User:Graeme E. Smith/Collections/Model Series/Datamining/Redescription and Memory Lag
Redescription and Memory Lag
[edit | edit source]
Annette Karmiloff-Smith introduced the idea of memory phases in her book Beyond Modularity: A Developmental Perspective on Cognitive Science another concept she introduced was the idea that one phase of memory had to complete before the next could begin, and that completing a phase of memory took around 2 years at least in language formation in childhood. This jibes with learning theories that suggest that after learning a fact, the integration of that fact into our knowledge can be traced over time, and takes as long as 2.5 years to complete. After about 2.5 years all we have learned about a fact is stabilized and no new learning can be traced, unless a new fact is learned that impacts on the first fact.
She hypothesized that this phase dependency of memory was due to something called Representational Redescription, that each phase corresponded to the emergence of a completed redescription of a former state of representation. Until the redescription was completed, the next phase of learning did not start. That the redescription took an average of about 2 years for many different types of learning during childhood, suggested to me at least that the retrograde Amnesia found in people with injuries of the hippocampus, might indicate a role for the hippocampus in learning.
Meta-Index Representation in Cerebral Cortex
[edit | edit source]
The fact that the retrograde amnesia was not total in the previous chapter, however suggested that a representation of the meta-index existed after the hippocampus was damaged. This suggested that the Lag between the Learning of a fact, and its eventual complete storage, included about a 2 year lag associated with the redescription of the symbolic storage in the Meta-Index, into a Cerebral Cortex representation of the Meta-index, possibly in the Inferior Temporal Cortex area. The other side of this was the dependency of learning on the completion of the new representation. Why would learning not be complete until the index was archived? After all, the Meta-Index was available in healthy organisms to be used.
The answer I thought, was that something had to happen to update the implicit memory, as a result of new relationships between information found in the Meta-Index, but not otherwise transferred back to the Cerebral Cortex, possibly because they were processed solely in the Meta-Index and did not require rehearsal of the actual contents of the memories. During the Archiving of the Meta-Index, the mechanism of archiving obviously resulted in rehearsal of the new relationships, and thus, triggered storage of the CLUMPS that were needed to process the relationships. It was only once we had new Meta-Index representations of the new relationships that we could process them, and find out more information about the relationships that were deeper in the environment and thus not immediately obvious.
Learning/Memory Lags
[edit | edit source]
Since this was not exactly the interpretation that Annette Karmiloff-Smith suggested, but explained the link between redescription and memory lag better than her theory (I thought),
It made sense to me, that learning lags are likely the result of Redescription, and that each type of redescription would be associated with its own lag, although some forms of redescription might be associated with more than one lag, especially if they involved sleep stages.
The lag of redescription from implicit to explicit form may only be milliseconds, but it impacts on short term memory size. The lag of redescription from explicit to declarative form may involve the refinement of memory into gradually smaller descriptive clumps, and then the mapping of those clumps on the index/maps of the Meta-Index which might require multiple passes in order to index the same information in multiple indexes. This might translate out to a 2 hour lag for express indexing, and possibly a 1 to 2 day lag for complete indexing. (if the less important indexing were offloaded into sleep periods such as REM sleep.) But the full force of learning would not be experienced until the completion of the redescription of the Meta-Index probably also requiring some sleep phase to complete, some 2.5 years after the memories were laid down. Other terms might be found to associate with other forms of redescription, and some of my terms may be skewed from known values, but I merely wanted to introduce the idea that most memory lags in learning, are associated with different forms of redescription.
Beauty
[edit | edit source]
To really make this model work, there needed to be some sort of meta-cognition that limited the learning of a new phase until the model of the previous form of representation was completed. This would probably include an impetus to complete the model, and a sort of brake that kept the model from being relied upon until it was completed. To add to that, there would be an emotional hook, a sense of beauty, perhaps that would attract us to use the completed model in spite of alternate and presumably less impressive models that we were more familiar with.