Nearly seven decades after the recognition of autism as a clinical syndrome, and four decades after its recognition as a biogeneic rather than a psychogenic syndrome, autism remains a Janus of psychology and biology. Though everyone (except for a few French psychoanalysts) acknowledges autism’s biological causation, treatment has yet to benefit wholly from this connection and remains the province of behavioural and educational psychology. No biological marker has been singled out, no “smoking gun” on a brain scan or a blood test that would say, unequivocally, that this person is or that person is not autistic. Instead, diagnosis relies, as it always has done, on detailed clinical observation of behaviour. Part of the trouble pinning autism to a single biological cause is the potential for one and the same behavioural symptom to arise from many different biological antecedents: as an analogy, consider a motorcar in which a single behavioural symptom, abnormal engine speed, could be caused by many distinct mechanical issues in the carburetor, the ignition timing, or the cylinders. In the autistic brain, many distinct genetic and environmental factors — each of which carries only a miniscule effect by itself — converge and synergise to influence the strength of connections between brain cells. These small variations in neural connectivity effect a balance between egocentric and allocentric perspectives in how we view the things and people around us, influencing modes of sensory perception, communication, and social function. They can lead to autism when they combine in large amounts, but also to normal and often beneficial cognitive variations when they occur in moderation — for instance, some of the same genetic variations that lead to autism also confer cognitive differences that can produce aptitude for science and engineering. Thus in some sense there is no such thing as an “autism gene,” as the vulnerability to autism inheres not in any single gene but rather in the ways in which genes’ effects combine with each other and with environmental factors. The study of autism and its causes, then, is inseparable from the study of human cognitive diversity and its determinants, and autistic behaviour can be most productively construed as the response of a normal human mind to an abnormal perceptual and cognitive environment. This realisation of the continuity between autistic and non-autistic cognitive variations can usefully inform therapies, the aim of which should be to provide the communicative avenues through which people with autism spectrum conditions can develop and share and their unique insights with the broader social world. Progress in understanding autism will depend on making connections between biology and psychology, between genes, brains, and environment, and between autistic and non-autistic people.