It is sometimes possible for this circuit to go haywire. Your autonomic response to something arousing manifests as increased sweating, heart rate, muscular readiness, and so on, to prepare your body for action. In extreme cases this surge of physiological arousal can feed back into your brain and prompt your amygdala to say, in effect, “Wow, it’s even more dangerous than I thought. We’ll need more arousal to get out of this!” The result is an autonomic blitzkrieg. Many adults are prone to such panic attacks, but most of us, most of the time, are not in danger of getting swept away by such autonomic maelstroms.
With all this in mind, our group explored the possibility that children with autism have a distorted salience landscape. This may be partially due to indiscriminately enhanced (or reduced) connections between sensory cortices and the amygdala, and possibly between limbic structures and the frontal lobes. As a result of these abnormal connections, every trivial event or object sets off an uncontrollable autonomic storm, which would explain autistics’ preference for sameness and routine. If the emotional arousal is less florid, on the other hand, the child might attach abnormally high significance to certain unusual stimuli, which could account for their strange preoccupations, including their sometimes savant-like skills. Conversely, if some of the connections from the sensory cortex to the amygdala are partially effaced by the distortions in salience landscape, the child might ignore things, like eyes, that most normal children find highly attention grabbing.
To test the salience landscape hypothesis we measured galvanic skin response (GSR) in a group of 37 autistic and 25 normal children. The normal children showed arousal for certain categories of stimuli as expected but not for others. For example, they had GSR responses to photos of parents but not of pencils. The children with autism, on the other hand, showed a more generally heightened autonomic arousal that was further amplified by the most trivial objects and events, whereas some highly salient stimuli such as eyes were completely ineffective.
If salience landscape theory is on the right track, one would expect to find abnormalities in visual pathway 3 of autistic brains. Pathway 3 not only projects to the amygdala, but it routes through the superior temporal sulcus, which—along with its neighboring region, the insula—is rich in mirror neurons. In the insula, mirror neurons have been shown to be involved in perceiving as well as expressing certain emotions—like disgust, including social and moral disgust—in an empathetic manner. Thus damage to these areas, or perhaps a deficiency of mirror neurons within them, might not only distort the salience landscape, but also diminish empathy, social interaction, imitation, and pretend play.
As an added bonus, salience landscape theory may also explain two other quirky aspects of autism that have always been puzzling. First, some parents report that their child’s autistic symptoms are temporarily relieved by a bout of high fever. Fever is ordinarily caused by certain bacterial toxins that act on temperature-regulating mechanisms in the hypothalamus in the base of your brain. Again, this is part of pathway 3. I realized that it may not be coincidental that certain dysfunctional behaviors such as tantrums originate in networks that neighbor the hypothalamus. Thus the fever might have a “spillover” effect that happens to dampen activity at one of the bottlenecks of the feedback loop that generates those autonomic-arousal storms and their associated tantrums. This is a highly speculative explanation but it’s better than none at all, and if it pans out it could provide another basis for intervention. For example, there might be some way to safely dampen the feedback loop artificially. A damped circuit might be better than a malfunctioning one, especially if it could get a kid like Steven to engage even just a little bit more with his mother. For example, one could give him high fever harmlessly by injecting denatured malarial parasites; repeated injections of such pyrogens (fever-inducing substances) might help “reset” the circuit and alleviate symptoms permanently.
Second, children with autism often repeatedly bang and beat themselves. This behavior is called somatic self-stimulation. In terms of our theory, we would suggest that this leads to a damping of the autonomic-arousal storms that the child suffers from. Indeed, our research team has found that such self-stimulation not only has a calming effect but leads to a measurable reduction in GSR. This suggests a possible symptomatic therapy for autism: One could have a portable device for monitoring GSR that then feeds back to a body stimulation device which the child wears under his clothing. Whether such a device would prove practical in a day-to-day setting remains to be seen; it is being tested by my postdoctoral colleague Bill Hirstein.
The to-and-fro rocking behavior of some autistic children may serve a similar purpose. We know it likely stimulates the vestibular system (sense of balance), and we know that balance-related information splits at some point to travel down pathway 3, especially to the insula. Thus repetitive rocking might provide the same kind of damping that self-beating does. More speculatively, it might help anchor the self in the body, providing coherence to an otherwise chaotic world, as I’ll describe in a moment.
Aside from possible mirror-neuron deficiency, what other factors might account for the distorted salience landscapes through which many autistic people seem to view the world? It is well documented that there are genetic predispositions to autism. But less well known is the fact that nearly a third of children with autism have had temporal lobe epilepsy (TLE) in infancy. (The proportion could be much higher if we include clinically undetected complex partial seizures.) In adults TLE manifests as florid emotional disturbances, but because their brains are fully mature, it does not appear to lead to deep-seated cognitive distortions. But less is known about what TLE does to a developing brain. TLE seizures are caused by repeated random volleys of nerve impulses coursing through the limbic system. If they occur frequently in a very young brain, they might lead, through a process of synapse enhancement called kindling, to selective but widespread, indiscriminate enhancement (or sometimes effacement) of the connections between the amygdala and the high-level visual, auditory, and somatosensory cortices. This could account both for the frequent false alarms set off by trivial or mundane sights and otherwise neutral sounds, and conversely for the failure to react to socially salient information, which are so characteristic of autism.
In more general terms, our sense of being an integrated, embodied self seems to depend crucially on back-and-forth, echo-like “reverberation” between the brain and the rest of the body—and indeed, thanks to empathy, between the self and others. Indiscriminate scramblings of the connections between high-level sensory areas and the amygdala, and the resulting distortions to one’s salience landscape, could as part of the same process cause a disturbing loss of this sense of embodiment—of being a distinct, autonomous self anchored in a body and embedded in a society. Perhaps somatic self-stimulation is some children’s attempt to regain their embodiment by reviving and enhancing body-brain interactions while at the same time damping spuriously amplified autonomic signals. A subtle balance of such interactions may be crucial for the normal development of an integrated self, something we ordinarily take for granted as the axiomatic foundation of being a person. No wonder, then, that this very sense of being a person is profoundly disturbed in autism.