Charting the progression in semantic dementia: Implications for the organisation of semantic memory

John R. Hodges,Naida Graham,Karalyn Patterson

DOI: https://doi.org/10.1080/09658219508253161

1995-09-01

Memory

Abstract:A patient, JL, with the syndrome of semantic dementia was assessed longitudinally over a two-year period. The data presented here address the controversy concerning the hierarchical organisation of semantic memory. On a range of category fluency tests, when first tested JL was just within the normal range on the broadest categories of animals and household items, but was virtually unable to produce any instances of specific categories such as breeds of dog or musical instruments. Longitudinal fluency data for the animal category demonstrate that while JL continued to produce the most prototypical responses (cat, dog, horse), other animal labels dropped out early from his vocabulary. On the picture-sorting tests from our semantic memory test battery, JL's discrimination between living things and man-made objects was preserved for a substantial time in conjunction with a marked decline in his sorting ability for more specific categories, particularly features or attributes (e.g. size, foreign-ness, or ferocity of animals). An analysis of naming responses to the 260 Snodgrass and Vanderwart pictures on four occasions suggests a progressive loss of the features of semantic representations that enable discrimination between specific category instances. There was a progressive decline in circumlocutory and category co-ordinate responses with a rise in broad superordinate and cross-category errors. The latter are of particular theoretical interest; on session I, all cross-category errors respected the living/man-made distinction, but by session IV almost half of such errors failed to respect this distinction. The emergence of category prototypes was another notable feature, particularly in the living domain: at one stage, land (or four-legged) animals were all named either cat, dog, or horse. By contrast, within the man-made domain, items were frequently described in terms of their broad use or function, until eventually no defining features were produced. These findings are discussed in the context of competing theories of semantic organisation.

psychology, experimental