Heredity advance online publication, 3 October 2012; doi:10.1038/hdy.2012.64In a recent paper, Megens and Groenen (2012) observed thatthousands of generations of human-mediated (artificial) selectionhas produced a huge number of distinct phenotypes, which constitute‘the kinds of long-term genetic experiments researchers usually canonly dream about’. Few people would challenge the power of artificialselection, but my last visit to the vet made me think that perhapssomething more complex is going on. He has recently managed todefine the standards of a new dog breed that blends the bestcharacteristics of local herding dogs, which we call ‘tabui’. Afterhaving visited my Australian shepherd Rocco (which, by the way, washealthy as a horse), the vet patted his head and told me: ‘He is four-eyed. Just like my tabuis’. He was referring to the two little brownspots on Rocco’s forehead, which mimic the shape and color of hiseyes (Figure 1). Australian shepherds come in three recognized colorvarieties, all characterized by the presence of the false eyes. Tabuis alsocan vary in coat color, but they are almost always four-eyed. The vettold me that he thinks the false eyes help the guardian dog to keep thecattle under pressure by confounding them. This seems pretty likely.Using eye spots to intimidate or confound is actually a commonadaptation in many animal species (Ruxton, 2005; Ohno and Otaki,2012). Perhaps, a four-eyed dog could also look awake while sleeping,thus being an effective guardian even when resting.I do not think local shepherds have been aware of the potentialeffect of false eyes while breeding tabuis generations, and I am surethat being four eyed was not one of the main traits that were underdirect artificial selection. But being a good guardian was. Many otherbreeds of herding dogs are four-eyed. It is hard to believe that this hasresulted from a widespread conscious selection associated with thepresence of the false eyes. Yet, we may hypothesize that, acrossgenerations, four-eyed dogs have worked better than their colleagues.On one hand, this is coherent with the concept of artificial selection:the breeder chooses to mate the individuals most suited for particularpurposes (Driscoll et al., 2009). On the other hand, being four eyed isnot exactly the same as being short-legged and having a strong tail,which makes it easier for a hunter to pull his hound dog out of a hole.The false-eye issue cannot be assimilated to the Darwinian concept ofartificial selection (Darwin, 1890), or to the modern one (Driscollet al., 2009), or interpreted as a correlated response to artificialselection of particular traits (Careau et al. 2010). It is a matterbetween dogs and cows.I wonder how often the path of artificial selection is affected bycharacters that make an individual more likely to be selected, but arenot under the breeder’s control. Maybe scenarios of human-inducedbut not human-controlled, quasi-artificial selection are quite common.I will keep my (true) eyes open.CONFLICT OF INTERESTThe author declares no conflict of interest.ACKNOWLEDGEMENTS