Thursday, August 13, 2009

A new study claims to have identified mirror neurons in the human brain

In case you haven't seen it yet, there is a new paper in J. Neuroscience that reports the existence of mirror neurons in human inferior frontal gyrus (~Broca's area). It used a repetition suppression fMRI paradigm and found a suppression effect (different > same) both when subjects executed and then observed the same action and when subjects observed and then executed the same action. This appears to be the best evidence yet for the existence of mirror neurons in humans: an effect was found in both directions, execute-->observe & observe-->execute, and it is showing up in the right place, Broca's area, the presumed human homologue of monkey area F5. Another thing I like about this study is that it used object directed actions rather than pantomime which makes it more comparable to the monkey studies.

I haven't yet read the paper closely. I'd be very interested to hear what folks think, so please post a comment. My only general concern is with the repetition suppression effect itself. We've used it previously and it strikes me as a bit on the sketchy side. The effects are very subtle and susceptible to the boredom critique: the suppression effect is not neural adaptation but instead reflects the fact that the subject is bored with the repetition of stimuli and therefore allocates less attention for same trials versus different trials. I'm not sure this critique applies here though given that the same trials cross modality. I'll have to think about that one.


Kilner, J., Neal, A., Weiskopf, N., Friston, K., & Frith, C. (2009). Evidence of Mirror Neurons in Human Inferior Frontal Gyrus Journal of Neuroscience, 29 (32), 10153-10159 DOI: 10.1523/JNEUROSCI.2668-09.2009


Greig de Zubicaray said...

Interesting. The peak is actually in the left rolandic operculum, lesions to which tend to produce anarthria without a comprehension deficit (admittedly, my memory is a little hazy on this, so feel free to correct). Note Kilner et al.'s comments in the methods section where they considered confining a search to BA44 to be inappropriate, citing Morin and Grezes' (2008) review. Petrides (2005) concluded that the homologue of F5 is BA44 on cytoarchitectonic grounds and makes it clear that this area is immediately anterior to ventral BA6, not part of it. The cited Morin & Grezes paper does not review cytoarchitectonic studies, only those involving functional imaging.

James Kilner said...

I just thought I would give a quick reply to Greig's comment. Greig is correct that the peak of the group level SPM lies more posterior than the probabilistic maps defining BA44. However, the peaks for the individual subjects, whether the conjunction analysis or the xmRS effect, lie more anterior to the peak for the classical second level analysis - within the area defined probabilistically as BA44 (Figure 2 and Tables 1&2).

Greg Hickok said...

No matter where they are, it would surprise me if mirror neurons didn't exist in humans. After all, they are just sensorimotor cells like so-called "canonical cells" in F5 that respond both to the perception of objects and the execution of actions toward objects. Similar sensorimotor cells have been noted in monkey IPS for years. The more important question is what are they doing. Greig's point that damage to the supposed mirror neuron regions (whether they are BA44 or BA6) do not seem to cause receptive deficits is key. The mirror neuron circuit just plain isn't the "basis" of action understanding.

Greig de Zubicaray said...

James, like Greg I think it is more likely than not that mirror neurons exist in humans, and the group response appears reliable (just not in an area associated with comprehension deficits). However, of the individual subject peaks reported in Table 2 for the important xmRS effect, only 4 are significant at the p < .001 (uncorrected) threshold and the caption for Figure 2 makes it clear that an extraordinarily lenient p < .05 (uncorrected) threshold was used for viewing this single subject data. The curmudgeon in me thinks that corrected thresholds are useful for avoiding false positives.

Anonymous said...

I think this blog's authors are correct in being skeptical of the study's authors use of suppression effect for even more reasons than they or the commentors cite. The study is trying to identify mirror neurons (plural only in referring to a number of such neurons in the aggregate) as neurons which are somehow differentiated in their function such that they serve as both receiver and transmitter of a specific signal. Aside from that violating unidirectionality of transmission within one neuron, it would of necessity 'lose' the information embodied in that signal by virtue of its relation to other neurons. Even if the definition is modified to a small group of neurons working as 'a or one' mirror neuron, there's still a problem with respect to specific mirror neuron/s then being tied to the specific information encoded by 'observing' any particular action at some point for it to be doing anything other than, or in addition to, what non-mirror neurons do already. I hate using computer analogies but it would take the equivalent of an entire mirror drive or raid array in the brain for that to be the case.