Thursday, December 8, 2011

Why autism has nothing to do with 'broken mirrors'

I've argued that the mirror neuron theory of action understanding is backwards.  Mirror neurons do not fire because they are critically involved in action understanding (the typical claim), they fire because perceiving an action is critically involved in selecting actions for movement (Hickok & Hauser, 2010).  This point has broad implications for any theory that builds on the mirror neuron theory of action understanding.  I've had plenty to say about the broader implications in the domain of speech (e.g., Hickok, 2009; Hickok, et al. 2011; Lotto, et al. 2009); now it's time to take a stab at the claims made beyond speech.

One of the most prominent and potentially important extrapolations of the mirror neuron theory of action understanding concern autism spectrum disorder (ASD).  The "broken mirrors" hypothesis of ASD, exemplified by a Scientific American article by V.S. Ramachandran, is built on the following logic.

1. The mirror system allows us to understand the action of others.
2. The mirror system, by extrapolation, allow us to understand the emotions, intentions, and perspectives of others.
3. ASD involves a lack of sensitivity to the emotions, intentions, and perspectives of others (in particular a lack of empathy).
4. Therefore ASD results from functional disruption to the mirror system, that is, from "broken mirrors".

Since I've argued that assumption #1 is false, the logic simply falls apart.  However, let me attack another of these assumptions, namely #3, that ASD involves a lack of sensitivity to emotion, etc.  I'm going to argue instead that ASD involves, in fact, hyper-sensitivity to emotional states, both their own and others.  This hypothesis is not new.  In fact, Henry Markram, Tania Rinaldi, and Kamila Markram proposed exactly this in their 2007 paper titled, The Intense World Syndrome -- An Alternative Hypothesis for Autism. I would just like to underline their perspective here.

An analogy is useful for seeing why hyper- rather than hypo-sensitivity makes more sense.  Imagine a person who is hypo-sensitive to sound.  Is such a person more or less likely to walk into very loud environments?  More likely!  If you are less sensitive to sound you might actually prefer loud environments because that's how you get your acoustic sensation to a normal level. Conversely, a person who is hyper-sensitive to sound is going to avoid loud environments because it hurts.

Now consider the same scenario translated to the social/emotional domain.  For starters let's agree that a large part of one's emotional stimulation comes from social situations.  Not all emotional stimulation comes from the social domain, to be sure: I can get pretty emotional when I can't figure out how to fix my leaky faucet.  But we get a lot more emotional and more often when a family member is sick or injured, or when a colleague rolls his or her eyes when we try to make a point, and so on. Now, imagine a person who is hyper-sensitive emotionally. Is such a person more or less likely to engage in social situations?  Less!  For someone who is hyper-sensitive to emotion, engaging in a normal social situation would be like walking into an excessively loud environment: it's uncomfortable and causes an avoidance response. ASD individuals may avoid social interaction not because they lack empathy but rather because social interaction is simply too stressful.

Some data:  A characteristic of autism is decreased gaze fixation, particularly on faces.  Correspondingly, the fusiform face area (FFA) appears to be less active in autistic individuals during face processing tasks.  This might be interpreted as social indifference, a lack of interest in faces.  However, another interpretation is that looking at faces, which are a major source of emotional information, is stressful for individuals who are hyper-sensitive to emotion.  An imaging study of face perception published by Dalton et al. in 2005 supports this view.  This study found that the autistic group did indeed spend less time fixating the eyes and that there was less activity overall in the FFA for the autistic than control group.  However, this is not all that surprising because if you spend less time looking at a stimulus, your brain activation in regions sensitive to that stimulus will naturally be lower.  And in fact, when Dalton et al. looked the correlation between FFA activity and gaze duration they found a strong and positive correlation in autistic subjects.  In other words, the FFA of autistic folks is working fine, it's just that they spend less time looking at the stimulus.  More importantly, they report that amygdala activity was strongly correlated with gaze duration in the autistic group but not in the control group.  Interpretation: looking at faces is more emotionally arousing in autistic than control individuals.

Hyper-sensitivity to emotion is perfectly consistent with the well-known sensory sensitivity noted in ASD.  In other words, there are other reasons to believe that hyper-sensitivity is a key feature of the syndrome across the board.

The mirror neuron folks got it perfectly backwards again.  You don't have to take my word for it though.  Here are some links to essays written by individuals with ASD or by a parent.  The links were provided by Morton Gernsbacher, an author on the study referred to above.  I found them particularly illuminating.


Dalton, K., Nacewicz, B., Johnstone, T., Schaefer, H., Gernsbacher, M., Goldsmith, H., Alexander, A., & Davidson, R. (2005). Gaze fixation and the neural circuitry of face processing in autism Nature Neuroscience DOI: 10.1038/nn1421

Hickok, G., Eight problems for the mirror neuron theory of action understanding in monkeys and humans. J Cogn Neurosci, 2009. 21(7): p. 1229-43.

Hickok, G. and M. Hauser, (Mis)understanding mirror neurons. Curr Biol, 2010. 20(14): p. R593-4.

Hickok, G., J. Houde, and F. Rong, Sensorimotor integration in speech processing: computational basis and neural organization. Neuron, 2011. 69(3): p. 407-22.

Lotto, A.J., G.S. Hickok, and L.L. Holt, Reflections on Mirror Neurons and Speech Perception. Trends Cogn Sci, 2009. 13: p. 110-114.


Guy said...

You accuse "mirror people" to make backwards arguments, but the logic of your sound sensitive person analogy is backwards by itself.
A sound insensitive person would prefer noisy environment that is fine. However, it does not mean that any person who prefers noisy environment is sound insensitive. Some people like noisy bars, rockNroll shows etc, are they all sound insensitive?
Similarly, the observation that a person avoids social interactions does not mean that he is socially hypersensitive, though this hypothesis should not be neglected.

drbrocktagon said...


Something else I've never quite understood about the whole "Broken Mirrors" Theory is where these mirror neurons get their mirror properties from. Unless we accept an extreme form of nativism whereby each neuron is genetically programmed to represent what it represents, then we have to conclude that mirror properties are somehow 'learned' through experience (ie social interaction and observation).

Essentially, Ramachandran et al are trying to explain social interaction impairments in terms of the dysfunction of mirror neurons, the normal development of which depends on the social interaction they're trying to explain the impairment of.

Peter said...

You're right, you'd expect kids with autism to have an exaggerated galvanic skin response to faces with direct eye gaze. If they're right you'd expect a muted response. The correct answer is..........exaggerated.
Here's the ref:
Journal of Autism and Developmental Disorders
Volume 36, Number 4, 517-525, DOI: 10.1007/s10803-006-0091-4
Skin Conductance Responses to Another Person’s Gaze in Children with Autism
Anneli Kylliäinen and Jari K. Hietanen

neuromusic said...

"Interpretation: looking at faces is more emotionally arousing in autistic than control individuals."

My amygdala is lighting up, so my reverse inference detector must have just gone off.

neuromusic said...

Also, the "intense world syndrome" does not invalidate Rama's claims.

It would not be hard for Rama et al to say that a hyper-sensitivity to emotion is an example of a "broken mirror" (a funhouse mirror poorly represents the environment whether convex or concave).

I suggest focusing on the evidence against (1), the illogical extrapolation of (2), and the bad logic of (4).

Greg Hickok said...

Nice call, neuromusic. I'm glad your reverse inference detectors are working. Look a little more carefully though and you'll see that the hypothesis was generated on the basis of logical inference from behavior. The neural data was interpreted as consistent with the independently generated hypothesis, not the driver of the hypothesis. Nonetheless, it is a fair point that we should not assume that amygdala activation means increased emotional reaction.

As long as you've got your reverse inference detectors highly sensitized, please read through the mirror neuron literature and underline all 4,502 reverse inferences ;-)

neuromusic said...

I suppose I should call it my "reverse inference neuron system".

VilemKodytek said...

Peter, the guys do know about the raised skin conductance. In fact in the Scientific American paper there are so many hypotheses that some must inevitably be correct. And that’s the point: they’re probably going to find something useful regardless the truth value of the broken mirror hypotheses. The “broken mirror” looks like a mere logo of their R&D activities.

The following papers for me confirm Greg’s view at a different level of research:
E. Ochs – O. Solomon – L. Sterponi (2005): Limitations and transformations of habitus in Child-Directed Communication. Discourse Studies 7, 547–583. doi 10.1177/1461445605054406.
E. Ochs – O. Solomon (2010) Autistic sociality. doi 10.1111/j.1548-1352.2009.01082.


Mal said...

I'm a bit lost --maybe because I'm not a neuroscientist-- about the analogy autism-sound. It looks to me it is not about being sensitive to sound, but being able to feel emotions listening to music. Exactly like empathy means feeling emotion looking at someone else's face. And if I were more sensitive to music, I would avoid modern rock more than I do know, but would go to jazz concerts more than I do know.
If I were hyper-sensitive to emotions, I'd probably stay more with happy people. I'd be more helpful not just to my family members but also to unknown people. I'd be such a cool guy, you know:)
I'm not necessarily a fan of mirror neurons (I'm not a neuroscientist, then I do not care that much), but I am a fan of empathy. I do believe that empathy is what made apes quite successful as social animals. I think that evolution teaches that humans with higher sensitivity would be more successful in society, not less (see e.g. Preston, S. D., & de Waal, F.B.M. (2002) Empathy: Its ultimate and proximate bases. Behavioral and Brain Sciences, 25, 1-72).

Greg Hickok said...

You are thinking about hypersensitivity within a normal range. I'm talking about abnormally high sensitivity. If you have excellent hearing sensitivity (higher than average) you might engage more in musical activities. But ramp up the gain to abnormal levels and any acoustic stimulation, even the most melodic, pleasing piece will be unbearable noise. It's the too much of a good thing principle.

Anonymous said...

I'll openly admit that I don't know much about the specific science involved (though I am always willing and trying to understand it), but I do know that your description of how the autistic mind interacts with and understands the world, makes a lot more sense of my autistic son than the theory you are critiquing.

And then at the end I realised one of the stories you found to be illuminating, was one I wrote about my son :)

It's nice to feel listened to and understood. And it's encouraging to read these discussions and see these ideas being aired. I know the aim of science isn't to give us that feel-good groove, or to be encouraging, but there it is anyway. So thank you.

Anonymous said...

I know next to nothing about mirror neurons, but I am curious. I've known people who seem to have some difficulty empathizing, to put it mildly. One says he has 'mild' Asberger's. In his case, he actively, doggedly seeks out social interaction. He calls many acquaintances 'friends,' but I am sure I'm categorizing the relationships more accurately. He is somewhat dependent on the kind of social interaction where others have no real choice but to allow his participation - church groups, volunteering at nonprofits, etc.

There is almost a palpable annoyance factor present when he tries to relate to others (many related issues). The point is, he is actually trying very hard to interact socially and build/have relationships, but few find it satisfying. He is included in gatherings and events more from a sense of politeness or not wanting to hurt his feelings, and is also sometimes 'forgotten' to be included. It's quite sad, actually.

I've been a friend of his for a couple of years and it is often quite difficult. I know this is anecdotal, but it makes me skeptical of the hypothesis here. I wonder if the degree of autism matters. Asberger's is in the autism spectrum as a mild form. I think there is certainly a difficulty in empathizing - but I do not have the sense that it's because he is hyper sensitive to emotions. Neither does he seek to avoid social interaction.
I would appreciate having a better understanding. I currently find I'm constantly reminding myself he cannot help how he is and he means well. I think. :)

Andrew said...

Hi Greg

I've been banging my head against mirror neurons for a little while; I agree that they've been mischaracterised. In my field (perception-action) motor theories tend to get replaced by event perception and my hunch has been for a while that mirror neurons, if they are up to anything, are involved in event perception, not 'action-recognition'. Your appeal for a sensorimotor account seems in this vein as well, I think, and the event perception literature might be of use to you.

I've laid some of this out here; I'd be interested in your thoughts, I'm not a neuroscientist but I'm interested in ways of pursuing this empirically. (I'll also trawl through your reference sections, it looks like there are useful results for me there!)


Rachel Cohen-Rottenberg said...

As the editor and publisher of the Autism and Empathy site, and as a highly empathic autistic woman, I want to thank you for this article. Based on my experience of myself and of other autistic people, I consider your analysis to be absolutely spot on. Your piece articulates many of the things that autistic people have said for years about our acute sensitivity to the sensory and emotional worlds. Far too often, our voices and our perspectives are left out of discussions of the science. By providing links to the Autism and Empathy site, you enable your readers to hear from those of us who experience autism from the inside out. Thank you.

Amanda Forest said...

I don't know Anonymous, I'm sure people like your friend exist and I do hear about them a lot. But lots of people I know with autism aren't like that at all and basically fit the hypersensitive theory. I'm pretty into it because I feel like we tend to get erased with all this "autism means no empathy" stuff being thrown around, and it's nice to read scientists talking about stuff that actually sounds familiar.

At the same time saying everyone with autism is like me doesn't seem much better than saying everyone with autism is like your friend. (I have to say, though, it doesn't sound like you're friends.) Sometimes I wonder if it is just two different disabilities accidentally being given the wrong name, and I'm wondering how the writers of this paper would describe his problems.

Rachel Cohen-Rottenberg said...

Anonymous, I have Asperger's and I'm very empathic. I always get a very strong sense of what is going on with people, which makes social gatherings very difficult. I walk into a room and I can feel all the emotions of the people there. A great many of us have to shut down our empathic connection when we're overwhelmed like that. So we come across as being unempathetic in the eyes of people who don't understand the extreme level of sensitivity involved. We often learn to simply shut down our responses in general, rather than be overwhelmed by them. And then people see the result -- rather than the process -- and think we lack empathy.

Add extreme sensory sensitivity, which is both physically exhausting and anxiety producing, to the mix and yes, we're not going to be able to express empathy the way that others do. But it's a big mistake to assume it isn't there.

Anonymous said...

Greg and David, I think you are right on.

I wasn't diagnosed with an ASD until I was in my 20's. Throughout my entire childhood and adolescence, as I tried desperately to figure out what was "wrong" with me that no one else would admit to, over and over again I kept coming back around to realizing that I was too sensitive. Both to sensory aspects of the environment and to emotionally difficult situations.

Mal: "It looks to me it is not about being sensitive to sound, but being able to feel emotions listening to music. Exactly like empathy means feeling emotion looking at someone else's face."

It is both. I am not only hypersensitive to sound volume and pitch (and other qualities which can make some quieter sounds far more painful and irritating than some loud ones--for instance, the sound of a nail file is far more painful to me than being at a rock concert), but also to the emotions conveyed by music. Likewise with human faces and eyes, we can be hypersensitive both to the physical medium AND to the emotional information conveyed by the physical medium. So sensitive that we have no chance of processing it in a usable way like most people can.

Think of it this way: even for music that you would otherwise like, there is such a thing as it being too loud for you to enjoy the music or process the emotion contained therein. That's what a lot of the sensory and emotional world is like for us. Not being able to process the emotional content into actionable form doesn't mean that we're lacking awareness of it, just that it's at a "pitch" or "volume" that we can't tolerate or use.

Add to this the fact that I am constantly having to explain empathy (mostly for marginalized/vilified people or minorities of various stripes) to non-autistic people (who have, practically in the same breath, supported torture and then told me I had no empathy), and I think the "intense world" theorists are several steps ahead of the pack on this one.

Ben Stansfield said...

I appreciate you being so honest about your feelings for your friend.
I have Asperger's syndrome, and though nobody really says these things to my face, it would not surprise me to learn that people who know me describe me in many of the ways you describe your friend.
I also try very hard to make friends, but I get that many people find it difficult or unsatisfying to befriend me.

For me, and possibly for your friend, I think that sometimes the differences between the way I work, and the way most people work, have gaps too exhausting for them to bridge. In other words, people who are more like one's self are less work to be friends with, and there are fewer people with brains like mine, than say, brains like yours. I don't get angry about this the way I did when I was young, but I still feel hurt by rejection or indifference.

It would probably be healthier for people to recognize annoyance as a feeling to be responsible for, not something that can be blamed on another person. This openness may not make your friend any easier to get along with, but it might. I can say from experience that most people expect me to adapt to the way they think, the way they want me to behave, rather than the other way around.
Ultimately, in a pragmatic sense, the majority will continue to rule, regardless of whether or not it's fair, moral or ethical. Or nice.

Unknown said...

Greg, do you feel that Kraskov et al 2009 (Kraskov, A., Dancause, N., Quallo, M., and Shepherd, S. (2009). Corticospinal neurons in macaque ventral premotor cortex with mirror properties: a potential mechanism for action suppression? Neuron.) adds some support to the idea of MN's as being involved in action selection rather than suppression? For myself, I can't make sense of a spinal cord projection if they are solely involved in action comprehension....


Greg Hickok said...

Hi Archy,
I hadn't seen this paper before. Thanks for pointing it out. It certainly seems to be more consistent with the action selection model. The suppression effect during action observation is particularly interesting in that much of motor selection involves suppression.

Anonymous said...

Hello, I am multitasking so I cannot read comments and am likely duplicating things said already. Apologies if that is so.

I appreciate this article, as a few of my fellow autists and I just this morning were ranting on Mirror Theory and the stupidity involved (that and abusive carers basing their presumption that I do not feel things because of said theory).

I dislike the head pat cookie reaction over all but, you seem to already know you have done a good job. You both. I still want to say thank you, not just for pointing out what to us Autistic folk is bleeding obvious but using references written BY autistic people. That is so rare that I want to say thanks.

Liz Ditz said...

Thanks for this article, and also thanks for quoting my friend Rachel. It is great to see researchers actually reading and quoting autistic people.

I posted a link to this at The Thinking Person's Guide to Autism's Facebook page:, where a discussion is going on in the comments. I'm not sure how many of the readers are coming over here.

Anonymous said...

"In his case, he actively, doggedly seeks out social interaction. He calls many acquaintances 'friends,' but I am sure I'm categorizing the relationships more accurately. He is somewhat dependent on the kind of social interaction where others have no real choice but to allow his participation - church groups, volunteering at nonprofits, etc."

Anonymous, how likely is it that your friend has been subjected to therapists or teachers who have told him that his social problems are all in his imagination, or a result of his own insecurities, and that he should just "put himself out there," or "keep trying?" And so he keeps trying, being told that he's doing the right thing, despite continuous lack of satisfactory results?

Because I had several people like that in my life. Eventually I learned to ignore everything that came out of their mouths.

Don't underestimate how much it hurts to desperately want to have friends, to try and try and try to do everything right, and be continuously treated as an annoyance or an obligation.

Bernard (ben) Tremblay said...

Is that explanation rather ... mmmmmm ... I want to say teleological. Neurons firing because of this or that intention? It's as though we're anthropomorphizing neuronal activity.

Neurons fire because of the chemical conditions in their immediate environment.

Sue Gerrard said...

You’ve questioned a fundamental assumption of the ‘broken mirrors’ model. But there appear to be some other assumptions implicit in your alternative hypothesis that mean it might not hold water for all cases.

Here are the assumptions I identified and why I would question them:

1. Everybody diagnosed with autism has the same underlying condition.

Technically, ‘autism’ refers to a set of signs and symptoms, not to the cause of those signs and symptoms. The two uses are often conflated. The evidence suggests that ‘autism’ is a set of emergent outcomes from a wide range of different causes. Autism is notorious for its variation across individuals.

2. Everybody diagnosed with autism is hypersensitive to sensory stimuli in general.

Rogers and Ozonoff’s 2005 review paper found that there is no pattern of sensory sensitivity that is characteristic of people diagnosed with autism. The evidence suggests that autistic people often show different levels of sensitivity in different facets of different sensory modalities. For example, a common issue in young autistic children is hyposensitivity to pain coupled with hypersensitivity to light touch. Since pain and touch are mediated by different types of receptors, there’s a mechanism for that phenomenon.

3. Autistic people are hypersensitive to emotional states. This explains reduced gaze-fixation in autistic people.

The evidence suggests that some autistic people have trouble getting emotional information from faces at all. And with identifying emotions - their own and other people’s. (Up to 3% of the general population have problems with facial information.) Some autistic people report deliberately using peripheral vision because they find eye contact too intense. They also report finding foveal fixation uncomfortable per se – whether on faces or cluttered rooms or any other complex visual stimulus - so an equally plausible explanation for gaze-avoidance is that some autistic people have abnormally large foveal avascular zones and suffer from too much detailed visual information if they use foveal fixations. It’s also possible that if auditory processing difficulties are involved, ‘gaze-avoidance’ is actually due to focusing on the lips to get visual cues to support speech monitoring.

In short, your hypothesis could explain some people’s autism, but given the wide variation between individuals, is unlikely to explain every case.