Kyaga et al ^{Reference Kyaga, Lichtenstein, Boman, Hultman, Långström and Landén1} have produced an excellent analysis based on the Swedish registers, which finds an increased rate of creativity in patients with schizophrenia or bipolar disorder, and their relatives. This lends support to the model for a correlation between schizophrenia, creativity and fitness that was developed jointly by one of us. ^{Reference Shaner, Miller and Mintz2} However, the authors claim that this finding supports the balancing selection hypothesis that aims to explain why psychiatric disorders have persisted throughout evolution. The theory stipulates that if patients with such severe disorders have fewer children, then the genetic variants responsible for the illnesses should be filtered out from the general population, unless this effect is balanced by adaptive advantages harboured by these variants. Relatives of patients, who also carry such variants but are free from illness, might therefore have an increased fitness.

A higher level of creativity could indeed be advantageous and increase fitness, as outlined by the authors. However, there are many other qualities that can also increase fitness, for example being faster, stronger, having a higher cognitive ability, being more attractive or living longer. The only outcome that matters for evolution is how many children an individual will leave, because if one does not pass on his or her genetic variants, these variants will disappear from the population. However, a systematic review found that patients with schizophrenia have a fertility ratio of only 0.39 compared with the general population, ^{Reference Bundy, Stahl and MacCabe3} and a more recent study of the Danish population also found strongly reduced rates for both schizophrenia and bipolar disorder. ^{Reference Laursen and Munk-Olsen4} More importantly, any possible increased fertility among relatives is too small to compensate for the strongly reduced fertility of patients. ^{Reference Bundy, Stahl and MacCabe3,Reference Keller and Miller5}

In contrast, the alternative mutation selection hypothesis, which the authors also discuss, can explain this apparent paradox, provided new (de novo) mutations replenish those that are lost because of reduced fitness. We found that about 5% of probands with schizophrenia had a de novo copy number variation (CNV), a twofold higher rate than in controls. ^{Reference Keller and Miller5} A large proportion of these CNVs appear to be under strong selection pressure. In fact, the ten best supported CNV loci that increase risk to develop this disorder have high mutation rates (a de novo CNV occurring in between 1:3500 and 1:30 000 individuals), and are under strong selection pressure. This leads to the elimination from the general population of each new mutation at these loci in less than five generations on average. ^{Reference Kirov, Pocklington, Holmans, Ivanov, Ikeda and Ruderfer6} We anticipate that ongoing next-generation sequencing studies will also implicate the more numerous point mutations, and could help resolve this debate.

Increased creativity among individuals with severe psychiatric disorders is an advantage to them and their relatives and could cause increased fitness in relatives, but de novo mutations appear to be more relevant for the persistence of these disorders in the population.