In this and related articles, Fincher & Thornhill (F&T) present an innovative and bold analysis of cross-cultural differences in psychology and behaviour. It is clear that parasitism – like predation – has acted as a powerful selection pressure on animal cognition and behaviour. However, a critical assumption of F&T's thesis is that pathogens carried by out-groups (neighbouring families or communities) will be more dangerous than those of one's own family and community. Empirical support for this proposition is tenuous.

Like footballers, pathogens tend to play better at home. In the ecological literature, this pattern is termed local adaptation and is defined as a situation where a pathogen has greater fitness within its local host population than within a non-local group. Local adaptation is not universal – F&T cite an example where fungi were better able to infect “out-group” plants (Kaltz et al. Reference Kaltz, Gandon, Michalakis and Shykoff1999) – but a recent meta-analysis concludes it is the general pattern (Hoeksema & Forde Reference Hoeksema and Forde2008). Selection on pathogens favours traits that enable them to spread within their current host population, not those that help them in neighbouring families or groups. Hence, when pathogens spread from their usual host population to another, they often have a reduced capacity to overcome host immune defences or spread from person to person. F&T present no clear evidence that the diseases used to compute their Parasite Stress Index exhibit local maladaptation in humans across relevant spatial scales – that is, neighbouring kin or religious groups.

What F&T do emphasize is that pathogen genomes show considerable geographic variation (Rougeron et al. Reference Rougeron, De Meeus, Hide, Waleckx, Bermudez, Arevalo, Llanos-Cuentas, Dujardin, De Doncker, Le Ray, Ayala and Banuls2009). However, this variation may be due to a range of factors, including adaptation to ecology, secondary host biology, behaviour of host(s), or processes besides natural selection. Pathogen genetic variation in and of itself does not imply higher virulence or transmissibility in neighbouring groups. Likewise, examples of host immune specialisation cited by F&T may simply reflect hosts with generally better immune systems (e.g., more major histocompatibility complexdiversity offering protection from a broad array of pathogens; Corby-Harris & Promislow Reference Corby-Harris and Promislow2008) or host groups whose immune systems differ because of founder effects (Miller et al. Reference Miller, Fadl, Mohamed, Elzein, Jamieson, Cordell, Peacock, Fakiola, Madhuri, Eltahir, Elhassan, Musa, Muntaser and Blackwell2007). While the authors refer to cases where inter-group contact has resulted in catastrophic epidemics, these rare occurrences are due to the evolution of highly virulent “crowd diseases” in large and completely isolated populations, and are not relevant to coevolutionary processes in neighbouring kin or religious groups.

Our point is not to claim that human groups are never more susceptible to the pathogens of neighbours due to coevolution, only that this is not the general pattern, and often the opposite will be true. Pathogen avoidance strategies involve critical compromises: out-groups may have valuable mates, allies, tools, resources, or good ideas. All of these can be fitness enhancing, and we are unconvinced that pathogen-host coevolution results in a world where forgoing these benefits generally makes adaptive sense.

If coevolution doesn't result in out-groups with more dangerous pathogens, how are F&T's results explained? One possibility is that assortative sociality is more beneficial in high-pathogen stress areas because of how it influences the shape of people's social network. Assortative sociality means people are clustered in groups such that people are well connected with each other, and poorly connected with other clusters. When a population is organised in this fashion the capacity of epidemics to spread is reduced (Keeling Reference Keeling1999; Salathé & Jones Reference Salathé and Jones2010). Hence, assortative sociality may be increasingly beneficial where pathogens are more common, independent of host/parasite coevolution. Another possibility is that ontogenetic changes in the immune system leave people more vulnerable to out-group pathogens. Illness in childhood, for example, often results in memory B-cells that respond quickly and effectively to subsequent exposure to the same pathogen. Adults may therefore be somewhat more vulnerable to pathogens of other groups, again independent of any coevolutionary processes. No doubt other factors – some unrelated to pathogens – also affect the payoffs of different social systems. Given the complexity and diversity of possible causal relationships between social relationships and disease transmission, we would encourage more formal modelling of how inter-group behaviour is optimized under different conditions of pathogen prevalence.