We thank readers for their interest in our editorial. ^{Reference Kelleher, Jenner and Cannon1} The main purposes of the editorial were threefold: (i) to highlight the relatively recent identification and characterisation of a non-clinical psychosis population (for review see Kelleher & Cannon ^{Reference Kelleher and Cannon2} ); (ii) to point out that there might be important overlap in the genetics of the clinical and non-clinical psychosis phenotypes; and (iii) to discuss the potential value of this population for empirically testing evolutionary theories of psychosis.

Dr Euba points out that hypervigilance may lead to an individual being ‘handicapped by an inability to trust others in the social group’ and as a result being less likely to procreate. However, hypervigilance is not in itself a disadvantage. In fact, the more vigilant an animal, the more likely it is to identify threats such as predators and to protect both itself and its progeny, allowing the propagation of associated genes. Increasing levels of vigilance, however, would promote survival of the organism and its progeny only to a point. As this trait becomes ever more pronounced, it would eventually lead to the dysfunction identified by Euba – paranoia. Nesse referred to this as cliff-edged fitness, ^{Reference Kipling3} whereby traits may increase fitness up to a critical threshold, but beyond this point, fitness drops precipitously (the cliff edge here being the transition from hypervigilance into paranoia). Thus, while in its extreme form – paranoia – hypervigilance will be negatively selected owing to negative fitness consequences, a ‘subthreshold’ level of this trait would be positively selected.

We agree with Treffurth that it is possible that non-clinical psychotic symptoms may be neither advantageous nor disadvantageous and that associated genes may have been passed on alongside other fitness-enhancing phenotypes. Our argument, however, is that if, as has been suggested by many researchers to date, the genetics of psychosis encode for positive as well as negative traits, then people with the recently characterised non-clinical psychosis phenotype may provide a valuable population in which to conduct empirical research.

Hubbeling makes the very point that we wished to emphasise in our editorial – that the non-clinical psychosis phenotype provides us with a population in which to test hypotheses about the evolutionary benefit of psychosis genes. It is clear why genes that promote certain traits, such as language development, hypervigilance and complex social cognition, would be selected in evolution. The ‘how’ questions, as Hubbeling points out, require attention, for instance how these traits differ in (non-psychotic) persons with psychosis genes compared with persons without (or with fewer) psychosis genes. This type of research is precisely what we wish to encourage by highlighting the validity of the non-clinical psychosis phenotype for empirical investigation. This population provides a potentially valuable means for moving beyond ‘just-so’ stories ^{Reference Nesse4} into the realm of testable hypotheses.