This paper defends the conditional thesis: if non-human animals engage in instrumental practical reasoning involving individual innovation, then such animals are likely self-conscious. Against neo-Kantian views that tie rationality to conceptual self-consciousness, and against empiricist critiques that decouple animal rationality from self-consciousness, I argue for an intermediate position. Drawing on cases of animal tool use (hook-bending in birds), I show that representing non-actual states of affairs and combining such representations with goals—a hallmark of individual innovation—requires a form of active cognition best explained by positing nonconceptual self-representation. I distinguish this from teleofunctionalist explanations, which attribute representational combination to selection mechanisms (evolutionary, social, etc.) rather than individual cognition. Where teleofunctional accounts fall short, self-representation plays an explanatory role akin to the essential indexical in human action. The upshot is a novel case for the essential connection between self-consciousness and rationality that accommodates animal cognition while preserving key neo-Kantian insights.