On the nature of blending as a cognitive phenomenon

Abstract

This paper argues for the integration of Turner and Fauconnier's concept of blending within the general architecture of cognition provided by production models. It also revises the notion of blending as proposed by Turner and Fauconnier and observes that blending obeys general constraints of cognitive economy and of consistency; more specifically, it is argued that blending responds to Lakoff's Invariance Principle, of which a slightly modified version is provided here and defined as the Extended Invariance principle. The Extended Invariance principle accounts for the possibility of preservation of all generic-level structure (not only image-schematic structure) in metaphoric mappings, and allows for the possibility of more than one input space in the mapping process. Finally, blended spaces are seen as one possible outcome of the activity of working memory, while the apparent conceptual irregularities and asymmetries which Turner and Fauconnier observe in their emergence are alternatively explained as contextual implications motivated by and derived from the convergence of information from different Idealized Cognitive Models (or ICMs) in the action part of productions.