Cognitive definitions cannot accommodate fear as it occurs in species incapable of sophisticated cognition. Some think that fear must, therefore, be noncognitive. This paper explores another option, arguably more in line with evolutionary theory: that like other "biological universals" fear admits of variation across and within species. A paradigm case of such universals is species: it is argued that they can be defined by ostension in the manner of Putnam and Kripke without implying that they must have an invariable essence. (...) Emotions can be defined in this way too, in principle, but the theoretical understanding of homology necessary to do so is lacking at present. (shrink)