A model of "distributed cognition" is contrasted with the "mental representation" model exemplified by Tomasello and Call's Primate Cognition. Rather than using behavior as a basis for inferences to invisible mental events such as intentions, the distributed approach treats communicative interactions as, themselves, directly observable cognitive events. Similar to a Vygotskian approach, this model characterizes cognition as "co-constructed" by the participants. This approach is thus particularly suitable for studying primates (including humans), whose reliance on multiparty negotiations can undermine the researcher's ability to extrapolate from observable outcomes back to individual intentions. Detailed (e.g., frame-by-frame) analyses of such interactions reveal cross-species differences in the relevant media of information flow (e.g., behavioral coordination, relative gaze) as well as in the flexibility and complexity of the trajectories observed. Plus, with its focus on dynamics, the distributed approach is especially useful for modeling developmental and evolutionary processes. In discussing enculturation and the ontogeny of imitation, its emphasis is on changes in how expert and novice participate in such events, rather than how either may represent them. Primate cognitive evolution is seen as involving changes in context sensitivity, multi-tasking, and the coordination of social attention. Humans in particular – in, especially, the context of teaching – are seen as having specialized in linking co-perception with the refined sensory-motor coordination that enables them to translate observed behavior into strategically similar action. Highlighting the continuity between human and nonhuman development, this promising, complementary model enables us to tap the richness of micro-ethology as a cognitive science.