“No evidence of Human Mirror Neurons”

That claim can be found in the latest issue of the Journal of Neurosciences.

If I were a sociologist of science, I would jump on mirror neurons – they are the perfect object if you want to study a scientific controversy today.

On the one hand, Mirror Neurons, found in several regions of the macaque cortex, have been hypothesized, by mainstream cognitive scientists, to underlie language [1], theory of mind, culture [2], empathy [3], art, social cognition in general [4], the success of an advertisement campaign promoting a famous brand of underpants (as Jonah Lehrer [5] noted: “You know that mirror neurons have jumped off the shark when they’re used to explain Abercrombie and Fitch”), and many other things, that have only one thing in common: they are extremely discreet, not to say absent, in the only species we are 100% sure has mirror neurons – macaques.

On the other hand, in a piece published in Current Biology [6] based on data published in the Journal of Neurosciences [7], I. Dinstein suggested that they can hardly be found in humans, and J. Neurosciences editors seem to concur. To understand how that is possible, we need to step a decade back in time.

The story of human mirror neurons is confused by frequent changes in definition. In the first papers published about them, most “mirror” neurons found in the macaque brain are merely neurons who have visual properties – they “fire” when the macaque sees movements although they are situated in motor areas. In addition to these visual properties, many of them also fire when the animal performs a movement – as one would expect in a motor area.

Now, a small percentage of these neurons investigated in the first studies also displayed a variable degree of congruence between their visual properties and their motor properties. In other words, they fired when the animal performed a gesture and when it saw the same gesture being performed. Few people cared to remark that this property of high congruence, from which “Mirror” neurons got their name, concerned only a tiny minority of actual macaque mirror neurons.

Yet that property of high congruence – neurons firing both when the animal performs an action and when he sees the same action being performed – came to define mirror neurons in subsequent works. Whenever a neuron or a brain area with high congruence was found, you could bet that someone was going to call it a mirror neuron or a mirror area, even if it was not a motor neuron at all. For example, a neuron that reacted both to pain a human subject experienced and to pain he saw others experiencing was regarded as a human mirror neuron. Likewise, a brain area that responded both to disgust felt in oneself and to disgust perceived in others was a “mirror” area. In that same vein, mirror areas for touch, auditive mirror neurons, etc. were found in humans and macaques.

But high congruence was also taken for granted when it was only one way among many others of interpreting neuroimaging data. Neuroimaging studies showed human brain areas lighting up both when subjects performed an action and when they viewed an action (not necessarily the same). The macaque data had shown that mirror brain areas contained a majority of motor neurons, visual neurons, and non-congruent visuo-motor neurons. These neurons could perfectly explain the observed activations without appealing to highly-congruent, full-blown Mirror neurons. This interpretation was all the more warranted since the same brain areas could light up when the subjects saw a different gesture from the one they performed, or when the putative mirror activity only showed up when contrasted with visual noise.

A few studies tried to show that “mirror” brain areas – meaning those human brain areas that light up both when perceiving an action and when performing one – were indeed selective for some aspects of the action seen and performed – for example, that some brain areas react both when the subject sees an action and when he performs one, but only if the action is performed with a foot. But this was still a long way from proving that these activations were due to highly congruent mirror neurons.

Why were highly congruent visuomotor neurons favoured as the mainstream interpretation of these findings? Because an extremely influential theory demanded it. The simulation theory, also known as direct-matching theory, holds that (highly congruent) mirror neurons underlie our understanding of other people’s intentions. When we see someone doing something, the story goes, we automatically activate the cortical circuit that allow us to do the very same thing. In other words, we covertly imitate – we “mirror” – everything we see others do, and that is what allows us to understand what mental processes underlie the actions we see them do. Of course, if mirror processes happened not to mirror anything at all – if they were merely provoked by visuomotor, motor and visual neurons that have very little selectivity and even less congruence – the theory would tumble down.

In a very insightful 2005 online paper [8], Gergely Csibra wondered why, if mirroring actions was the function of mirror neurons, most mirror neurons found in the first studies were not mirror at all – the vast majority was either merely visual neurons, or they were visuomotor, but not congruent. He went on to suggest that most “mirror” activations in humans did not reflect mirroring at all, but rather, anticipated motor responses to the actions the subject observed. It is nice to see that, although Csibra’s paper could not be published in a peer-reviewed journal at the time, Dinstein’s Current Biology piece retains many an argument that Csibra used.

I’ll let fMRI geeks appreciate Dinstein’s very clever methodology, that allows him to use fMRI to show real mirror activity, that is, mirror activity due to highly congruent visuomotor neurons only. This method had been used by two other studies who found some mirror activity, but in somewhat unexpected places. Dinstein and his colleagues found none. They are extremely careful not to overstate this negative result – way more careful than the editors of the Journal of Neurosciences. But the mere fact that a negative result, accompanied by open criticisms of the methodology used in previous fMRI studies of mirror neurons, gets so much publicity in two major journals, is a sign that the scientific community is becoming suspicious. After all, positive evidence for mirror activity got published the same day, also in Current Biology – but this time, the critical view of mirror neuron stole the light from these results.

A tentative explanation of the rise – and fall ? – of mirror neurons

Why now? The frailty of the evidence for human mirror neurons was evident from the start. I think part of the reason why a backlash may be under way, is the increasingly far-fetched speculations surrounding mirror-neurons. Mirror-neurons theories of autism and mindreading were already difficult to swallow (see [9] for a critical review), but the recent mirror theories of art might have been the last straw. This, I suspect, is a constant factor in the fate of mirror neurons: their rise and fall was determined by the use that could be made of them in fields related to philosophy and the social sciences – social cognition, imitation, culture, art.

As one may see in the first papers, the study of macaque visuomotor neurons might have taken a very different road. One may have interpreted their activity as reflecting a preparation for action, as Csibra did. Or the time and the resources that were devoted to mirror neurons might have gone instead to canonical neurons, neurons found in similar areas that react to the view of an object, but not of an action. Canonical neurons were just as sexy as mirror neurons from a neurophysiological point of view. They had nothing to do with Empathy, Society and Togetherness; but they displayed amazing visual properties, that could change the way we thought about object perception, and the way our motor system is involved in it.

The direct matching theory, that states that the chief function of these neurons was to make social life possible, changed everything – I argued above that it changed the very definition of mirror neurons. Even though there was no clear-cut evidence for human mirror neuron activity – and for a long time the evidence was even weaker than it is now – the theory was one major motivation to continue the search. In the heydays of mirror neuron speculation (the late- 1990s and early 2000s), the field of social cognition was booming, with links between autism and theory of mind being feverishly explored. In an increasingly antiscientific climate, major journals, widely accused of being too dry, and losing readers and suscribers, were feeling the need for more sexy topics – results that talked not only of BOLD signals and p values, but also of empathy, bonding, identification, and other things we think make us human. Had it not been supported by the appeal of speculations about social life, by many philosophers and social psychologists, and all the images conjured by the beautiful (and woefully misleading) label “mirror neuron”, I suspect the results would have been seen with much less indulgence.

Social scientists are often concerned about the importation of poorly understood concepts from the hard sciences into “softer” fields like anthropology or sociology. In my view, the story of mirror neurons shows that the reverse also happens: scores of brilliant neuroscientists were set to work on entities celebrated and, one might say, partly invented because they allowed neuroscience to relate to social concepts, like imitation, culture, and art. Once it got started, the story was one of mutual seduction. The disenchantment, if it takes place at all, may take years. Will the speculative bubble burst?


References

[1] Rizzolatti, G., & Arbib, M. A. (1998). Language within our grasp. Trends in neurosciences21(5), 188-194.

[2] “Ramachandran, V. S. (2000).

[3] Viggiani, E., & Watson, G. (Eds.). (2010). Friends and Foes Volume II: Friendship and Conflict from Social and Political Perspectives (Vol. 2). Cambridge Scholars Publishing.

[4] Hurley, S. (2008). The shared circuits model (SCM): How control, mirroring, and simulation can enable imitation, deliberation, and mindreading. Behavioral and brain sciences31(1), 1-22.

[5] “Buyer, Beware” by Jonah Lehrer

[6] Dinstein, I. (2008). Human cortex: reflections of mirror neurons. Current Biology, 18(20), R956-R959.

[7] Dinstein, I., Gardner, J. L., Jazayeri, M., & Heeger, D. J. (2008). Executed and observed movements have different distributed representations in human aIPS. Journal of Neuroscience28(44), 11231-11239.

[8] Csibra, G. (2005). Mirror neurons and action understanding: is simulation involved?. Interdisciplines.

[9] Southgate, V., & Hamilton, A. F. D. C. (2008). Unbroken mirrors: Challenging a theory of autism. Trends in cognitive sciences12(6), 225-229.

7 Comments

  • comment-avatar
    guest guest 5 November 2008 (00:12)

    I was not previously aware that the mirror neuron finding was so tenuous. But, that’s not surprising since I am not a scientist and get a lot of information from journalists and ”pop” science writers. I will be cautious in my use of that idea in my philosophical writing and thinking. — Enjoy, Don Berg Site: http://www.Teach-Kids-Attitude-1st.com Blog: blog.Attitutor.com

  • comment-avatar
    guest guest 5 November 2008 (05:57)

    Hello, I subscribe to your blog and another, and was surprised to see both of you posting contrasting stories on my feed, just a few hours apart: http://talkingbrains.blogspot.com/2008/11/mirror-neurons-in-humans-revealed-by.html

  • comment-avatar
    Olivier Morin 5 November 2008 (09:19)

    Yes, Hazel, two studies have been published at the same time, using roughly the same methods and reporting quite different results: Chong et al. in Current Biology, Dinstein et al. in J. Neurosciences. You can find a good (if short) discussion of the differences between the two papers in Dinstein’s Current Biology piece (a third article, not a results article, published in the same CB issue as Chong’s paper…).

  • comment-avatar
    Dominique Baillot 6 November 2008 (21:03)

    Hey Olivier, From this point of view, it almost seems that we are stuck in a sort of ’bipolar attractor’ (one of the pole being ”hard sciences” – I guess the one that uses mathematics – and the soft sciences – the one that uses words -) whose movements are directed by fashion and economic interests. Oops, am I doing hard sciences again here ?!! interesting story anyway.

  • comment-avatar
    Ilkka Pyysiäinen 12 November 2008 (07:45)

    I don’t know what might be the consquences of this, but I heard from a colleague that Csibra — to whom this feature refers – had said in a workshop in Turin a month ago that he had originally misunderstood the idea of direct matching in monkey mirror neurons.

  • comment-avatar
    Gergely Csibra 13 November 2008 (16:00)

    What a nice example of propagation of information without context! At that workshop we heard that the idea of direct matching simply means that the goal of the observed action is represented in the motor system, and not that this representation is generated without the intervention of non-motor processes. In other words, we were told that ”direct matching” and ”resonance” are not intended to refer to the process by which action observation generates motor activation, but to the result of such a process. If this is the case, I indeed misunderstood the idea, but I guess, I was not the only one. This clarification of what direct matching means was provided by Corrado Sinigaglia, and was endorsed by Giacomo Rizzolatti and Christian Keysers.

  • comment-avatar
    Olivier Morin 13 January 2009 (16:45)

    I’ve added the references and changed the link to the J Neurosciences issue.