Word recognition in the brain

Abstract

Concrete words, such as ‘chair’ or ‘cat’, are generally recognized faster and more accurately than abstract words, such as ‘love’ or ‘liberty’, even if differences in word length and frequency of occurrence are controlled for. This has led some researchers to propose a ‘dual-code’ theory of word recognition: concrete words have both a verbal and a visual code, and, hence, are recognized more efficiently than abstract words, which have only a verbal representation. On the assumption that the image-based code involves right-hemisphere processing more than the verbal code, the processing of concrete words but not of abstract words would be expected to activate the right hemisphere. Neuroimaging studies, however, have so far shown the opposite: abstract rather than concrete words activate the right hemisphere1,2. Recently, Kiehl et al. reported results from an fMRI study, showing that when people performed a lexical decision task on real abstract words, real concrete words and pseudowords, both the left and right hemispheres were activated during the taks for concrete and abstract words relative to the pseudoword baseline3. Importantly though, abstract words also produced activation in the anterior part of the right superior temporal gyrus, a finding that is inconsistent with a dual-code model. Kiehl et al. hypothesize that the recognition of abstract words is slower and less accurate than that of concrete words because more extensive semantic processing is required for abstract words, and this involves the right superior temporal cortex. Exactly what this semantic processing consists of, however, will require further investigation.