few weeks I have been writing about how mirror neurons and emotional contagion work in normal adults, without delving into equivalent research for abnormal folks. One of the main points of contention for MN theories is that it may or may not act as a tool for aiding the process of empathy. Emotional contagion, on the other hard, is widely considered to be a sort of automatic precursor to the full experience of empathy. Dr Silas, my potential PhD supervisor at Roehampton, worked with his colleagues to examine whether adults exhibited any difference in MN activation depending on their self-reported empatyh and systemising levels. Their results, like those of many other studies, answered with "sort of yes, sort of no" (Silas, Levy and Holmes, 2010). They looked at two types of EEG measurement: Evoked Readiness Potentials (nerve activations responding to a presented stimulus) and Induced Mu Desynchronisation (brain signals within the frequency range of 8-13 Hz that are inhibited when individuals observe, imagine or execute certain movements).

I love the look of this equipment so much, here's another picture. Dunno about you, but I'd be sitting there thinking: "Maybe they can actually read my thoughts. Don't think of tits. OH GOD now I'm thinking of them! Chairs! Chairs!"

They had participants go through a number of trials in which they had to press one of two buttons when the corresponding sound was played (execution). The same participants later had to watch another person perform the trial (observation). These simple tasks would allow the researchers to examine ERPs and IMD during aciton and observation. They had participants rate their levels of empathy and systemising using some standard tests that have been shown to fairly accurately measure both traits in previous studies (I am a little confused about which tests were used exactly, as there seem to be a number of different ones used in vairous experiments and I am unfamiliar with the differences between them). As expected given the vast majority of research, males had higher systemising and females had higher empathy to a statistically significant level. The researchers would have been glad to find that females showed stronger IMD during observation (indicating higher MN activity) but then were probably stumped at discovering that there was no correlation between IMD and the empathy/systemising scores. This result has been pretty typical of other studies, such as Cheng et al. (2008)'s.

The experiment had non-autistic adult males and females observing either a moving dot (non MN inducing control) or hand gestures (known to activate MNs). They also asked them to fill in three questionnaires: The Empathizing Quotient (pretty self explanatory for what it does), the Systemizing Quotient (again, self explanatory), and the Emotional Contagion Scale (Hah, now there's an interesting thing). Results? Males and females had significantly different empathy and systemising scores in the expected direction, and for both sexes higher systemising meant less IMD (thus less MN activity). Sounds like a step in the right direction, but although the correlation was statistically significant it was also very small, and measures of empathy did not correlate with MN activation in the main. Silas et al. (2010) mention several other studies which demonstrate some correlations but with many conditions which emphasise the current uncertainty about the relation between MNs and empathy.

Southgate and Hamilton (2008) examined the evidence for this theory that autistic people have abnormally functioning MNs (the broken mirrors theory), and came to the conclusion that it was not sufficient for explaining the difference between the imitation behaviour of un-autistic folks and the abnormal imitation of autistic people. They did not discount MNs as a contributor but just wanted to make clear that they believed the current explanations forwarded by many, to be premature. As my two examples had pretty uncertain results, they may be right. Focussing specifically on imitation abilities as a link between MNs and autism, the writers reviewed relevent studies and then offered their criticism. Imitation problems has been well documented as an issue in autistic individuals, with 21 studies involving 281 children with autism reviewed by Williams et al. (2004). The reviewers noted that overall, children with autism performed worse on imitative tasks. The nature of many of the tasks suggested that the issue was more to do with neural complications between sensing and motor control than with motivation or any other executive function. This is strongly supported by MN research, including that by Lacoboni et al. (1999) which identified the 'direct-matching' function of MNs. All in all, a good amount of research existed for Southgate and Hamilton (2008) to look into.

D'aaaaw. Imitation is awesome.

Along with all of the direct "Hey, people seem to have a mechanism for understanding and learning other people's movements" MN research, there are brain damage studies showing that lesions to important areas of the brain (in most cases, Broca's area) can mess with imitation abilities. Buxbaum, Kyle and Menon (2005) asked 44 patients who had had a stoke in their left hemisphere (where Broca's area is located in most brains) to observe and perform various pantomime movements. 21 of the participants had a condition called ideomotor apraxia which s characterised by the inability to imitate hand gestures or perform voluntarily pantomime actions. They were graded on how accurately they imitated what they observed. In both groups, the ability to recognise the movement they are observing (i.e. having experienced that movement in some way previously, and recall having done so) correlated positively with the ability to 'pretend' doing the movement. What this possibly  showed was that destroying internal representations of an observed action corrupted MN activity and so buggered up the usual imitation system.

Southgate and Hamilton's (2008) first of two big arguments against the direct-matching theory of MN supporters is that there is more to imitation than simple mirroring. Young infants (who tend to be as of yet unaware that they are a separate being) have been shown to imitate a model adult doing an action with her head less often when her hands were also occupied. When the model had her hands free but still used her head to switch on a light, the children also used their heads in imitation. When the model occupied her hands with the blanket, the 14 month old infants seemed to decide the only reason she wasn't using her hands was that they were limited, so they used their hands to hit the switch (Gergely et al. (2002). Infants who were not communicated with by the model, would use their heads irrespective of conditions. This showed more complexity than mirroring. The children were influenced by communication and some form of reasoning. Autistic children given the imitation task were nto influenced by communication, and imitated with their head most of the time in either condition. Southgate and Hamilton suggested that this demonstrated that it was the acceptance of communication at fault rather than an impaired internal representation a la broken mirrors theory. Additionally, if direct-matching existed as a way of gradually replacing less efficient actions for more efficient ones as one develops new skills, why did the children swap efficient means (using hands) for inefficient ones (using head)?

The second of their arguments rested on the admission by MN researchers than the MN system is not purely involved in imitation. Rizzolatti and Craighero (2004) were involved in much of the initial macaque MN studies, and noted that although they definitely have them, they rarely imitated each other in the way humans do. Wilson et al. (2005) argued that MNs play a big role in predicting the behaviours of conspecifics (members of the same species). Rizzolati et al. (2001 and many other big names in the area) suggested that their primary role is the understanding, not the imitation, of other people's actions. Impaired MNs should thus impair not only imitation, but these other functions as well if broken mirrors theory is to be credible. In Buxbaum et al.'s (2005) study, the participants with ideomotor apraxia (lesions in MN-areas) showed significantly poorer action recognition as well as imitation abilities, but there is no evidence of similar problems existing in autistic individuals. Carpenter et al. (2001) gave autistic and non-autistic children (2 - 5 years old) a test which asked them to view actions which ultimately failed in their intended purpose. There was no significant difference in ability to identify what the intended action was between groups. I.e. both groups seemed to be able to understand the action purpose fine.

Is the mirror completely intact, then? Or perhaps just chipped a little.

Williams et al. (2004) mentioned above, did a big review and found that autistic kids weren't ace at imitation, but this has more recently been criticised by Hamilton, Brindley and Frith (2007). They asked autistic and non-autistic children to take part in a series of tasks designed to test their ability to copy actions. Both groups performed equally well, and the autistic children even had superior performance in action recognition tasks (though not by much). The reasoning behind the contrasting results? In Hamilton et al's study the children were explicitly instructed to imitate, rather than just hinted at as in Gergely et al's (2002) research. Bird et al. (2007) had the clever idea to pre-specifc the imitation that the participants should perform in reaction to a human or robotic hand making an action. The robot/human would hold their hand with thumb and fingers parallel and facing the roof, they would then either open or close their fingers and the participants (non/autistic) would have to imitate the action. In a second condition the participants were asked to do the opposite action. For both autistic and control groups the compatable imitations occured faster, and the reactions were faster when watching a human hand (animacy bias). All in all, there was no significant difference between the groups. So autistic children can imitate so long as they are specifically instructed to do so. Southgate and Hamilton (2008) suggested that their problems with imitation thus stem from knowing when and what to imitate (social problems) rather than any problem with the actual imitation mechanism. We know that autistic children tend to dislike making eye contact or engaging with infant-directed speech, so having an impaired ability to judge when it would be right to imitate is perfectly plausible.

So I will conclude, that on the basis of the reviewed papers the role of the mirror neuron system in autism is far from certain. Neurological studies have found no clear or reliable link between traditional measurements of empathy and systemising, and direct research of MNs in autistic people have found little to no reliable results. Autistic childen have been shown to have poor imitation, but more recent studies showing that direct instructions result in nromal levels of imitation bring that largely into quesstion. Additionally, if MN impairment was a major contributor to autism then more than just imitation would be impaired, there is little evidence that this is the case. Due to all of this, it very much seems like broken mirrors theory is premature and we have a lot of work to do to narrow down what contribution MNs make to autism.