Introduction

The human brain is not only a machine that processes many kinds of information, it is also an information storage system. The plasticity of the brain indicated by its ability to alter the synaptic connections between its neurons in several ways (see Rose, 1992) suggests that most regions of the brain, if not all, are capable of storing memories. There is a view, widely held by neuroscientists, that information is stored where it is processed, or, more precisely, represented, by the brain (for example, see Squire, 1987). Although there is relatively little direct evidence for this view, it is plausible to believe that, when a neuronal system is active because it is representing information, this activity will give rise to synaptic changes selectively within the neuronal system so that this information can be more readily re-represented from appropriate cues on future occasions. There is also overwhelming evidence that different brain regions process and represent different kinds of information (for example, see Kolb & Whishaw, 1990 passim), which implies that different kinds of information will be stored in different parts of the brain rather than in the same specialized ‘memory storage system’. As there is good evidence for this implication (see Squire, 1987; Mayes, 1988), the view that information is stored where it is represented has considerable indirect support.

Until about 20 years ago most people assumed that memory (at least long-term memory) was a monolithic system, so that there was only one kind of memory that needed to be understood.