Abstract
Dissociations between different aspects of learning and memory in patients with amnesic syndromes and various neurodegenerative diseases has led to a fundamental revision of the taxonomy of human memory, and a greater understanding of its neurobiological basis. It has been suggested that, while lesions to temporal and diencephalic structures leave procedural learning intact, neurological disorders involving the striatum lead to deficits in this aspect of learning. However, as comparative evidence has accumulated on a variety of procedural tasks, it has become clear that different neurological diseases lead to varying patterns of impaired and intact performance. This has led to a revision of the concept of a unified procedural system in favour of parallel, independent systems, only some of which involve the striatum. There is a danger, however, that the effort to fit the patterns of functional association and dissociation into a neuroanatomical framework may be premature, It is suggested that the way forward lies in more comparative studies, but ones which (1) adopt a broader perspective, assessing other patient groups or single cases, (2) tackle the question of inter-subject variability, (3) adopt a more analytic approach to the assessment of individual performance, and which utilise theory-driven methods to assess the processes of learning at different stages, and (4) which employ complementary methods such as pharmacological manipulations in normal subjects to assess the neurochemical basis of procedural knowledge.