Analogical priming via semantic relations

The metaphor of semantic memory as a rich network of associated concepts has become both an assumption and a tool of psychologists studying diverse cognitive processes. Many studies of semantic memory take advantage of the fact that concepts are associated; however, relatively little work has investigated the effects of the specific content of those associations, which are typically semantic relations. Yet such fundamental cognitive processes as discourse comprehension, inference, problem solving, and analogical reasoning depend on the human ability to represent and process the semantic relations between concepts. Research has shown that people are able to decide quickly whether basic semantic relations, such as superordination, antonymy, and partonymy, hold between a pair of concepts (e.g., Chaffin & Herrmann, 1988; Collins & Quillian, 1969; Glass, Holyoak, & Kiger, 1979). However, there has been little investigation of the possible role of such relational decisions in promoting subsequent access to analogously related concepts (but see Chaffin & Herrmann, 1988). Standard network models of semantic memory depict object concepts as nodes and relations as links between these nodes. When one concept (typically corresponding to a noun) is activated, that activation is assumed to spread to linked concepts. In Quillian’s (1968) original formulation, it was assumed that the semantic content of some links (typically corresponding to verbs) would also be activated, because those links were themselves concepts1 (Collins & Loftus, 1975). Despite the extensive research on semantic priming since Quillian’s proposal, little work has examined relational priming per se—that is, the role of links as structured semantic concepts, as opposed to simple conduits for activation. This gap is troublesome because of the general importance of relations and role bindings in cognition.