Impairment in category fluency in ischemic vascular dementia.

Drexel University and Crozer-Chester Medical Center The underlying mechanisms for impaired output on letter (F, A, and S) and category (e.g., animal) word list generation (WLG) tasks in subcortical ischemic vascular dementia (1VD) were investigated. Normal control (NC) and Alzheimer's disease (AD) participants were also studied. IVD and NC participants performed better on category than letter WLG tasks, whereas the opposite was observed among AD participants. IVD participants produced fewer responses than AD participants on letter WLG tasks, but there was no difference between AD and IVD participants on the "animal" WLG task. AD participants scored lower than 1VD and NC participants on animal WLG indexes measuring semantic knowledge. There were few differences between IVD and NC participants. The reduced output on the animal WLG task for IVD participants is consistent with search-retrieval deficits. The reduced output of AD participants may be caused by degraded semantic knowledge. Tests of letter and category word list generation (WLG) have been extensively used to study various aspects of cognitive functioning in both normal aging and a wide range of dementing illnesses. In an attempt to investigate the underlying mechanisms for successful performance on cat- egory WLG tasks, Gruenewald and Lockhead (1980) admin- istered the animal WLG task to normal control (NC) participants. Two findings emerged from this study. First, the authors observed that clusters of semantically related re- sponses (i.e., responses that shared many attributes) were produced relatively quickly. Second, Gruenewald and Lock- head (1980) observed that participants" shifted" or switched into other semantic clusters only after a comparatively long response latency. They proposed a two-component model to describe the performance of normal participants on category WLG tasks and concluded that the ability to generate successive clusters of semantically meaningful responses is independent from the ability to search and shift from one cluster to another. Category WLG tasks have been used to investigate the semantic knowledge deficits associated with Alzheimer's disease (AD). For example, it has been shown that output on semantically related WLG tasks declines as the illness Tania Giovannetti Carew, Melissa Lamar, and David J. Libon, Neuropsychology Program, Drexel University, and the Neuropsy- chology Service, Department of Psychiatry, Crozer-Chester Medi- cal Center, Upland, Pennsylvania; Blaine S. Cloud, Department of Neurology, the Graduate Hospital, Philadelphia, Pennsylvania; Murray Grossman, Department of Neurology, Hospital of the University of Pennsylvania, Philadelphia, Pennsylvania. This work was supported, in part, by the U.S. Public Health Service (AG09399). We thank Gail Silverstein, Laurel Buxbaum, Myrna Schwartz, and Laura P. Sands for their helpful comments. Correspondence concerning this article should be addressed to David J. Libon, Neuropsychology Service, Department of Psychia- try, Crozer-Chester Medical Center, Upland, Pennsylvania 19013. advances (Ober, Dronkers, Koss, Delis, & Friedland, 1986; Rosen, 1980). In addition, AD participants tend to produce fewer responses on semantically based WLG tasks than phonemically based WLG tasks when compared to NC participants (Butters, Granholm, Salmon, Grant, & Wolfe, 1987; Weingartner, Kawas, Rawlings, & Shapiro, 1993). AD participants have also been shown to make more category violations on semantic WLG tasks and to produce fewer responses per cluster (i.e., farm animals, fruits, and vegetables) than NC participants (Binetti et al., 1995; Martin & Fedio, 1983; Mickanin, Grossman, Onishi, Auriacombe, & Clark, 1994; Ober et al., 1986). Whether these findings are due to an actual loss or degradation of semantic information (Chan et al., 1993; Chertkow & Bub, 1990; Hodges, Salmon, & Butters, 1992; Martin, 1992) or to an inability to access semantic information that may be largely intact (Bonilla & Johnson, 1995; Nebes, 1989; Nebes, Martin, & Horn, 1984; Nebes, Brady, & Huff, 1989) has been hotly debated. A different pattern of performance on letter and category WLG tasks has been reported among participants suffering from dementing illnesses known to disproportionately im- pair executive control functions, such as Huntington's disease (HD), Parkinson's disease (PD), and progressive supranuclear palsy (PSP). For example, over time, HD and PSP participants exhibit greater decline on phonemically based rather than semantically based WLG tasks, a pattern of performance opposite from that found in AD. Similarly, demented participants with PD have been shown to produce more responses on semantic as opposed to phonemic WLG tasks (Bayles, Trosset, Tomoeda, Montgomery, & Wilson, 1993; Beatty, Staton, Weir, Monson, & Whitaker, 1989). Comparisons between AD, HD, and PSP participants have shown that HD and PSP participants produce fewer re- sponses on phonemic WLG tasks. On semantic WLG tasks, however, HD, PSE andAD participants are equally impaired (Butters et al., 1987; Hodges, Salmon, & Butters, 1990; Rosser & Hodges, 1994; Monsch et al., 1994). 400