Almost a century ago, the great Swiss psychiatrist Eugen Bleuler coined the terms schizophrenia, as the splitting of psychic functions in Kraepelin's dementia praecox, and autism, as withdrawal from reality in people with schizophrenia. The term autism was, of course, later redefined by Leo Kanner ^{Reference Kanner1} for a childhood psychiatric condition first considered as a subset of schizophrenia, then regarded as separate and distinct by both himself and Michael Rutter. ^{Reference Rutter2} Most recently, an article by Craddock & Owen ^{Reference Craddock and Owen3} not only formally challenges the distinction between schizophrenia and bipolar disorder that followed from Kraepelin's work, but also returns autism to the scene of its Bleulerian birth, in proposing a new model for the relationships between major mental illnesses as a continuum – from intellectual disability, through autism, to schizophrenia and schizoaffective disorder, through to bipolar and unipolar mood disorder. By their model, autism is juxtaposed with schizophrenia on the basis of overlapping neurodevelopmentally based phenotypes of cognitive impairment and negative symptoms, and recent genetic data that show overlap in the genetic risk loci associated with both conditions, including a set of copy number variant loci and specific genes such as CNTNAP2 and NRXN1.

Do such molecular genetic data, dovetailed with data on select phenotypes, imply that Blueler was indeed correct, if not prescient, that autism and schizophrenia are manifestations of similar disease processes? In contrast to Craddock & Owen's ^{Reference Craddock and Owen3} proposition, a recent alternative hypothesis posits that not only are autism and schizophrenia not juxtaposed or overlapping, but they are actually diametric psychiatric opposites characterised by underdevelopment v. dysregulated overdevelopment of human social-brain phenotypes. ^{Reference Crespi and Badcock4} This alternative hypothesis for the relationship between autism and schizophrenia has recently been tested using data from the seven copy number variant loci that have been linked statistically with both conditions. ^{Reference Crespi, Stead and Elliot5} The data provide statistical support for the hypothesis that autism and schizophrenia are mediated by reciprocal variants, such that at four distinct loci, deletions predispose to one disorder, whereas duplications predispose to the other – clear support for the diametric model.

Near the end of his life, Kraepelin made it clear that schizophrenia could not be satisfactorily distinguished from bipolar disorder; his supposed dichotomy was indeed false from the near start, and recent genetic data strongly support a dimensional model for the phenotypes that make up these two conditions. ^{Reference Craddock and Owen3} But neither Bleuler nor Kanner should sleep soundly, nor should practitioners of psychiatric genetics or therapy, until the relationship between schizophrenia and autism becomes much better understood – the implications for diagnostics, pharmacology and psychiatry in general reach too far. Further targeted tests, based on clear alternative hypotheses with differentiating predictions, will be required – and as Craddock & Owen ^{Reference Craddock and Owen3} suggest, such investigations must ultimately focus on reconciling clinical categories and dimensions with normal and abnormal neurological and psychological architecture to dissect and define psychiatric conditions for the next 100 years.