Nature of the Functional Loss in Amnesia: Possible Role for a Highly Structured Neural Network

Abstract

Organic amnesia is a condition in which brain damage to structures in the medial temporal lobes, midline diencephalon or basal forebrain impairs the ability to recall or recognize recently experienced facts or episodes (anterograde amnesia) and also the ability to recall and recognize facts and episodes, memories for which may have been formed normally up to decades before the onset of brain damage. Despite these impairments, which can be very severe in some patients, many amnesics show preserved intelligence and short-term memory. Amnesics therefore show an impairment that is specific to certain kinds of memory, leaving other kinds of memory and cognitive function intact. In this paper, the precise nature of the preserved and impaired functions will first be described in more detail in order to facilitate an appropriate characterization of the disturbed function(s) and to help determine whether patients are suffering from only one functional deficit or several independent functional deficits. Work that is concerned with identifying the structures, damage to which is critical in producing the syndrome, will then be briefly reviewed. The anatomy and physiology of the critical structures will then be outlined and the nature of their informational inputs and outputs briefly considered. Finally, the conditions that must be met by a neural network model that can produce the kinds of memory that are deficient in amnesics will be discussed.