Advances in understanding self-organization over the past few decades have led to the temptation to extend it to a model of human cognition. The extension is supported by new insights in situated cognition and success in reproducing quite complex behaviors in robots without any centralized control. Dennett has been a vocal proponent of the extension, repeatedly invoking analogies with self-organizing systems and denying the existence of a self, conceived as an inner locus of information and control. I arguei argue that there is a difference between self-organizing collections and collectives. Only the latter are agents. And this difference is crucial for our understanding of selves.