This paper proposes an evolutionary account of the origins of tonality taken in its widest sense, i.e. as the hierarchical organization of the pitch material around a single, central pitch, which is often used to evoke stability and closure. While contemporary accounts of tonal movement appear to explain why musical pitches are organized within a functional hierarchy, and around a stable pitch, they fail to explain one of the most important aspects of tonal organization, i.e. the existence of recurring patterns that lead to the stable pitch-patterns technically known as ‘cadences’. In order to explain this phenomenon, I propose to conceive of the evolutionary origins of tonality in terms of a multimodal affective schema acquired early in life through the multimodal movement patterns that are at the heart of the exchanges between infants and caregivers. The purpose of these movement patterns is to provide a stable affective referential state so that all negative psychological states can be steered back to it by enacting, and re-enacting, the various vocal, kinesic, haptic and facial components of the schema as and when required. I argue that the origins of tonality are to be located neither in tones, nor in tonal perception as such, but in certain affective states. I also address the question whether our capacity for recognizing movement patterns that converge towards stable states in diverse phenomena originates in our cognitive understanding of the behaviour of physical objects and events. I further explore the implications of the conceptualization I propose regarding the origins of tonality by considering the human capacity to experience and structure psychological spaces and to create narrative structures. In this connection I draw attention to the significance of end-states or goal-states that appear to determine the trajectory of movement dynamics in locomotion.