<p>Humans visually assess materials to estimate their physical properties — such as judging the ripeness of fruits — and to plan actions — such as planning steps on an icy road. Despite the complexity and diversity of materials in the world, humans can infer a wide range of material properties from a brief glance. Much previous research has focused on the visual mechanisms that underlie material perception, and it is&#xa0;often described as a task of mid-level vision. In this Review, we discuss the broader importance of material perception in cognition and action planning. We begin by describing the visual inference of material properties and then discuss its relationships with recognition, language and intuitive physics. We conclude that material perception should be considered part of the perception–action loop and highlight the multisensory nature of material perception in guiding actions. We end by discussing how deep learning facilitates the discovery of meaningful representations and generates testable models of material perception. By framing material perception as a process that dynamically integrates perception, cognition and action, we may improve our understanding of how humans interpret and interact with the physical world.</p>