Greg Detre
Monday, March 03, 2003
Mirror neurons (MNs) fire both when the monkey sees an action performed, and when it performs the same actions. They fail to respond to just the sight of the object or of the agent alone, nor to mimicking the action without the object or with tools.
Gallese and Goldmain claim that there�s good evidence to suppose that a similar execution/observation matching system exists in humans. I didn�t fully understand their description of Fadiga et al.�s Transcranic Magnetic Stimulation study. It showed that while observing objects being grasped, people�s motor evoked potentials in hand muscles increase during TMS. As far as I can tell, they are arguing that this is because the TMS stimulates the motor cortex, the MNs are responsible for increased activity in the premotor cortex, and in conjunction these responses affect the MEPs in the hand muscles. Two further PET studies have shown the left superior temporal sulcus (Brodmann�s area 21), the left inferior parietal lobule (Brodmann�s area 40) and the anterior part of Broca�s region (Brodmann�s area 45) to be particularly activated when observing being grasped rather than just the objects. This is one source of the argument for F5 being a homologue to Broca�s area.
Gallese & Goldman argue against MNs� role in learning by imitation, but instead as being implicated in mind-reading (aka �folk psychologising�), the activity of representing specific mental states of others. They discuss the difference between the theory theory and simulation theory of mind. Theory theory considers that people mind-read by creating theoretical posits of others� mental states in order to explain their behaviour, in the same sort of way �that physicists appeal to electrons and quarks to predict and explain observable phenomena�. The simulation theorists argue that humans� mind-reading abilities are based on the ability to simulate the mental states of others. In other words, we are able to predict or retrodict another person�s mental states by putting ourselves in another�s shoes and simulating what mental state we would be in.
If the brain is utilising the same machinery for making one�s own decisions as for predicting/retrodicting others� mental states offline (as per the simulation theory), then finding neurological evidence for mental mimickry in the mind-reading process would support the ST over the TT. They postulate that externally-generated activity in the MNs plays this simulation role, or at least some primitive version of it in monkeys. They consider Fadiga et al.�s TMS study to provide further support for the ST, since it showed that downstream motor activity in the observer is facilitated by observation of that motor activity in the target. This result is unexplained by the TT, and implies that the simulation activity is not completely offline in humans. That there exists a group of prefrontal-lesioned compulsive imitators could be neatly explained by a lack of downstream inhibition of this partially-online simulation process.
If MNs are the neural correlate of a �cognitive continuity� of intentional-state attribution, then we might expect that MNs in humans have been coopted into doing something more complicated than they do in monkeys. I�m not sure quite what this might be. I was surprised to find that all mention of MNs focuses on hand- and mouth-related actions, so perhaps human MNs are able to relate a wider range of body-parts at the action-level, as well as more abstract motor programs at the program-level. After all, there is evidence that when writing one�s signature, certain shape characteristics are retained even when using one�s big toe, or when writing on the blackboard instead of a piece of paper, even though fMRI studies indicate that the movement parameters are stored in secondary sensorimotor cortices of the dominant hand (Rijntjies et al. 1999).
With regard to the theory theory vs simulation theory debate, I had one major query. As I understand it, theory theorists are claiming that the theoretical posits we produce about about other people�s beliefs and desires are like scientific hypotheses to best explain the data of their behaviour, without being produced by simulating what they themselves would feel, believe, or desire in a similar position. If this was really the case, then wouldn�t we occasionally expect the beliefs and desires that people attribute to others to differ in kind from the beliefs and desires they understand themselves to have? After all, the only reason that I attribute beliefs and desires like mine to other people is because I assume that we are fundamentally alike. If, in TT, the beliefs and desires are really only theoretical posits, wouldn�t there be a propensity to imagine other people to have strange and alien mental mechanisms generating their behaviour? If theory theorists were to respond by saying that we have a kind of censor that restricts our hypotheses to those akin to our own mental machinery, this makes their theory more ugly, and more similar to simulation theory than they might like to admit.
They characterise the autistic spectrum disorders in terms of impairments in social interaction, imaginative ability and repetitive and restricted patterns of behaviour. Autism has an onset before 3 years and is associated with delayed/abnormal language development. Baron-Cohen et al. suggested that autism may result from a deficit or delay in development of theory of mind. Williams et al. give three good reasons for thinking that this cannot be quite the right or whole story:
Rogers and Pennington first suggested that deficits in imitation might play this precursor role in the development of autism, and be able to answer at least the first two of the above objections. They suggested that the root of autism may be �impaired formation/co-ordination of specific self-other representations� resulting from impaired imitation. After all, intuition and the attribution of mental state both �involve translating from the perspective of another individual to oneself�. The link between the two is even closer for the simulation theory. We discussed Meltzoff and Gopnik�s hypothesis that a new-born�s capacity for translating and imitating the seen behaviour (facial expressions) of others might underlie emotional mind-reading. Moreover, the 21 experimental studies of the imitative competence of individuals with autism seem to provide fairly strong evidence of impairment.
The rest of the paper revolves around the idea that MNs could be neural correlates of our imitative abilities, and that their early impairment might be related to the development of autism. This is evidenced by the observation of imitative dyspraxia in patients with left frontal lobe lesions, who are unable to repeat actions performed by others, or to replicate the gestures on a� mannequin. This makes sense, given MNs� purported locations in left Broca�s area in humans (as well as in the premotor and parietal cortex).
Rizzolatti and Arbib have suggested that MNs may have evolved to subserve speech in humans by building upon a prelinguistic grammar of actions in the primate brain. Few details of how this might work are given in this paper, but it seems a little implausible at first hearing. However, vervets have a roughly 4-signal verbal vocabulary, and bonobos chatter amongst each other in large groups for hours at a time. I would want to see how well Rizzolatti & Arbib�s suggestion fits with neural activity during these kinds of pre-linguistic behaviour.
It�s fairly easy to see how MNs could be precursors to full ToM, e.g. in retrodicting others� mental states by reconstituting an executive plan to perform an action like the one being watched. Similarly, emotional contagion and emotional mind-reading may be related (cf Meltzoff & Gopnik). Although Williams et al. mention that shared attention may also be crucial to the development of a full ToM, the case for implicating MNs here is surely much weaker.
Finally, some impairment or delay in the development of (some of the) mirror neurons may be heavily implicated in autism. Failure to develop self-other representations may result in failure to develop reciprocal social abilities: e.g. shared/joint attention, gestural recognition, language (especially social/pragmatic aspects), breakdowns in the development of empathy/ToM. They explain the repetitive/inflexible/stereotyped behaviour as something to do with perception-action linkage, and because the child may have failed to learn some of their high-level executive functions (also apparently impaired in autism) from others (program-level imitation). As I argued briefly towards the end of the earlier critique, I�m open to the idea that MNs could play a crucial developmental role in ToM (and probably autism), I can�t help but feel that human MNs must be involved in more wide-ranging, multi-modal and abstract functions than simply the low-level gesture matching that has been observed in single-unit recording in monkeys.
�