Psychiatric disorders are disadvantageous behavioral phenotypes, yet they have been existing in humans for long time. A recent epidemiological survey has reported decreased fecundity in patients with psychiatric disorders such as schizophrenia and autism spectrum disorders, with fecundity of relatives of the patients not exceedingly higher than normal, predicting that prevalence of psychiatric disorders among humans should be decreased. Nevertheless, in reality, prevalence rate of psychiatric disorders in humans have been either constant or even increasing more recently. Several attempts have been made to explain this fact, including de novo gene mutations or variants associated with psychiatric disorders, although none of them fully explains it. Here we propose a novel, theoretical model that psychiatric disorders may have been emerged in evolution of organisms as adaptation against peculiar environmental conditions associated with stress. As such, we propose that, for understanding of the biological mechanisms of psychiatric disorders, a novel and fruitful direction of research is aimed to examine that (1) behavioral traits associated with psychiatric disorders yield advantages for reproduction depending on the environmental contexts; (2) Neuronal and neurodevelopmental changes caused by stressful environments are understood as adaptation to the specific stressful environments; and (3) Such neuronal changes caused by the stressful environments are inheritable. We discuss these issues along with recent studies including our own preliminary data showing transgenerational inheritance of stress-induced behavioral and epigenetic changes and the effects of prenatal-postnatal environmental interaction that cause adaptive neurodevelopmental changes.