REM and memory consolidation

consolidation of memories - being the enhancement and stabilization following encoding of information - primarily occurs during sleep. there are, as to be expected, several propositions as to the manner in which this takes place.

the medial temporal lobe, including the hippocampal system, is said to mediate consolidation through enhancing innervation to the neocortex. it is proposed that this occurs in several possible mechanisms:

1) the MTL system consolidates by signaling the neocortex to form a new representation of information, and the neocortex subsequently imprints and stores

2) the MTL is the rehearsal mechanism, strengthening connections with the particular cortical regions so they are all activated when the experience responsible for the information input repeats in the future

3) the MTL encodes AND imprints memories into the neocortex. this final option suggests that once enhanced, the neocortex is the final repository for memories. long-term memory is then distributed by coactivation through the higher neocortices specialized for analysis, each contributing differently to the storage of the complete memory. in this way, the neocortex is primed to reconstruct the representation of the information from partial cues.

regardless of which of these mechanisms is more correct than the other, their common function depends on adequate sleep cycles. it has been shown that not only does consolidation occur primarily during slow-wave and REM sleep (Kandel 2001, Sei et al 2000), but also that MTL activity during the information encoding process decreases dramatically in sleep deprived subjects (Drummond et al 2000). both of these observations correlate with decreased cortical metabolism and reproducible memory deficit.

disruption of encoding and deficit of retention in sleep-deprived subjects map onto particular physiological correlates that pronounce REM sleep as crucial for the memory consolidation process. REM deprivation reduces excitability of hippocampal neurons responsible for imprinting information into the neocortex (Kandel). this, as you might imagine, directly impairs long-term potentiation (LTP), or decays any LTP that does occur [ aka, early-LTP is vulnerable to decay within 90 min if not pushed into late-LTP by excited hippocampal neurons which facilitate the genetic transcription of proteins which are biochemically responsible for the actual "long term"]. in addition to LTP, acetylcholine (ACh) accompanying slow-wave sleep just prior to REM mediates consolidation (Power 2004, see neato pic).

the other important gesture sustained by REM sleep is the production of several nerve growth factors, primarily NGF and BDNF. these are both key regulators of LTP (again referring to the strengthening of neural circuit connections in learning and memory formation). this means that their reduction in REM sleep-deprived subjects probably causes the impairment in memory consolidation - supported by the most drastic deficits occurring in the hippocampus.

apparently, episodic memory is most profoundly affected by REM sleep-deprivation... but we don't know why (Rauchs et al 2004). although it wouldn't surprise me in the least if it had something to do with consolidation involving the prefrontal cortex (which has been suggested by almost everyone in the field to be the primary region of activation and most distinguishable candidate during episodic memory retrieval, and also during MTL mediated consolidation (see entry on "episodic memory and autonoetic awareness")). both the hippocampus and pfc are main players in episodic memory encoding and retrieval (id est, emotional memory depicting information in the context which it was learned). the other half of declarative memory, semantic, seems to involve a much different array of hippocampal efferent circuits. and although i don't actually know what the profiles of ACh, NGF or BDNF are in those circuits... i would imagine that they're also different from the activity seen in the circuits associated with episodic consolidation.

ergo, REM sleep: fighting amnesia since the evolution of the higher limbic system.

P.S.

do dreams arise from the stochastic hippocampal inputs to the neocortex?

did dreams evolve from the prolonged quiescent state in reptiles to promote calcium-dependent memory consolidation?

o.O