This paper fits with the notion I've always had about the hippocampus, which is that it participates in the sequential linking, and subsequent retrieval of, episodic memories. This property provides the so-called "remembered present" quale of consciousness, and may be the mechanism by which people produce spatio-temporally extended imaginings.

So why should patients with hippocampal damage have difficulty encoding new non-episodic information? Names and the like. Is it because experiences have to be saved in an episodic buffer and then slowly stripped down to their components to be stored extra-hippocampally?

I remember hearing once that the right hippocampus is involved in encoding semantic info, the left, spatial-episodic. So I should have said the left hippocampus. Your point is well-taken though.

earlier today i was thinking that maybe it would be okay if we never had a grand unifying theory of the hippocampus. what if it actually does more than one thing? it's pretty complicated circuitry, it can have different oscillatory states, and it has differentiated subregions along every axis. still you'd like to be able to predict the effects of perturbations. it's gonna take a pretty wide-ranging integrative mind/model to get the thing under control.

An important thing to remember about the HF is that its connectivity and neurochemistry is very different along the dorso-ventral axis. The differential connectivity of the dorsal and ventral hippocampi has been very well elucidated by LW Swanson in his work with Risold. One neurochemical difference that comes to mind is the differential expression of estrogen receptors. In the dorsal CA1, nuclear estrogen receptors are found primarily within inhibitory interneurons. In the ventral CA1, they are found predominantly within pyramidals. Consequently, the level of ERalpha expression is much higher within the ventral HF. And this fits quite well with the connectivity, since the ventral HF projects mainly to the ventral forebrain and hypothalamus-- the home of many other steroid-sensitive nuclei.

The reason I mention this is because it agrees with your point-- that unless you take into account the massive differences in connectivity between the north and south HF, there never will be a unified account of the entire hippocampus. Most people, including myself, think of it as a region involved in declarative memory consolidation. Yet I doubt that the ventral HF's connectivity with the hypothalamus has much to do with that!

Considering the well-demonstrated role of the hippocampus in spatial processing and navigation (in addition to memory, both spatial and non-spatial), I wouldn't be surprised if some of the lack of detail in this imagination of the museum has more to do with a lack of being able to set down a mental "coordinate system" on which to draw the layout of the museum, than a lack of imagining events per se. I think it's useless to say "hippocampus=memory" when these other roles of that brain region are very plausible confounding factors.

I would find it frustrating to be in that patient's position, but in fact I have the opposite situation. I find it very easy to imagine complex scenes that look nothing like what I've seen before. As a demonstration, my dreams are full of stunning landscape layouts, interesting buildings, etc. On the other hand, they often don't have much plot and frequently lack many people (besides as sort of disembodied voices I hear). I also have a good sense of direction, and a good memory for certain (but not all) things, so maybe part of my hippocampus is on the overactive side.