|
pubmed:abstractText |
We examined the interaction of content and process in categorizing novel semantic material. We taught patients with Alzheimer's disease (AD) and healthy age-matched seniors a category of plausible novel tools by similarity- and rule-based processes, and compared the results with our previous parallel study of categorization of novel animals, in which AD patients were selectively impaired at rule-based categorization. AD patients demonstrated learning in the novel tool study; however, in contrast to the novel animal study, they were impaired in similarity-based as well as rule-based categorization relative to healthy seniors. Healthy seniors' categorization strategies reflected process irrespective of category content; they frequently attended to a single feature following similarity-based training, and always attended to all requisite features following rule-based training. AD patients' categorization strategies, in contrast, reflected category content; they frequently attended to a single feature when categorizing novel animals by either categorization process, but rarely did so when categorizing novel tools. AD patients' ability to categorize novel tools correlated with preserved recognition memory, a pattern not found in the novel animal study. The category-specific role of memory, along with AD patients' performance profile, suggests content-specific distinctions between the categories. We posit that tool features are relatively arbitrary, placing greater demands on memory, while prior knowledge about animals such as constraints on appearance and feature diagnosticity facilitates the assimilation of novel animals into semantic memory. The results suggest that categorization processes are sensitive to category content, which influences AD patients' success at acquiring a new category.
|
