It looks like the prevailing view of what our hippocampus does is about to undergo a major revision. Greg Miller writes a review in Science Magazine about a PNAS article from Maguire's laboratory. The textbook view is that the main function of the hippocampus is to encode new memories, creating an initial memory trace that is eventually filed away to the cortex for long-term storage. In this view, the hippocampus is not needed to maintain or retrieve memories once they've been stored in the cortex. Maguire and colleagues examined five amnesic patients who had severe memory deficits caused by damage to the hippocampus, but no damage to the cortex; they had great difficulty forming new memories and recalling events that happened after their injuries.
The new and interesting result is that these patients, compared with normal controls, had difficulty envisioning commonplace scenarios they might reasonably have expected to encounter in the future. The healthy subjects provided rich descriptions, remarking for example on the curve of a beach, the sound of waves hitting the shore, and the feel of burning hot sand. The amnesic patients were able to follow the researchers' instructions, but their descriptions were far less vivid. They described fewer objects, fewer sensory details such as sounds and smells, and fewer thoughts or emotions that might be evoked in the imagined scenario.
This suggests that the same system we use to remember the past we also use to construct possible futures. The emerging view is that to have vivid constructions of the past, the future, or of imaginary events, you always need the hippocampus. A forthcoming report from Addis and Schacter of Harvard University in fact demonstrates that a similar network of brain regions, including the hippocampus, is activated when healthy volunteers are asked either to recall a vivid memory or to envision a future experience.