The progress that is now possible by combining cognitive theories, functional and structural brain imaging in genetically sensitive designs, should yield some long awaited answers. In particular, it should become clear which features of autism have separate and independent causes, and which arise from one and the same origin in the brain. When the genes for susceptibility to ASD are identified, the diagnosis may be revolutionized. Then cases that are now considered to fall on the same spectrum may be revealed to belong to completely different etiological subgroups, while previously unidentified cases may be identified within genetic pedigrees. Once genes are isolated, then animal models will become truly useful for identifying the neurophysiological mechanisms and devising means of repairing and preventing neurological abnormalities.All of this progress, however, requires a real understanding of how autism unfolds through development, which are core elements and which merely secondary and avoidable knock-on effects. This long-term aim can only be achieved by integrating bottom-up approaches, such as genome-wide screening, and top-down research such as establishing the neural basis of hypothesised cognitive assets and deficits.