This paper aims to contribute to discussions on the multimodal nature of cognition through an elaboration of the ways multimodal aspects of thinking are exploited by learners doing mathematics. Moving beyond the fact "that" multimodality occurs, this paper focuses on "how" it occurs, with particular attention drawn to the complex network of perceptual, bodily and imaginary experiences of students. Through an analysis of 2 selected episodes of the work of 1 primary school child learning to graph motion, the paper shows how the notion of blending spaces is central to understanding the mechanism of multimodality.