Mirror neurons aren’t really all that bad…
I’ve been too busy writing manuscripts and teaching to blog much lately, but since The Evil Monkey over at Neurotopia insists everyone should post about mirror neurons, I feel compelled to oblige. Mirror neurons seem to be getting the short end of the stick, so I’ll play devil’s advocate and try to convince you they’re not all that bad (though I agree they’re probably not “the DNA of psychology”!)
The posts up at Mixing Memory, Frontal Cortex, and Neurotopia express several different concerns about mirror neurons. One is the standard “they’re getting a disproportionate amount of attention” complaint. I agree with this worry to an extent–so far mirror neurons haven’t miraculously solved any of cognitive science’s biggest problems–but excitement of this sort accompanies quite a few major discoveries in the cognitive neurosciences. And the existence and localization of mirror neurons is without a doubt a major discovery. For one thing, mirror neurons offer what is clearly the most plausible current model of imitative behavior, which isn’t a trivial matter, since imitation turns out to be pretty rare in the animal kingdom. For another, it wasn’t all that long ago that people were pretty skeptical about the prospect of knowledge being grounded in perceptual-motor processing. While we’re nowhere near grounding concepts like justice and mercy in monkeys’ left elbow joints, these days no one seriously doubts that at least some kinds of concepts are likely to be pretty intimately related to perceptual and motor representations. Mirror neurons provide an elegant way to study those kinds of representations.
Another kind of worry about mirror neurons is the potential lack of generalizability of monkey findings to humans. Well, it’s true that there haven’t been any electrophysiological studies of the human mirror system to date. But there’s any amount of indirect evidence supporting a system in humans that’s (a) located in brain regions homologous with regions F4 and F5 in monkeys (i.e., Broca’s area and adjoining premotor regions), (b) displays virtually all of the functional characteristics displayed in monkeys, and (c) displays additional functionality that appears to be unique to humans and can plausibly be related to the more basic functionality already present in monkeys. Given the existing evidence, it’s vastly more parsimonious to suppose that what fMRI and EEG studies are picking up on is something very much like what’s been observed electrophysiologically in monkeys, and not some highly dissimilar system.
A third concern is that it’s a leap to go from “mirror neurons represent actions” to “mirror neurons represent meaning”. I’m not sure this is the case; I personally find the evidence that mirror neurons in monkeys specifically code for the meaning of actions pretty compelling. Mirror neurons are often described as being sensitive to both the visual presentation of an action and the commission of that same action by the monkey itself. That’s true, but it’s only part of the story. It turns out that a large minority of mirror neurons respond not only to visual observation of an action, but also to associated sounds, even when the monkey can’t see the action being performed. Now I don’t know what definition of ‘meaning’ mirror neurons need to satisfy, but I feel pretty comfortable with one that accords it to neurons that respond selectively to polymodal presentation of a specific action.
Along the same lines, what’s equally striking is that a given visual stimulus can elicit very different patterns of activation in mirror neurons depending on the context it’s presented in. For example, ahand reaching behind a screen for an object that the monkey knows is there (but can’t directly see) might increase firing in a given neuron, but the same neuron may fail to respond at all when the hand reaches behind the screen and there’s no object present. Is the neuron representing the meaning of the action? Well, why not? It’s clearly tracking something like grasping, as opposed to something like hand moving in particular direction at particular orientation.
Finally, there seems to be some pretty serious skepticism about the mirror neuron-language evolution theory proposed by Rizzolatti and colleagues. I don’t have much to say here, because (a) I don’t know very much about this debate, and don’t feel qualified to evaluate Rizzolatti’s claims; and (b) based on what little I do know, it does seem like the least convincing aspect of Rizzolatti’s research program. That said, I’m not sure it’s fair to subject Rizzolatti and Arbib’s (1998) paper to scrutiny in 2006 without considering the body of work accumulated since then. A more recent and comprehensive review of the mirror neuron system (including relation to language) can be found in Rizzolatti’s recent (2004) Annual Review of Neuroscience chapter. While it certainly doesn’t come close to providing a fully-worked out model of either the neuron system or its connection specifically to language, the views Rizzolatti’s been proposing most recently at least don’t strike me as being absurd on their face. And the payoff will be very large if he’s even partly right. These are still early days, but I think it’s already pretty clear that there’s a large and potentially very important ongoing research program into the structure and function of the mirror neuron system.