I describe a key implication of cognitive linguistics that has been ignored or undertheorized. At the center of metaphors and blends, I argue, is an act of conflation, in which abstract meaning is thought in or as a concrete form. Cognitive linguistics asserts this key fact but has failed to theorize explicitly about 1) abstract meaning and 2) its conflation with concrete domains. I examine the role of abstract domains in cognition and suggest ways that the conflation of abstract and concrete domains can be addressed from a cognitive perspective. Scientific theories go astray when they lose sight of the very phenomena they are intended to explain. A theoretical explanation or "map" is helpful only if one also possesses accurate knowledge of the "territory" itself. For cognitive science, the territory is the rich semantics of lived meaning, cognition as it is experienced. In this essay I argue that two of the main schools of cognitive linguistics, Lakoff's and Johnson's metaphor theory and Fauconnier's and Turner's blending theory, have neglected some key dimensions of actual, lived cognition. Both schools, as I see it, have been too eager to develop theories to explain cognition without obtaining a sufficient description of cognitive life itself. By performing such descriptions, I show, we confront a paradox at the heart of human thought. This paper addresses metaphors and blends but views these as nearly ubiquitous in, and central to, human cognition. Any general theory of human cognition and meaning-making must address these phenomena as central factors. For the sake of simplicity, then, I will merely use the broader term "cognition" when referring to these complex cognitive acts. (Tim Adamson)