Anders Sandberg
To study how the medial temporal lobe (MTL) and the cerebral neocortex (NCX) interacts, especially conceptual models for complementary learning systems and how they may support memory consolidation and context integration through ``fast'' and ``slow'' learning.
The medial temporal lobe system (including parahippocampal, perirhinal and entorhinal cortices as well as the hippocampal formation [ZMS93]) seems to be crucial for declarative long-term memory and memory consolidation, the process where declarative memories are gradually encoded into the neocortex [SA95] The MTL appears to be more active in a non-consolidated declarative memory state compared to a well encoded consolidated memory state, i.e. the interaction between the MTL and NCX seems to be dynamic and changing as a consequence of repeated activations of the neocortical representations [PEI]. There is a broad consensus that the NCX is the final storage site for declarative long-term memories (cf. for example [Fus94]). Presently, it is not clear what role the MTL plays in the consolidation and reinstatement processes, and there are many suggestions concerning the functional role of this system in long-term memory.
It has been suggested that the hippocampus has a role in spatiotemporal memory [ON78]. This is supported by the existence of place cells in the hippocampus which fire when the rats are in a certain location. In humans and other primiates it is still unclear if the hippocampus proper has an exclusive role in spatiotemporal memory but it is clear that the MTL has a role in both spatial/spatiotemporal memory and non-spatial memory.
One model proposes that the hippocampus binds together different representations into a ``chunk'', ``index'', or ``pointer'' which is later learned by the neocortex [Squ92,BWP95]; the function of the place cells, in this model, are to represent spatiotemporal context [MG97].
A different group of models propose that the hippocampus acts as intermediate storage of memory representations before it is transferred into the final long-term storage in the neocortex, an idea originally proposed by Marr [Mar71]. This could be done using a rehearsal mechanism that strengthens the connections between different cortical areas corresponding to the same original activation pattern [Mis90]. The MTL only needs to store an ``index'' pattern which allows the retrieval of the original pattern of cortical activation. This is similar to the pseudorehearsal model of Robins [Rob96].
There exists large differences in learning speed between neocortical memory and MTL memory, as is evidenced by the need for repetition in procedural learning and semantic learning rates in amnesiacs contrasted by the fact that episodic memory in general can be encoded with a single experience. For example Milner has suggested that the hippocampal synapses are ``soft'' and capable of changing quickly, both in learning and forgetting, while neocortical synapses are ``hard'' and change slowly in response to new stimuli and weaken slowly. The hippocampus can reactivate the cortical representations during recall, and through repeated activations create hard links that remain as long term memories long after the hippocampal representations have faded [Mil89,AS94]. Functional Neuroimaging data consistent with this position has recently been provided by [PEI].
This model of complementary memory systems based on the necortical-MTL interactions fits together with the conceptual models of McClelland et al. [MMO95] where the difference in plasticity is proposed as a solution to a fundamental problem for learning systems, the so called stability-plasticity dilemma or serial learning problem, which in some systems give rise to a catastrophic interference effect (where new learning interferes with already acquired older memories) which is observed in some simple memory models but not in humans [Rat90]. Catastrophic interference can be avoided within some frameworks, for example through rehearsal and pseudorehearsal which would correspond to the MTL storing information which is then gradually replayed to train or taught back to the NCX, possibly during sleep [Rob96] or together with ongoing experience [MG97].
So far most models of the MTL has been purely conceptual and with little ties to biological realism or concentrated on low-level properties of the hippocampal network. This project intends to study conceptual models of the MTL and its interactions with neocortex with an intention to work towards biological plausibility, and to test the viability of the hypothesis of of complementary memory systems.
Another important aspects of memory function and memory consolidation are how these functions are influenced by for example relevance of information to the subject or the emotional state of the subject, attention, working memory and control processes.
It is known that various modulatory brain systems, like cholinergic or monoaminergic transmittor systems as well as certain endocrine systems [MSW91]. The MTL seems ideally placed for influencing associative strengths due to its limbic and mesencephalic connections which may influence consolidation and the level of internal plasticity. It is then natural to keep an eye open for if these forms of memory function can be accommodated in the model of complementary memory systems.
Presentation as a seminar and/or essay.
Develop a model for MTL-neocortical interaction and consolidation with different learning time constants.
Presentation as an article.
Presentation as an article.
Sleep, limbic system, cholinergic pathways, frontal lobe memory systems.
This document was generated using the LaTeX2HTML translator Version 97.1 (release) (July 13th, 1997)
Copyright © 1993, 1994, 1995, 1996, 1997, Nikos Drakos, Computer Based Learning Unit, University of Leeds.
The command line arguments were:
latex2html -split 0 project.tex.
The translation was initiated by Anders Sandberg on 10/22/1997