Abstract
Identification deficits in individuals with Alzheimer's disease (AD) can target specific categories of objects, leaving other categories unaffected. Using line drawings to investigate naming deficits unfortunately carries severe limitations because object form is yoked to object meaning. We therefore used computer-generated stimuli with empirically specifiable properties in a naming paradigm that decouples object form and object meaning. We recruited individuals with Alzheimer's disease, mild cognitive impairment (MCI), as well as healthy matched controls, and asked them to learn to label visually similar and visually distinct novel shapes with the names of semantically similar items (leopard, lion, tiger) or semantically distinct items (donkey, frog, robin). The results revealed that participants confused visually similar novel shapes more often than visually distinct shapes objects, and that they confused shapes labelled with the names of semantically similar items more often than when the same shapes were labelled with the names of semantically distinct items. These main effects were further qualified by interactions with participant group. Differences in confusion errors between visually similar and visually distinct shapes were greatest for AD participants and smallest for healthy age-matched controls. Similarly, differences in confusion errors between shapes associated with semantically similar and semantically distinct labels were greatest for AD participants and smallest for healthy age-matched controls. In conclusion, both visual and semantic similarity contributed to the identification errors of participants. However, the impact of each variable was always smallest for the healthy controls and largest for the AD participants, suggesting a continuum of severity.