Improvisation is currently enjoying an intellectual vogue across fields as diverse as the musicology of free jazz to management science. But what are the theoretical moorings of this far-reaching new enterprise? First, the article offers a brief review of some potential foundations for studies of improvisation. The hypothesis that humans possess neurons for mirrored interaction because they have evolved as social animals is arguably as plausible as the notion that interactive, social behaviour is a product of a neural architecture primed for interactive cognition. Durkheim responded to a similar unresolved set of arguments about brains and cognition at the end of nineteenth century by taking his well-known late ethnographic turn (towards Australia). This takes us to the second part of the article. The ethnography of performance retains its value to nourish our understanding of larger questions regarding properties of human sociality. Specifically, the article seeks to suggest that a focus on the ritual shaping of embodied actions is crucial to understand and address the emergence of a range of competing "styles of thought." An example helps show that the "bubbles" and "echo chambers" of opinion, of which contemporary political commentators complain, are not (as supposed) products of the internet and social media, but rooted in more fundamental differences in social ordering reinforced by variations in practical and ritual performance. The article seeks to bring out the persistent "deafness" of development agencies to connections between shifting cultivation and social practices of marriage and death in a West African farming community. Calls by development agencies to abandon shifting cultivation have no effect. Approaching agrarian intervention via joint improvisation might help two circular arguments sustained by institutional differences to connect.