function of hippocampus

[Ali Minai]

	From trengove at socs.uts.EDU.AU  Thu Sep  3 04:37:08 1998
	X-Sender: trengove at linus
	Subject: Re: function of hippocampus
	MIME-Version: 1.0

	  ...the above quotes concerning the role of hippocampus in categories of
	memory such as episodic memory as well as tasks such as spatial cognition
	suggests to me we consider from a top down, psychological standpoint why
	spatial cognition and episodic memory should be tied together
	functionally, and hence why we shouldn't be surprised that a single part
	of the brain is involved in both.
	  From my own subjective observations, if I am trying to remember a
	thought that I had, or a particular piece of information that came up in a
	conversation, often the best way to proceed is for me to remember where I
	was when I had the thought.  Once I have remembered the place where I had
	the thought I have access to a rich pool of cues that can help to trigger
	the particular thought I'm after.
	  It thus makes sense to me that spatial cognition should be involved in
	the laying down of new memories.  In the course of a day I will have
	perhaps hundreds of disctinct cognitive experiences to remember, but the
	number of distinct _places_ in which I dwell whilst having those
	experiences is likely to be at least an order of magnitude smaller.
	  Thus it makes good sense to organise memory around the memories of the
	places one has been during the course of a day.

There has been debate among hippocampal theorists (mainly in the context of
the rodent hippocampus) about whether the hippocampus is dedicated solely or
predominantly to spatial cognition. The alternative --- advanced most prominently
by Howard Eichenbaum and co-workers --- is that the hippocampus helps construct
memories involving complex relationships between cues, conditions, contexts, etc.
(relational memory). In this view, purely spatial representations, such as the
cognitive map of O'Keefe and Nadel, are special cases of relational memory of
complex episodes.

Let us think of an episode as a spatio-temporal structure in some very
high-dimensional space, where the dimensions correspond to sensory cues,
features, contexts, motivations, internal states, etc. In any particular
case, most of the possible dimensions will be irrelevant, and the episodic
representation will lie in a subspace of the full space of possibilities.
When this subspace is purely spatial, we see a place representation. When
the subspace is predominantly spatial but also includes other dimensions,
we see conditional place representations (e.g., context-dependent,
directional, reward-related, etc.) In particular cases --- e.g., experiments
designed to make place irrelevant --- we would see non-spatial representations
(e.g., in Eichenbaum's olfactory experiments). In general, as the quotation
above points out, place is an extremely important component of any episodic
experience, and it would not be surprising to find strong place-dependencies
in any neural representation of episodic memory. I believe that most
hippocampal theory now implicitly acknowledges this.

Theories which see the hippocampus as primarily involved in spatial cognition
are based on rodent data with its place and head-direction cells. There is no
question that, in many instances, one is hard pressed to find any correlate
of a hippocampal pyramidal cell's activity other than location. However, in
strongly directed environments (e.g., arm mazes), direction becomes a factor
too. And several reports have demonstrated that cells fire in response to
task contingencies when these are made important. It is tempting to think that,
given a rat's limited mental life (am I being speceist:-), place is almost
the entire default sub-space of experience, except at particular times. Thus,
episodic representations appear --- especially when measured one cell at a
time --- as place representations. In higher animals such as primates, episodes
have much richer content, and space is only a part of the picture --- hence
the absence of place cells. Interestingly, one does find view cells in primates
(e.g., in the work of Rolls and co-workers) which fire when the animal is looking
at a particular scene. Perhaps this means that, for primates, what is seen during
an episode is more significant than where one is located. Also, it is possible
that primates have a better ability to project experience to places that they
can see but where they are not currently located.


        ......  I believe one idea in
	circulation e.g. discussed by Rolls and coworkers, is that the hippocampus
	provides a short term 'buffer' for storing memories, which are later
	'transferred' to the neocortex for long term storage.

This idea has been around for a while in various forms. In terms of modeling,
I think the work by Gluck and Mayers, Squire and Alvarez, O'Reilly, McClelland
and McNaughton, and recently, by Redish and Touretzky, offers interesting
perspectives.


	  Back to the specific idea given above, I'm curious whether specialists
	of the hippocampus find it (a) highly dubious, implausible or naive; or
	(b) too obvious to be worth mentioning; or (c) a potentially useful way
	to look at the role of the hippocampus.

Without claiming specialist status, my answer is (c), provided we think of the
big picture that includes the whole constellation of available data. I believe
strongly that big theoretical ideas --- such as the cognitive mapping theory
--- are crucial to our understanding of neural function, even when the theories
turn out to be less than perfect in the end.

Ali Minai


-----------------------------------------------------------------------------
Ali A. Minai
Assistant Professor
Complex Adaptive Systems Laboratory
Department of Electrical
   & Computer Engineering
               and Computer Science
University of Cincinnati
Cincinnati, OH 45221-0030

Phone: (513) 556-4783
Fax:   (513) 556-7326
Email: Ali.Minai at uc.edu

Internet: http://www.ececs.uc.edu/~aminai/