Intended for healthcare professionals

Clinical Review

Diagnosis of autism

BMJ 2003; 327 doi: (Published 28 August 2003) Cite this as: BMJ 2003;327:488

Limits of the underconnectivity theory of autism

Limits of the underconnectivity theory of autism

A reader of this E-forum has sent me a comment, in which he relates
the underconnectivity theory of autism with what I have suggested in my
previous rapid responses about the pathogenesis of the disorder. This
colleague refers to the newly emerging concept that local
hyperconnectivity coupled with distant underconnectivity of brain
organizations could be the basis of autistic symptomatology.
I have been reading a few interesting articles in support of the
underconnectivity theory of autism [see for example references 1 and 2,
although the points of view of the two research groups do not coincide
completely]. Very briefly, this theory entails that brain computational
centers work in a less synchronized way in autism, especially when higher
- order integrative functions are at stake. Dynamic imaging studies have
been conducted which support this theory, and I have been taking these
results as a confirmation of what clinicians have been inferring until now
from the observation of the behavior of persons with autism, id est the
presence of a deficit in the multiple parallel processing of receptive and
executive data [3].

However, there are some aspects of normal brain functioning and of
autism which the underconnectivity theory seems unable to explain. I will
quote some of them.

1) How in individuals without autism (hence with a diffusely good
neural connectivity) are computational centers involved in the processing
of a specific perceptive and/or motor and/or speculative task included in
the synchronous web of active processing while unnecessary centers are

2) How are previously excluded neural areas flexibly reintroduced in
that web when the individual is coping with a new task, a task which is
completely different and requires the contribution of different areas?
Think about shifting from observing an event which happens in one
particular site, to listening to a dialogue taking place elsewhere, to
speculating on the two perceptions while filtering out the data which are
still coming from both of them.

3) How can normally developing children so rapidly and precisely
change their operational targets, in strict connection with the requests
of the environmental and/or as the result of their free choice, while
children with autism show variable levels of impairment in this ability?
In autism, variations in the control of attentional shifts and probably of
thought processes seem to range from a virtual absence of control to less
readily detectable impairments. Nonetheless, it is reasonable to assume
that a certain ability to modify patterns of neural arousal, with a shift
in the distribution of neural connectivity is possible even in individuals
with autism, and especially in high functioning individuals, albeit much
less efficiently.

4) How can the underconnectivity theory account for the frequent
phenomenon of altered sensitivity in autism? Normal individuals are able
to regulate the intensity of all but the most strong and abrupt variations
in their perceptions of the environment, and even some physical pain can
be less disturbing when the average individual is involved in an absorbing
intellectual, emotional, or physical task.

It becomes then necessary to think about connectivity not as a stable
quality of neural organizations, but as a phenomenon subject to goal -
oriented variations.

I think that, to explain normal brain functioning and the autistic
dysfunction, we must draw the underconnectivity theory nearer to the weak
central coherence theory. We must presume that a smooth and rapid change
in the distribution of neural connectivity in normality is driven by the
active intervention of a modulatory circuit. My opinion is that it is
precisely this circuit that is out of order in autism. Two consequences
would derive from this impairment: a reduction in the extension of the
online web of synchronous neural arousal, wherein higher order
computational centers will be less represented, and a slower ability of
the same web to reconfigure itself following the continuously changing
requests stemming from the environment or from the willing self. In fact
this ability to synchronize, connect, activate, filter out and exclude
neural organizations could be the main factor of our physiological ability
to perceive ourselves with a distinct sense of personal continuity and
identity and a sense of independent choice.
A deficit in the coordination of neural functions in autism could act as
the “bottleneck” [4] that narrows the passage from lower-complexity brain
activities (e.g. the reproduction of visual images by primary cortical
areas in the occipital lobe) to higher-complexity brain activities (the
transformation of those images into more meaningful abstractions by the
fusiform gyrus and related circuitries).

It is possible that mechanisms for distinguishing internally
generated data from data coming from the five senses are included in the
functions of this modulatory circuit. This would explain the genesis of
the hallucinations of individuals with schizophrenia, and of the
hallucinations supposed to be present also in persons with autism. The two
pathologies would differ because in autism this putative dismodulation
would start very early in life, probably under the overburdening effects
of a rapid increase in the maturation of brain organizations and of their
hard-wired connections to the whole brain apparatus, while in
schizophrenia a breakdown of central integration would take place during
the final phases of enlargement of the computational potential of the
human brain, when the last cortical areas become part of the whole
processing web.
Neurobiological findings such as macrocrania, an increase of white and
gray matter and the minicolumnar changes in the cortex [4] might be
explained by a reduction in the physiological, pre-programmed neuronal
death, which could be related to the underfunctioning of this integrative
and maybe onthogenesis – regulating circuit.

This meta-circuit might be extensively distributed in the brain, and
linked with the prefrontal cortex, where plans for actions are constantly
produced and adaptively modified; with the temporal lobes, where previous
and actual aversive situations are detected); with the cerebellum, which
might actually configure the web of arousal depending on signals coming
from the other nodes of the system; and with the thalamus, where the
monitoring and regulation of sensory inputs might take place.


1. Just MA, Cherkassky VL, Keller TA, Minshew NJ. Cortical activation
and synchronization during sentence comprehension in high-functioning
autism: evidence of underconnectivity. Brain 2004; 127: 1811-21.

2. Belmonte MK, Allen G, Beckel-Mitchener A, Boulanger LM, Carper RA,
Webb SJ. Autism and abnormal development of brain connectivity. J
Neurosci. 2004; 24: 9228-31.

3. Loddo, S. An attempt to define the core dysfunction of autism.
Autism, (2004); 8, 335-336.

4. Castelli F, Frith C, Happe F, Frith U. Autism, Asperger syndrome
and brain mechanisms for the attribution of mental states to animated
shapes. Brain. 2002. 125:1839-49

5. Casanova MF, Buxhoeveden DP, Switala AE, Roy E Minicolumnar
pathology in autism. Neurology 2002 58:428–432

Competing interests:
None declared

Competing interests: No competing interests

23 December 2004
Silvio Loddo
Child and Adolescent Psychiatrist
Neuropsichiatria dell'Età Evolutiva, AUSL n. 5 Oristano, Italia