No abstract

Some forms of memory have been shown to depend on a system of medial temporal lobe stuctlres that includes the hippocampus and the adjacent cortical areas (entorhinal, perirhinal, and parahippocampal cortex). The role of this system is only temporary, however, as indicated by the fact that, after damage to the medial temporal lobe, recent memories are impaired but very remote memories are Intact. Here we review the evidence that the medial temporal lobe memory system is involved in a process of consolidation: memories are initially dependent on this system but gradually become established in other areas of the brain. We then review some of the ideas that have been proposed about the phenomenon of consolidation and suggest a synthesis of these views. Finally, we describe a simple neural network model that captures some key features of consolidation.The importance of the medial temporal lobe (MTL) for human memory was established in the 1950s, when surgical removal of this region was found to produce a profound and selective memory impairment (1). Subsequent work, using a model of human amnesia in the monkey (2, 3), identified the anatomical components of this brain system-i.e., the hippocampus (together with the dentate gyrus and subiculum) and the adjacent, anatomically related entorhinal, perirhinal, and parahippocampal cortices. Continuing studies in humans, monkeys, and rodents have illuminated how this system contributes to memory functions (for review, see ref. 4).One important finding is that the MTL (and related structures in the diencephalic midline) is involved in a limited domain oflearning and memory. These structures support the capacity for conscious recollections of facts and events (i.e., declarative memory) but are not necessary for various nonconscious (nondeclarative) forms of memory that are expressed through performance, including skills and habits, simple forms ofconditioning, and the phenomenon ofpriming (5-7). Retrograde Amnesia and Memory ConsolidationA second important finding (and the focus of this article) is that the role of the MTL is only temporary. This conclusion is based on the fact that damage to the MTL produces temporally graded retrograde amnesia. Memory for events that occurred a short time before the damage is impaired, but memory for remote events (events that occurred a long time before the damage) is spared. This phenomenon was recognized more than a century ago as central to understanding memory and the brain (8). Since that time, temporally graded retrograde amnesia has been observed repeatedly in rodents and humans using a variety of memory tasks and disruptive treatments (9)(10)(11) Thus, as time passes after learning, there must be gradual reorganization (consolidation) of memory storage, whereby memories that are initially dependent on the MTL eventually do not require this system. A more permanent memory that is independent ofthe MTL develops gradually, presumably in neocortex (2,6,21).An alternative view is that consolidation occurs gradually but independ...

We thank D. G. Amaral for helpful discussions and R. P Glower for assisting in the development of the surgical procedure used for making the lesions in the hippocampal region. We also thank C. M. L&lair and A. K. Lockwood for behavioral testing and technical assistance and S. Ramus and N. Rempel for assistance in histological processing and analysis of the brain tissue. A research protocol describing all aspects of the present study that related to the use of animals (care and maintenance, surgery, behavioral testing, and euthanasia) was approved by the Animal Research Committees of the V.A. Medical Center,

Debate continues on whether the role of rodent hippocampus in memory is limited to the spatial domain. Recently, this controversy has been addressed with studies on the social transmission of food preference, an odor-odor association task with no spatial requirements. Multiple reports have concluded that damage to the hippocampal region impairs memory in this task, but there remain questions about the extent of damage essential to produce an impairment. Furthermore, a recent study (Burton et al. 2000) found no effect of hippocampal lesions on memory in this task. We tested animals with complete lesions of the hippocampus (H) lesions of the hippocampus plus subiculum (HS), and lesions of the adjacent, anatomically related cortices of the parahippocampal region (PHR). H lesions produced an impairment on spatial delayed alternation, but not on memory for the social transmission of food preference, whereas HS and PHR lesions produced severe and equivalent impairments on memory for the socially acquired food preference. We discuss possible explanations for the discrepancy with the results of Burton et al. (2000) and conclude that the hippocampus and subiculum together play a critical role in the formation of this form of nonspatial, relational memory.