Mixing Memory brings up some excellent points regarding mirror neurons in primates, and Frontal Cortex follows up with his thoughts. To both of them I say “bravo, but your skepticism probably doesn’t go far enough”.

We give Rizzolatti et al too much credit with their conclusions. After all, they’ve only demonstrated the existence of mirror neurons in monkeys. Due to the obvious inherent difficulties associated with recording from human neurons in vivo, no one has yet (to my knowledge) published anything that demonstrates the existence of mirror neurons in people. Instead, we stick people in scanners and infer that they have mirror regions, or mirror neural systems, that are at least in part composed of mirror neurons. These regions are associated with language and imitation, but any evidence that mirror neurons are involved with either behavior in humans is circumstantial at best.

In fact, there is nothing in the human literature to suggest that mirror neurons are required for imitative behaviors. It is possible that population-based mirror function is much, much more efficient for this task than are individual mirror neurons, and indeed such a diffuse mirror system could function without any mirror neurons at all, at least in principle, as long as the same information is carried in parallel and not able to be generalized to any particular cell. Is this likely? I doubt it. Mirror activity in different brain regions is probably driven by populations of mirror neurons. So I will caveat my caveat by stating my personal opinion; mirror neurons likely exist in humans, and honestly I would be very surprised if they didn’t. However, certainty is the perfect barrier to learning, and if we remain convinced of their existence without hard evidence then we run the risk of creating faulty models.

So the underlying question remains: what do mirror neurons actually do? Are they merely a system for recognizing actions? Do they allow an organism to infer intent or mental state? Do they mediate imitation? Perhaps they represent a dynamic system; while their basest function may merely be an action recognition system, they might be periodically recruited to perform other functions such as mental state inference, imitative learning, and yes even learning language. Such a phenomenon would not be unheard of. For instance, prefrontal cortex is known to actively recruit the attentional capacities of the basal forebrain cholinergic system when pressed to perform cognitively demanding tasks.

With all that in mind, how do we study mirror neuron-like activity in humans? Primarily we rely on electroencephalography, or EEG, which records electrical fields through the placement of electrodes on the skull, and the dreaded fMRI, which has been dissected to death here on SEED. The advantages of EEG are that it actually measures electrical field potentials produced by populations of neurons in a very noninvasive fashion. The drawback is that spatial resolution is shit. fMRI allows for better spatial resolution but by using it, one runs the risk of having electrical signal decoupled from blood flow and thus not providing an accurate index of functional changes.

Autism Spectrum Disorders are a frequent target for investigating mirror functions in humans. Those with ASDs are thought to lack empathy, a theory of mind, and to be impaired in their ability to imitate and relate socially. Obviously ASD researchers see mirror neurons as the logical place to start, and with all the hype I can’t say I blame them. One group of ASD researchers decided to see if patterns of brain activity were altered in ASD patients. Specifically, they examined alterations of mu spectrum activity, which is known to be decreased in mirror neuron system brain areas upon the observation of biologically relevant motions. For example:

The three light gray bars represent mu activity from different scalp regions following the observation of a bouncing ball. The medium gray bars are activity from observing another person’s hand motion. The dark gray bars are observation of the subject’s own hand in motion. As we would expect, control subjects show suppression when observing another’s hand motion or observing their own hand in motion, while ASD subjects only show suppression when viewing the motion of their own hand.

Similarly, when using fMRI to study activation patterns in people asked to imitate the facial expressions of others, control subjects (a) show markedly more activation than do ASD subjects (b). (C) represents areas where the differences in activation were significantly greater in controls, mainly the pars opercularis. It should be pointed out that the pars opercularis is part of Broca’s Area.

So what do we learn here? Admittedly, not much. We learn about the neural networks involved in mirror system activity, and the anatomical areas involved. But we aren’t studying mirror neurons themselves. Likewise, it is easy to be suckered into pointing at the activity patterns and involvement of language-oriented areas like the pars opercularis, and claim that mirror neurons are somehow involved with a slew of exotic hypotheses. Hell, they might even be correct! These hypotheses are very interesting and may drive some fascinating, satisfying research in the future, but we must remember that correlation is not causation, especially when evidence for the cause itself is lacking in humans. Even when the logic is sound, the argument can still be false if the premises do not comport with reality.

References

Dapretto M, Davies MS, Pfeifer JH, Scott AA, Sigman M, Bookheimer SY, Iacoboni M. Nat Neurosci. 2006 Jan;9(1):28-30. Epub 2005 Dec 4.

Oberman LM, Hubbard EM, McCleery JP, Altschuler EL, Ramachandran VS, Pineda JA. Brain Res Cogn Brain Res. 2005 Jul;24(2):190-8.

Oztop E, Kawato M, Arbib M. Neural Netw. 2006 Apr;19(3):254-71. Epub 2006 Apr 3.