How the human brain organizes semantic information has been the subject of
debate over the past 20 years. Numerous cases of modality-specific and cate
gory-specific semantic deficits have fueled this controversy. Based on this
literature, we have proposed a matrix theory of semantic organization (Cro
sson, Cato, Sadek, & Lu, 2000) that includes three dimensions: modality, ca
tegory, and attribute. More recently, functional neuroimaging has been used
to interrogate the neural substrates of modality- and category-specific pr
ocessing. However, the contribution of imaging to this endeavor is relative
ly young. Consequently, findings across imaging:: studies have led to dispa
rate conclusions. This article reviews competing theories of semantic organ
ization in the brain and provides supporting and refuting evidence for each
, incorporating examples from the imaging Literature. This literature does
provide some support for distinct neural substrates that process specific s
emantic attributes. Theory-driven research is needed to dissociate modaliti
es, categories. and attributes.