Abstract
The remarkably widespread belief that people with dementia are incapable of new learning is manifestly false. Results are briefly reviewed of experimental list-learning findings, based on the levels of processing framework, which suggest ways to optimise a grossly impaired but still residual capacity to learn limited amounts of new information in mild to moderate dementia. Studies of a more clinical bent are dealt with in some detail, demonstrating that is possible to utilise the methods of spaced retrieval and fading cues to train persons with dementia to associate a behaviour with contextual cues, and retain the information over time. Two case studies are also presented to illustrate clinical application to serious behavioural problems in dementia. It is suggested that research such as this, at the problem-specific level, may be more relevant for cognitive and behavioural scientists at this stage of the research enterprise than attempts, so far unsuccessful, to produce generalised cognitive improvement of any practical value in dementia.