Biology may be able to modify philosophers' views of the question of realism in science. Two classical notions of realism are considered: 'scientific realism" (which assume the objective reality of theoretical entities), and the metaphysical thesis of the independent reality of the world. The realism of theoretical entities, which is a modern version of the "realism of universals", is rendered problematic by the extreme difficulty these sciences face in constructing laws out of their generalisations, and thus their difficulty in identifying natural kinds. The category of scientific realism raises particular problems in the case of the biological sciences: a science that is confronted throughout its range by the historical singularity of its objects. On the other hand, modern biology, and particularly evolutionary biology, provides some strong arguments that pose a problem for realism as a postulate of an independent external world: we are not pure intelligences. In a biological perspective, and more especially in an evolutionary perspective, what matters is not an independent world, but a surrounding world. This suggests that the goal of this mode of knowledge is to construct intelligible models that can respond to the complexity of the human environment, rather than to construct a global and unified image of the world.