People generally process concrete verbal materials (e.g., the word "chair") more quickly and accurately than abstract verbal materials (e.g., the word "importance"). This finding, while straightforward and easily replicable, has been a source of theoretical controversy for decades. Multiple semantics accounts (e.g., Paivio's dual-coding theory) explain concreteness effects by postulating separate representational or processing systems for pictorial and verbal meanings, with concrete words processed more quickly and accurately because they are processed in both systems, as opposed to abstract words which are processed in both. In contrast, unitary semantics accounts (e.g., the context-availability model of Schwanenflugel and colleagues) posit only quantitative differences between concrete and abstract words. Specifically, they argue for a single type of knowledge representation, with concreteness effects resulting from differences between concrete and abstract words in either externally provided or internally available context. In this talk, I review cognitive neuroscience investigations of this issue, focusing in particular on event-related potential (ERP) studies.