Measuring cognitive processes underlying picture naming in Alzheimer's and cerebrovascular dementia: a general processing tree approach.

Existing data from the Boston Naming Test were analyzed using standard statistical methods and with a General Processing Tree (GPT) model in an attempt to differentiate between patients with Alzheimer's Disease (AD) and Cerebrovascular dementia (CVD), matched for severity, and age-matched healthy controls. The GPT approach enables the estimation of parameters reflecting underlying cognitive processes (e.g., perceptual analysis, lexical access) based on categorical data. Compared to traditional analyses of proportion correct and of errors, the analysis with the GPT model was more sensitive in detecting differences between patient groups, as well as the source of these differences. Among the differences were two, evident in a comparison between very mild AD and very mild CVD patients: standard analyses did not reveal a significant difference between these groups; but the GPT analysis revealed that the CVD patients had significantly higher estimates than the AD patients on parameters reflecting lexical access and phonological realization.