To Sleep, To ream, Perchance to Remember Jessica. Payne & Lynn Nadel ept. of Psychology, University of rizona Tucson, rizona W.J. Jacobs M. Moscovitch Richard ootzin ob Stickgold ruce McNaughton Thanks Stages of Sleep can the analysis of dreams tell us how sleep relates to memory consolidation? in the study of dreams, a major distinction has been drawn between REM and NREM REM is more prevalent in the latter part of the night NREM (SWS) is more prevalent in the early part of the night Structure of reams REM SLEEP NREM SLEEP REM reams highly fragmented with often bizarre content episodic elements, if any, are disconnected and hard to relate to waking life events NREM reams veridical episodic memories possible recent episodes are predominant; remote episodes appear but are less frequent NREM dreams late at night are often more REM-like N ream reports contain fragments that are hard to relate to specific episodes H N ream reports contain veridical episodic memories H (e.g. avallero et al., 1992; icogna, avallero & osinelli, 1986, 1991; Foulkes, 1962; Hobson et al., 2000; Stickgold et al., 2000, 2003) (e.g. uzsaki, 1996, 1998) (e.g. Sirota, sicsvari, uhl, & uzsaki, 2003) 1
Hippocampal-ortical Interactions: onsensus View Memory onsolidation post-registration events that stabilize memory subiculum short-term vs. long-term hippocampal - neocortical interaction as basis of longterm consolidation effects 1 G : N > H (to G) : H > H (in 3) : H > N (from 1) : N > N 3 Hippocampal-ortical Interactions: onsensus View ENHNEMENT OF MEMORY ONSOLITION S FUNTION OF SLEEP STGE subiculum REM NREM Episodic NO YES : N > H (to G) 1 G Procedural/ Implicit (e.g. Plihal & orn, 1997, 1999) YES (e.g. Plihal & orn, 1997, 1999; Wagner et al. 2002) (e.g. Plihal & orn, 1997, 1999) MYE (e.g Plihal & orn, 1997; Walker et al. 2002) : H > H (in 3) : H > N (from 1) : N > N 3 Sleep ondition Plihal & orn, 1997 slow-wave sleep group Percent improvement in recall after early or late retention interval. (Plihal & orn, 1997) failure of consolidation in REM learning Early sleep retrieval REM sleep group 50 50 sleep learning Late sleep retrieval 10:15pm 11:00pm 11:00pm 2:00am 2:15am 3:00am 3:00am 6:00am 6:15am 7:00am 40 30 20 10 sleep group wake group 40 30 20 10 sleep group wake group Wake ontrol ondition 0 early (SWS) late (REM) 0 early (SWS) late (REM) learning Wakefulness retrieval Retention interval Retention interval sleep learning Wakefulness retrieval 10:15pm 11:00pm 11:00pm 2:00am 2:15am 3:00am 3:00am 6:00am 6:15am 7:00am Paired-associate list (episodic) Mirror tracing (procedural) 2
What accounts for: differences in episodic content of dreams differential effects on memory consolidation N-->H REM Hasselmo (1999), uzsaki (1996) NREM Sirota, sicsvari, uhl & uzsaki (2003) differences in hippocampal-neocortical interactions as a function of: REM/NREM Early vs Late Sleep Variations in Neurotransmitters H-->H H-->N N-->N Louie & Wilson (2001), Hirase et al (2001) Hasselmo (1999) (Hippocampal output diminished during REM sleep). Hasselmo (1999) Wilson & McNaughton (1994); Pavlides & Winston (1989); Lee & Wilson (2002); Hirase et al (2001) Sirota, sicsvari, uhl & uzsaki (2003), uzsaki (1996) Hasselmo (1999) Neuromodulation across the Sleep-Wake ycle h NE 5-HT ctive Wake Quiet Wake SWS h: acetylcholine NE: norepinephrine (noradrenalin) 5HT: serotonin REM UT none of these dichotomies seem to capture all the results for example, REM-like dreams can be seen in NREM late in the night something else must be involved New Possibility GLUOORTIOIS glucocorticoid receptors in hippocampus ourtesy of ruce McEwen http://www.sfn.org/w/presentation/zpresentation entate gyrus fields 3
ORTIOSTERONE INING in RT Hippocampal subregions: Subiculum (S) 1 pyramidal cells 2 pyramidal cells 3 pyramidal cells 4 pyramidal cells entate gyrus (G) 405 ± 49 365 ± 20 572 ± 30 250 ± 61 192 ± 6 258 ± 58 Entorhinal cortex 100 ± 0.4 Lateral septal nucleus Medial septal nucleus 117 ± 8 40 ± 1 ortical amygdala Medial amygdala 77 ± 15 34 ± 1 Parietal cortex 27 ± 2 Stress response system The HP axis from Sapolsky, McEwen & Rainbow, rain Research, 1983. ose - Response Functionality HP MYG Level of Stress Latency to find target platform stress impairs LTP in 1 region of hippocampal slices stressed animals had LTP impairments 100 90 invisible target 80 70 stress involved 60 tail shocks while under restraint, then rats were sacrificed and slices taken 60 50 40 stressed animals given diazepam had LTP impairments visible target 30 20 10 0 Vis 1 Vis 2 cq 1 Group 1 cq 2 Group 2 cq 3 Group 3 cq 4 cq 5 Group 4 cq 6 Vis Group 5 Nadel, L., Thomas, K.G.F., Laurance, H.E., Skelton, R., Tal, T. & Jacobs, W.J, 1998. (Kim, Foy & Thompson, 1996) 4
Why ORT? cort variations with REM/NREM cort variations with early/late sleep *ortisol increases steadily across the night, with mini - peaks corresponding to REM episodes. From, Lavie (1996). Why ORT? Why ORT? cort variations with REM/NREM cort variations with early/late sleep cort modulates critical neurotransmitters cort variations with REM/NREM cort variations with early/late sleep cort modulates critical neurotransmitters similar phenomenology with cort variations in wake Episodic Memory: inding by ontext Episodic Memory under Stress habits/ procedures emotional procedural emotional features features meanings neocortex e.g. PTS meanings X neocortex disconnected fragments contexts hippocampus contexts hippocampus 5
Why ORT? similar phenomenology with cort variations in wake cort variations with REM/NREM cort variations with early/late sleep cort modulates critical neurotransmitters 1 and 3 have different ORT receptor profiles MR (type I) and GR (type II) Why GR-MR istribution Matters istribution of GR in the mouse () and rat (). subiculum Patel & ulloch, 2003 only when MR and GR are co-localized within a hippocampal subregion - as in 1 and dentate gyrus - are the two corticoid receptor subtypes capable of mediating modulatory effects on 5-HT responses according to a Ushaped corticosterone dosedependency. (Karten, 2000, p. 120) 200 mm 1 In REM output from H > N via 1 is reduced but there is normal activity within 3 G 3 Plihal & orn, 1999 Why ORT? Sleep ondition learning similar phenomenology with cort variations in wake cort variations with REM/NREM cort variations with early/late sleep cort modulates critical neurotransmitters Early sleep retrieval infusion of ORT 10:15pm underlying cort distributions in H match, in that 1 and 3 have different profiles ort injections overturn consolidation facilitation normally seen in NREM (Plihal & orn, 1999) 11:00pm learning 11:00pm 2:00am 2:15am 3:00am Early sleep retrieval 3 2 1 0-1 -2-3 -4-5 -6-7 % diff in recall Placebo ort infusion of Placebo 10:15pm 11:00pm 11:00pm 2:00am 2:15am 3:00am dream content was not assessed in this study ORT injection did not alter sleep stage, but did raise ORT levels and eliminated the facilitation seen in memory consolidation in the placebo group 6
Thus it might be ORT level, not sleep stage, or time of night, that is the causal agent in determining both the nature of dreams and memory consolidation effects Implications what is going on in REM/Hi ORT what is going on in NREM/Lo ORT Hippocampal-ortical Interactions: uring REM/HI ORT during REM/Hi ORT: impaired H > N feedback dreams are fragmented with little contextual content consolidation of episodic memory is diminished disrupted H > N consolidation, but hyperassociativity within cortical circuits 1 G hyperassociativity in N dream output seems to come from N alone consolidation of procedural memory is enhanced semantic priming by weak primes is enhanced (Stickgold etal. 1999) 3 implications for location of episode memories Hippocampal-ortical Interactions: uring NREM/Lo ORT if, as appears to be the case, dream output during REM comes from N, with little or no contribution from H the nature of memories reported from REM should tell us something about whether N can store episodes as a result of consolidation the virtual absence of coherent episodic content in REM dreams suggests that N by itself cannot store and retrieve meaningful episodic memories normal consolidation strengthened H <--> N bonds expanded H <--> H traces slow integration of new knowledge with old 1 G 3 7
broader implications if episodic memory is not stored in N as a consequence of consolidation, as the content of REM dreams suggests, what then is the purpose of consolidation? Perchance to Speculate What dreams can tell us about memory and its consolidation what is outcome of N <---> H interactions over time? transfer? replication? integration/transformation? possibilities: episodic memory initially represented in H; then transferred to N episodic memory initially represented in H; then replicated in N episodic memory represented in N; needs H intervention to allow full accessibility (creation/strengthening of links) until consolidated is H merely an index? while standard consolidation theory seems to accept this index idea, multiple trace theory (Nadel & Moscovitch) supposes that H represents critical contextual information, that cannot be fully replicated in N what transpires in N, by this view, is a strengthening of linkages among semantic representations, leading to the extraction from episodes of statistical regularities -- the gist of experience, and the integration of this new knowledge with existing knowledge To ream, Perchance to Fly a great deal has been learned about the nature and role of dreams in recent years, and a role for sleep, and dreams, in memory consolidation seems secure we know little about why some dreams are so bizarre, how we can, for example, fly -- we suggest that high levels of ORT, typically seen in REM, may be responsible study of how these dreams break the boundaries of normal thought could illuminate mechanisms of creativity, consciousness, and the dynamical neural processes that maintain the logic of waking life 8