The Tell-Tale Brain

and intentions and to predict and influence their behavior. Calling it a theory can be a little misleading, since the word “theory” is normally used to refer to an intellectual system of statements and predictions, rather than in this sense, where it refers to an innate, intuitive mental faculty. But that is the term my field uses, so that is the term I will use here. Most people do not appreciate just how complex and, frankly, miraculous it is that they possess a theory of mind. It seems as natural, as immediate, and as simple as looking and seeing. But as we saw in Chapter 2, the ability to see is actually a very complicated process that engages a widespread network of brain regions. Our species’ highly sophisticated theory of mind is one of the most unique and powerful faculties of the human brain.

Our theory-of-mind ability apparently does not rely on our general intelligence—the rational intelligence you use to reason, to draw inferences, to combine facts, and so forth—but on a specialized set of brain mechanisms that evolved to endow us with our equally important degree of social intelligence. The idea that there might be specialized circuitry for social cognition was first suggested by psychologist Nick Humphrey and primatologist David Premack in the 1970s, and it now has a great deal of empirical support. So Frith’s hunch about autism and theory of mind was compelling: Perhaps autistic children’s profound deficits in social interactions stem from their theory-of-mind circuitry being somehow compromised. This idea is undoubtedly on the right track, but if you think about it, saying that autistic children cannot interact socially because they have a deficient theory of mind doesn’t go very far beyond restating the observed symptoms. It’s a good starting point, but what is really needed is to identify brain systems whose known functions match those that are deranged in autism.

Many brain-imaging studies have been conducted on children with autism, some pioneered by Eric Courchesne. It has been noted, for example, that children with autism have larger brains with enlarged ventricles (cavities in the brain). The same group of researchers has also noted striking changes in the cerebellum. These are intriguing observations that will surely have to be accounted for when we have a clearer understanding of autism. But they do not explain the symptoms that characterize the disorder. In children with damage to the cerebellum due to other organic diseases, one sees very characteristic symptoms, such as intention tremor (when the patient attempts to touch his nose, the hand begins to oscillate wildly), nystagmus (jerky eye movements), and ataxia (swaggering gait). None of these symptoms are typical of autism. Conversely, symptoms typical of autism (such as lack of empathy and social skills) are never seen in cerebellar disease. One reason for this might be that the cerebellar changes observed in autistic children may be the unrelated side effects of abnormal genes whose other effects are the true causes of autism. If so, what might these other effects be? What’s needed, if we wish to explain autism, is candidate neural structures in the brain whose specific functions precisely match the particular symptoms that are unique to autism.

The clue comes from mirror neurons. In the late 1990s it occurred to my colleagues and me that these neurons provided precisely the candidate neural mechanism we were looking for. You can refer back to the previous chapter if you want a refresher, but suffice it to say, the discovery of mirror neurons was significant because they are essentially a network of mind-reading cells within the brain. They provided the missing physiological basis for certain high-level abilities that had long been challenging for neuroscientists to explain. We were struck by the fact that it is precisely these presumed functions of mirror neurons—such as empathy, intention-reading, mimicry, pretend play, and language learning—that are dysfunctional in autism.¹ (All of these activities require adopting the other’s point of view—even if the other is imaginary—as in pretend play or enjoying action figures.) You can make two columns side by side, one for the known characteristics of mirror neurons and one for the clinical symptoms of autism, and there is an almost precise match. It seemed reasonable, therefore, to suggest that the main cause of autism is a dysfunctional mirror-neuron system. The hypothesis has the advantage of explaining many seemingly unrelated symptoms in terms of a single cause.

It might seem quixotic to suppose that there could be a single cause behind such a complex disorder, but we have to bear in mind that multiple effects do not necessarily imply multiple causes. Consider diabetes. Its manifestations are numerous and varied: polyuria (excessive urination), polydypsia (incessant thirst), polyphagia (increased appetite), weight loss, kidney disorders, ocular changes, nerve damage, gangrene, plus quite a few others. But underlying this miscellany is something relatively simple: either insulin deficiency or fewer insulin receptors on cell surfaces. Of course the disease is not simple at all. There are a lot of complex ins and outs; there are numerous environmental, genetic, and behavioral effects in play. But in the big picture, it comes down to insulin or insulin receptors. Analogously, our suggestion was that in the big picture the main cause of autism is a disturbed mirror-neuron system.

ANDREW WHITTEN’S GROUP in Scotland made this proposal at about the same time ours did, but the first experimental evidence for it came from our lab working in collaboration with researchers Eric Altschuler and Jaime Pineda here at UC San Diego. We needed a way to eavesdrop on mirror-neuron activity noninvasively, without opening the children’s skulls and inserting electrodes. Fortunately, we found there was an easy way to do this using EEG (electroencephalography), which uses a grid of electrodes placed on the scalp to pick up brain waves. Long before CT scans and MRIs, EEG was the very first brain-imaging technology invented by humans. It was pioneered in the early twentieth century, and has been in clinical use since the 1940s. As the brain hums along in various states—awake, asleep, alert, drowsy, daydreaming, focused, and so on—it generates tell-tale patterns of electrical brain waves at different frequencies. It had been known for over half a century that, as mentioned in Chapter 4, one particular brain wave, the mu wave, is suppressed anytime a person makes a volitional movement, even a simple movement like opening and closing the fingers. It was subsequently discovered that mu-wave suppression also occurs when a person watches another person performing the same movement. We therefore suggested that mu-wave suppression might provide a simple, inexpensive, and noninvasive probe for monitoring mirror-neuron activity.

We ran a pilot experiment with a medium-functioning autistic child, Justin, to see if it would work. (Very young low-functioning children did not participate in this pilot study as we wanted to confirm that any difference between normal and autistic mirror-neuron activity that we found was not due to problems in attention, understanding instructions, or a general effect of mental retardation.) Justin had been referred to us by a local support group created to promote the welfare of local children with autism. Like Steven, he displayed many of the characteristic symptoms of autism but was able to follow simple instructions such as “look at the screen” and was not reluctant to have electrodes placed on his scalp.

As in normal children, Justin exhibited robust a mu wave while he sat around idly, and the mu wave was suppressed whenever we asked him to make simple voluntary movements. But remarkably, when he watched someone else perform the action, the suppression did not occur as it ought to. This observation provided a striking vindication of our hypothesis. We concluded that the child’s motor-command system was intact—he could, after all, open doors, eat potato chips, draw pictures, climb stairs, and so on—but his mirror-neuron system was deficient. We presented this single-subject case study at the 2000 annual meeting of the Society for Neuroscience, and we followed it up with ten additional children in 2004. Our results were identical. This observation has since received extensive confirmation over the years from many different groups, using a variety of techniques.²

For example, a group of researchers led by Riitta Hari at the Aalto University of Science and Technology corroborated our conjecture using MEG (magnetoencephalography), which is to EEG what jets are to biplanes. More recently, Michele Villalobos and her colleagues at San Diego State University used fMRI to show a reduction in functional connectivity between the visual cortex and the prefrontal mirror-neuron region in autistic patients.

Other researchers have tested our hypothesis using TMS (transcranial magnetic stimulation). TMS is, in one sense, the opposite of EEG: Rather than passively eavesdropping on the electrical signals emanating from the brain, TMS creates electrical currents in the brain using a powerful magnet held over the scalp. Thus with TMS you can induce neural activity artificially in any brain region that happens to be near the scalp.