What is semantic in semantic dementia? The decay of knowledge of physical entities but not of verbs, numbers and body parts.

Background: Conceptual knowledge does not decay randomly in patients with cerebral damage, suggesting that dedicated neural substrates may support different categories of knowledge. Semantic dementia is an optimal natural model for studying the organization of semantic memory. Nevertheless, in a pathology primarily characterized by a semantic memory disorder categorical- and modality- specific effects are not obvious findings. In fact, there is no clear evidence of categorical effects, at least concerning two broad categories of knowledge, that is, natural items and artifacts. Furthermore, transmodal deficits do not seem to be the rule in SD. Also quite robust is the observation that some conceptual domains are relatively spared in this pathology, that is, numerical knowledge, abstract words, and action verbs. Aims: To explore category specific and modality specific deficit in SD and to support the evidence that semantic degradation in SD primarily involves knowledge of the objects in the real world, whereas categories of knowledge whose items can be less easily identified by surface attributes, such as verbs, numbers and body parts, are more preserved. Methods and Procedures: We investigated the semantic impairment in 8 patients with Semantic Dementia (SD). Voxel-based morphometry (VBM) in each patient was also obtained Outcomes and Results: In some patients manmade objects were significantly more preserved than natural items, verbs more preserved than nouns and the number system entirely preserved; the body parts category was the least impaired in all subjects; finally, in three patients visual semantic knowledge was significantly more preserved than verbal semantic knowledge. VBM showed that atrophy of the anterior inferior temporal regions was insufficient to impair knowledge about verbs, numbers and body parts, whose impairment was associated with more widespread atrophy. In subjects whose verbal semantic knowledge was significantly more impaired than visual semantic knowledge, atrophy was principally distributed in the left hemisphere. In patients with significant impairment for natural items compared to manmade objects, atrophy was not confine in the temporal lobes. Conclusion: We conclude that in SD semantic decay primarily involves the real-world items whose knowledge is processed by surface sensorifunctional features and that this is the type of knowledge stored in the temporal lobes. Our data support a model that associates a semantic hub with modality/category specific neural substrates. [ABSTRACT FROM AUTHOR]