Autism is largely defined in terms of behaviour. While there are related genetic conditions (e.g. fragile-x), these do not occur in every individual who is diagnosed as autistic. There is no biological abnormality with a one-to-one correspondence; to make the logic of this plain: biological factor x is present if and only if autistic behaviour is observed.
Because a set of behavioural traits can be observed in children diagnosed as autistic (e.g. echolialia, hand flapping, lack of normal eye contact, existential isolation), it is often assumed that there is one kind of underlying handicap, one kind of autism. This is a natural conculsion, but it is also a logical mistake. In medicine, it is possible for different illnesses to produce the same surface symptoms; in evolutionary biology, it is possible for quite different biological groups to evolve similar behavioural and structural characteristics so that they come to resemble one another, even though their deeper biology is quite different - this is termed convergent evolution. It is therefore possible that there is, at a deeper biological level, more than one kind of autistic handicap giving rise to similar behavioural traits.
Why is this important? If research into causal factors in autism assumed one basic set of biological deviance, and there is more than one set, then research will tend to pinpoint the intersection of the sets. But if there is no intersection of sets, then research will fail to find any causal factors, and fail.
The marked prevalence of males to females with autism has led to the modern therory that autism is a genetic handicap resulting from abnormalities at a deep biological level before birth. This is, however, not necessarily true. It is equally possible that there is an inherent fragility in all males, which could be triggered by an external catalyst at an early age, depending upon the degree of fragility. If this was the case, then research into genetic causes would run into problems, because only the inherent fragility might be considered, and not the external catalyst. It is also possible that more than there is more than one catalyst, particiarly if several disease or vaccine result in the same catalytic effect.
Causal Factors in Autism: A Tripartite Model (2002)