This project addresses a serious gap at the heart of the social and cognitive sciences. The proposal is to fast-start an interdisciplinary science of human communicative interaction focused on sequences of contingent actions. The research will be guided by a novel hypothesis, the ‘interaction engine’, which holds that human interactive abilities are distinct from and phylogenetically older than our language capacity. Inverting the usual assumptions, the hypothesis suggests that the human interaction system is fundamentally universal, while the language system lacks substantial universals and is diversified by cultural evolution. The interaction system and language system therefore do not mesh neatly, and this can be detected in the domain of contingent action sequences – the crucial characteristic of human interaction. Two specific areas, (a) the timing of turn-taking and (b) the ascription of speech acts or intentions, will be examined from three distinct disciplinary perspectives:(i) cross-cultural corpus and experimental studies in a dozen languages,(ii) developmental studies (corpus and experiment based) from early infancy up to middle childhood,(iii) cognitive studies employing both behavioural measures (reaction time, eye tracking), and neurocognitive methods (EEG, fMRI).The ‘interaction engine’ hypothesis makes detailed predictions about cross-cultural universals, developmental trajectories, and processing complexities in both turn-taking and action ascription.The project aims to build an integrated, interdisciplinary science of communicative interaction. New developments (multi-media corpora, dual eye-tracking, EEG, fMRI, baby-lab techniques) now make this feasible. The PI and host institution are uniquely equipped to undertake this project.