This commentary considers three key themes in the target article, from the perspective of the self-evidencing brain. In brief, I examine the notion of minimal selfhood in terms of inference to the best explanation; in particular, an explanation for sensations that are uniquely available to a self. This begs the question of how self differs from non-self and how non-self differs from the concept of another. From an (active) inference point of view, the building of generative models that entail minimal selfhood rests on structure learning and Bayesian model selection. In this setting, I focus on the role of generalized synchrony – and the implicit closure of action-perception cycles – in dyadic interactions. Finally, I touch on the importance of interoception awareness and the implications of how interoceptive signals are contextualized for translating the ideas in the target article into developmental psychopathology.