Slide 33

These results fit well within an emerging pattern in autism in which posterior brain regions associated with lower-level processing are abnormally strongly activated,

anterior brain regions associated with high-level or integrative processing are abnormally weakly activated or deactivated,

and long-range connectivity between brain regions is abnormally low.

This combination of results begins to sketch the neurophysiological underpinnings of the psychological concept of weak central coherence.

With the aid of functional anatomy, we can see that weak central coherence and competing explanations such as the executive dysfunction theory and the empathising-systemising theory may be but several aspects of a single idea, in which abnormally weak long-range connections between brain regions produce a lack of capacity for top-down coordination of cognitive resources, and in which abnormally strong local network connectivity within brain regions may be associated with intense but weakly modulated lower-level processing.

The idea of local network overconnectivity in autism is consistent with autism's comorbidity with epilepsy, with findings such as genetic linkage to a GABA receptor gene cluster and low levels of GABA binding, and with psychophysical results suggesting abnormally noisy neural processing, such as high motion coherence thresholds and broadly tuned auditory filters.

As you'll hear from Ruth Carper later in this symposium, there are anatomical data to support such a dichotomy between local and long-range connectivities.

You'll also be hearing from Lisa Boulanger about ways in which infectious aetiologies and immunological abnormalities may affect the development of neural connectivity,
and from Andrea Beckel-Mitchener about how different abnormalities at the cellular level may produce similarly abnormal properties in autism and in Fragile X syndrome.
In addition, Sara Jane Webb will be telling you about EEG evidence for long-range underconnectivity between brain regions that are abnormally highly activated but weakly coordinated.

Though the possible relationship between these complementary local and long-range aberrations of connectivity remains unclear, it's interesting to note that one abnormality may produce or reinforce the other. In a very strongly connected network, the capacity to represent information can be degraded: when an input can't activate one processing unit without also activating its neighbours, the effective number of independent processing units is lowered, and the capacity to discriminate signal from noise diminished. There is some indirect evidence from studies of minicolumn structure that autism may involve this sort of pathology of information processing.

Such a low-entropy local network would produce a high level of activity, but a low capacity to modulate this activity since the network would always be operating near the top of its range. This combination of high activity and low modulation is of course exactly what we see when we study selective attention in autism.

You'll hear shortly from Greg Allen about how this sort of indiscriminate neural activity may sabotage the normal activity-dependent development of long-range functional connections. And of course, vice versa, a lack of appropriate long-range connectivity can be expected to affect local functional specialisations.