Demonstrating the qualitative differences between semantic aphasia and semantic dementia: a novel exploration of nonverbal semantic processing.

Semantic dementia (SD) implicates the anterior temporal lobes (ATL) as a critical substrate for semantic memory. Multi-modal semantic impairment can also be a feature of post-stroke aphasia (referred to here as "semantic aphasia" or SA) where patients show impaired regulatory control accompanied by lesions to the frontal and/or temporo-parietal cortices, and thus the two patient groups demonstrate qualitatively different patterns of semantic impairment [1]. Previous comparisons of these two patient groups have tended to focus on verbal receptive tasks. Accordingly, this study investigated nonverbal receptive abilities via a comparison of reality decision judgements in SD and SA. Pictures of objects were presented alongside non-real distracters whose features were altered to make them more/less plausible for the semantic category. The results highlighted a number of critical differences between the two groups. Compared to SD patients, SA patients: (1) were relatively unimpaired on the two alternative forced choice (2AFC) decisions despite showing a comparable degree of semantic impairment on other assessments; (2) showed minimal effects of the plausibility manipulation; (3) were strongly influenced by variations in the regulatory requirements of tasks; and (4) exhibited a reversed effect of familiarity - i.e., better performance on less commonly encountered items. These results support a distinction between semantic impairments which arise from impaired regulatory processes (e.g., SA) versus those where degraded semantic knowledge is the causal factor (e.g., SD). SA patients performed relatively well because the task structure reduced the requirement for internally generated control. In contrast, SD patients performed poorly because their degraded knowledge did not allow the fine-grained distinctions required to complete the task.