The concept of innateness remains extremely unclear although it is widely used by biologists and cognitive scientists. However, when researchers talk about “innate traits”, they obviously mean “genetically specified traits”. But is it not the case for every trait to be “genetically specified” in one sense? Furthermore, in crossing distinct disciplinary boundaries or in following the folk's misconceptions like uninformative dichotomies (innate/acquired, innate/learnt), innateness seems to produce a confusing and unhelpful notion. Relying on the complexity of ontogenetic development, some researchers thus hold that this concept should be rejected (Lehrman, 1953; Oyama, 2000; Griffiths, 2002).

In this paper, I will argue that the complexity of ontogenetic development, instead of being a reason to abandon the notion of innateness, reveals its usefulness. With this in mind, I propose to show that Ariew's account of innateness (2006) as empirical developmental canalisation is perfectly relevant for some traits. Yet, I will argue that this account is not sufficient for other more complex traits. In other terms, Ariew's identification of three developmental patterns (innate/acquired/triggered) from isolation experiments with songbirds is not always feasible. I will then state that, as innateness seems to be a theoretical term depending on specified theoretical contexts, it has to be viewed as a primitive term intervening within explicit and empirically robust developmental scenarios. It means that every innateness ascription today is relied upon a specific developmental theory. Therefore, it is likely to change with regard to progress in understanding development.