In this talk I will illustrate four different perspectives that we as a community have embraced to study robotic manipulation: 1) controlling a simplified model of the mechanics of interaction with an object; 2) using haptic feedback such as force or tactile to control the interaction with an environment; 3) planning sequences or trajectories of manipulation actions to achieve long-horizon goals; and 4) using visual cues to guide manipulation actions. These are complementary perspectives, and building general dexterous robotic manipulation systems requires integrating them. I will discuss the key role that manipulation primitives play at integrating these perspectives. In particular I will present recent work on tactile dexterity to embed tactile feedback into the mechanics models of frictional contact, and on planning with visual affordances to execute dexterous long-term behavior on novel objects. I will illustrate this work in the context of a dual-arm dexterous robotic system.