Discussion 5

R

M

L a

b

a

A R R A

K M T R

C

o i n 5 w t v i W v fl n

0 d

Neuroscience and Biobehavioral Reviews 36 (2012) 1640–1645

Contents lists available at SciVerse ScienceDirect

Neuroscience and Biobehavioral Reviews

journa l h o me pa g e: www.elsev ier .com/ locate /neubiorev

eview

emory formation, consolidation and transformation

. Nadela,∗, A. Hupbachb, R. Gomeza, K. Newman-Smitha

Department of Psychology, University of Arizona, United States Department of Psychology, Lehigh University, United States

r t i c l e i n f o

rticle history: eceived 29 August 2011 eceived in revised form 20 February 2012 ccepted 2 March 2012

a b s t r a c t

Memory formation is a highly dynamic process. In this review we discuss traditional views of memory and offer some ideas about the nature of memory formation and transformation. We argue that memory traces are transformed over time in a number of ways, but that understanding these transformations requires careful analysis of the various representations and linkages that result from an experience. These

eywords: emory consolidation

ransformation econsolidation

transformations can involve: (1) the selective strengthening of only some, but not all, traces as a function of synaptic rescaling, or some other process that can result in selective survival of some traces; (2) the integration (or assimilation) of new information into existing knowledge stores; (3) the establishment of new linkages within existing knowledge stores; and (4) the up-dating of an existing episodic memory. We relate these ideas to our own work on reconsolidation to provide some grounding to our speculations that we hope will spark some new thinking in an area that is in need of transformation.

© 2012 Elsevier Ltd. All rights reserved.

ontents

1. The nature of memory formation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1640 2. Problems with the textbook view . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1641 3. What next . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1641 4. What is memory consolidation all about?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1642 5. What forms of memory transformation are possible? How are they instantiated in different brain systems? And what role does sleep play in the transformation process? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1643

Acknowledgement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1645 References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1645

These are exciting times in memory research. Just as with mem- ry itself, what seemed settled and fixed has become unsettled. The dea that memory formation is a highly dynamic process is hardly ew, but the implications of this dynamism were largely ignored for 0+ years. That has changed in the past decade, overturning some ell-established ideas, upsetting many apple-carts, and opening

he door to new ideas. In this review we first detail traditional iews of memory and then, in a frankly speculative way, offer some deas about the nature of memory formation and transformation.

e relate these ideas to our own work on reconsolidation to pro- ide some grounding to the speculation. It is our hope that this

1. The nature of memory formation

The textbook view of memory formation goes something like that shown in Fig. 1 below.

That is, we experience an event, some aspects of this event get encoded, this encoding initiates a series of processes, typically labeled “consolidation”, that take time, leading to a permanent memory trace. Memory consolidation plays a key role in this sequence, as it determines both what will be preserved after ini- tial encoding, and how long this process takes. Once consolidation is over, by definition, the permanent memory trace that results

ight of fancy will spark some new thinking in an area that is in eed of transformation.

∗ Corresponding author. E-mail address: [email protected] (L. Nadel).

149-7634/$ – see front matter © 2012 Elsevier Ltd. All rights reserved. oi:10.1016/j.neubiorev.2012.03.001

cannot be disrupted in the normal course of events. It is crucial to the very notion of consolidation that this end result be a fixed memory trace. Absent this, it would be impossible to state when consolidation has finished, since the operational definition of that end-point is precisely when some agreed-upon manipulation no longer disrupts memory.

L. Nadel et al. / Neuroscience and Biobehavioral Reviews 36 (2012) 1640–1645 1641

EVENT ENCODE CONS OLIDATE FIXED MEMORY

oding

2

a m i L d t c ( ( t r t m s t w o i r

F o

Fig. 1. Memory enc

. Problems with the textbook view

The problem with this view of memory formation is that it ppears to be wrong. As Fig. 2 shows, when an apparently fixed emory is reactivated, some process is initiated that renders what

nitially seemed fixed once again labile (Nader et al., 2000; see also ewis et al., 1968 and Misanin et al., 1968). A reactivated “consoli- ated” memory, that has become labile, is open to change in ways hat a non-reactivated “consolidated” memory is not. This change an take any of several forms, including weakening or even erasure Nader et al., 2000; Walker et al., 2003), strengthening, or alteration Hupbach et al., 2007). Perhaps the most typical outcome is that he reactivated and thus temporarily fragile memory will become estabilized, and that will be that. However, under other conditions hat perturb the system, a different result can be seen. In experi-

ental work with animals, for example, the injection of protein ynthesis inhibitors at this stage can disrupt the memory – an effect hat is not seen in animals that received the same injections but

ithout the prior memory reactivation (Nader et al., 2000). Another

utcome, that we have explored (see below), involves the updat- ng of the memory as a function of some novel experience. Thus, eactivation of an already consolidated memory is more than just

EVENT ENCODE CON

RE

PROTEIN SYNTHESIS INHIBI TORS

DISRUPTED MEMORY

RESTABILIZED

LABILE M

ig. 2. Possible effects of reactivation induced memory trace lability: disruption (in this c f the trace, possibly in strengthened form; updating of the trace resulting from incorpor

– traditional view.

a “read-out” of the memory trace. Instead, reactivation initiates a process, labeled “reconsolidation”, that under some circumstances can lead to a changed (Hupbach et al., 2007) or even obliterated memory (Nader et al., 2000).

This is not a small problem that can be handled by tweaking one or another parameter of the textbook story. This problem goes to the very heart of the textbook story, which helps explain why the re-emergence of the idea of memory “reconsolidation” was met with such fierce resistance.

3. What next

Our view is that the facts alluded to above represent not just a challenge to the classic story concerning memory formation. They also create an opportunity to re-evaluate some basic assumptions about memory. The remainder of this paper is devoted to doing exactly that, in a spirit of what we hope is productive speculation.

We consider the following questions:

• If memory consolidation is not about permanently fixing a mem- ory trace, what is it about?

SOLIDATE FIXED MEMORY

ACTI VATION

MEMORY

EMORY

UPDATE D MEMORY

NOVEL EXPERIENCE

ase caused by inhibition of protein synthesis; no change, other than restabilization ation of novel information.

1 behav

•

•

t T c u s t e s

1

2

3

4

5

6

e l a t t

642 L. Nadel et al. / Neuroscience and Bio

What forms of memory transformation are possible, and how do these different forms of memory transformation map onto brain systems? What role does sleep play in the normal progression of memory formation?

Before addressing these questions we must spend some time alking about memory itself: what shape does it take in the brain? here is no single answer to this question, and indeed drawing areful distinctions between several forms of memory is crucial to nderstanding what consolidation is all about. A systematic analy- is of memory is beyond the scope of this paper; readers are pointed o a recent review (Nadel and Hardt, 2011) that goes into consid- rably more detail on the following major points we will simply ketch out here:

. While psychologists have made some progress sharply distin- guishing processes such as perception, short-term memory and long-term memory, considerable neuroscientific evidence now suggests that such separations may make little sense in the brain (e.g., Barense et al., 2011).

. Instead, the brain seems to be organized into representational systems distinguished by the qualitative nature of the infor- mation they represent (Nadel and Hardt, 2011). The idea that brain circuits are organized in this qualitative way is not new – similar ideas have been expressed for nearly 200 years: the doc- trine of “specific nerve energies” first promulgated by Johannes Muller in the early 19th century, “labeled-line” theory, and con- tent addressability are three manifestations of this idea.

. Within this view, any given brain system represents a particu- lar type of information, and is accessed whenever that kind of knowledge is deployed for any psychological process, such as attention, perception, short-term memory, or long-term mem- ory.

. We focus in this article on two kinds of memory and the repre- sentations that seemingly underlie them: episodic memory and semantic memory (e.g., Tulving, 1983). We experience the world “episodically” – acquiring knowledge from specific experiences in particular contexts (spatial and temporal). These experiences yield representations that at least at the outset allow us to recall both the context-bound episodic experience and the context- free knowledge that can be extracted from experience.

. Current evidence suggests that information is processed simul- taneously in multiple representational systems, such that representations supporting several kinds of “memory” are cre- ated in parallel, including episodic and semantic memories (see below).

. It is misleading to talk about memories, either episodic or seman- tic, as being “retrieved”. This word implies a fixed entity that is accessed in its entirety. A better word is “remembered” or “rec- ollected” – where the literal sense of those words is implied. That is, an episodic memory is stored in the brain in “dismembered” form, as bits and pieces represented in distinct brain systems. As Bartlett (1932) and many others have argued, recalling an episodic memory involves “reconstruction” – putting together these bits and pieces (accurately or not).

Most discussions of memory consolidation are concerned with pisodic and semantic memory and the representations that under-

ie them, and how these forms of memory interact over time after n experience. As noted above, what makes things complicated is hat the retrieval of either form of memory (episodic or semantic) ypically engages brain systems important to both.

ioral Reviews 36 (2012) 1640–1645

With these assumptions about memory in place, we can now turn to our three questions, starting with the very purpose of mem- ory consolidation itself.

4. What is memory consolidation all about?

In contrast to the standard textbook story, we assume that con- solidation is not only, or even primarily, about permanently fixing an isolated memory trace. Taking into account the points raised above about the nature of memory, we see memory consolidation as accomplishing a number of different things, depending on what aspect of memory one is looking at.

Consider what happens when an organism experiences a novel episode, which we can view as consisting of a set of entities (indi- viduals, objects) interacting with each other in space and time. Let us leave aside, for present purposes, the various processes involved in “parsing” this episode into its constituent parts, although these are themselves fascinating and important. There is a consensus view of what happens in the nervous system during encoding: (1) representations of the various entities involved in the episode are activated, largely in cortical systems; (2) the conjunction of these entities activates representations within higher-order contextual systems, including the hippocampal formation. Representations in the hippocampal system encode spatial and temporal contextual attributes, while at the same time providing an “index” that can allow the system to access, and hence retrieve, the entities that played a role in a given episode. This much appears to be non- controversial, at least at the outset. What happens next during the phase of memory consolidation is not so clear, partly because there are two levels of memory representation, and the linkages within and between them that need to be considered, as seen in Fig. 3.

First, there are the various dispersed representations in cortex that capture the entities involved in the episode. These representa- tions appear to involve distributed overlapping coding mechanisms that promote generalization. To put this in concrete terms, within this system kitchens can be distinguished from living rooms, but all kitchens are lumped together. Cortical systems appear to be spe- cialized to extract the semantic features of entities, their statistical regularities, and to form categories and concepts.

Second, there is the representation formed within the hip- pocampal system that captures contextual attributes of the episode, most particularly the spatial context. These representations in the hippocampus appear to use sparse, distributed coding mechanisms that allow the system to separate closely related representations. This allows for sharp boundaries between quite similar contexts, and there is considerable evidence that hippocampal dysfunction manifests as, amongst other things, a loss of the ability to encode contexts and to discriminate amongst them (Nadel, 2008). This sys- tem can distinguish between my kitchen and that of my neighbor.

Then, there are the linkages among the various parts of the hippocampal contextual representations, the various parts of the cortical representations, and, finally, the linkages between the hippocampal-contextual representation and the distributed- cortical representations of the entities involved.

Memory consolidation could in principle affect each of these representations and linkages. As a result of consolidation pro- cesses, contextual representations in the hippocampal system, and entity representations in cortical systems, could be strengthened through enhancement of the changes in synaptic connectivity that is assumed to underlie memory at the cellular level. Although con- solidation involves such strengthening in both cases, the outcome

in the two systems is likely to be different: given the sparse coding in hippocampus, consolidation should have the effect of enhancing the unique, context-specific, nature of its representations. Given the overlapping coding that exists within at least some cortical

L. Nadel et al. / Neuroscience and Biobehavioral Reviews 36 (2012) 1640–1645 1643

nd cor

s m o m

o e c a h m w S m v t t M i r t s c q

t r

objects. Again, two days later, they return and are asked to recall (or

Fig. 3. Hippocampal a

ystems, consolidation should have the effect of enhancing com- on elements at the expense of outliers. The result is the extraction

f prototypes, or, to put it another way, the creation of semantic emory. At the level of linkages other things are happening: creation

f new linkages, or variations in the strength of linkages, within ither the hippocampal or cortical systems result in memories that an be remembered or not. Such variations in the strength of link- ges between hippocampal and cortical representations are at the eart of debates in the field of memory consolidation. Textbook odels of consolidation presume that it is these linkages that fade ith time, to be replaced by links within cortical networks (e.g.,

quire and Alvarez, 1995). In this view, there is no necessary infor- ation loss associated with such a shift over time, since in this

iew the hippocampal representation is merely an “index” and cap- ures no information other than which cortical nodes are part of he memory. In contrast, multiple trace theory (MTT, Nadel and

oscovitch, 1997) assumes that these linkages remain critical if one s to retrieve a detailed episodic memory, because the hippocampal epresentation is both an “index” and the bearer of specific spa- ial contextual information. Indeed, it is by virtue of being such a patial representation that it can serve to index the entities that o-occurred in a particular spatial arrangement in the episode in uestion.

Thus, memory consolidation, and memory formation, involve he delicate interplay of all these factors, resulting in multiple rep- esentations, some of which represent the context, while others

tical representations.

represent the entities. What happens over time in these traces reflects the shifting fate of specific episodic knowledge and more generic semantic knowledge, about one’s experiences. It is hard to avoid the conclusion that we should be using the term memory transformation to describe these processes. Indeed, recent work building on multiple memory trace ideas uses this terminology to describe what happens to memories over time (e.g., Winocur et al., 2010).

5. What forms of memory transformation are possible? How are they instantiated in different brain systems? And what role does sleep play in the transformation process?

Following on from the analysis above, what can we say about the kinds of transformations that occur during memory consolidation and reconsolidation?

It would be instructive to work through the various stages of memory transformation in terms of an actual example for which some empirical data exist. Fig. 4 portrays a paradigm we have recently been using to study episodic memory updating in humans. In brief, subjects learn a list of twenty common everyday objects, then two days later come back and are either actively reminded of that learning experience or not, then learn a second list of twenty

recognize) the objects from either the first or second list. In a series of studies (Hupbach et al., 2007, 2008, 2009) we have shown that when reminded prior to learning List 2, subjects routinely include

1644 L. Nadel et al. / Neuroscience and Biobehavioral Reviews 36 (2012) 1640–1645

Hour 96 Hour 48 Hour 0

Learn Set 1 Learn Set 2 Re call Test

(Set 1 or Set 2)

Full Reminder:

-Spatial Context

-Experimenter

-Reminder Question

No Reminder

OR

Fig. 4. Paradigm for the study of episodic memory reconsolidation. In the Full Reminder condition subjects are brought back to the same spatial context with the same e o in t t

( o t i o t e e s e

m ( f 2 f u

t c r o i s r e n i ( t t r r c

t l l l

xperimenter and are asked if they remember what they were previously asked to d o learn List 2 with a different experimenter and get no reminder question.

on the final test day) a number of the List 2 objects in their recall f List 1. They do not include List 1 items on a recall of List 2, so his is not a simple source confusion effect. And, although there is nterference in recall of List 1 from having learned List 2 in terms f fewer List 1 items recalled, this interference is independent of he intrusions resulting from the reminder manipulation (Hupbach t al., unpublished data). Finally, the most effective reminder to licit this memory updating effect is the initially rather unfamiliar patial context in which the entire study is carried out (Hupbach t al., 2011).

Recently, the proponents of Temporal Context Models (TCM) of emory have simulated some of our results using a TCM-like model

Sederberg et al., 2011), but their approach cannot as yet account or the special role played by the spatial context (Hupbach et al., 011). Further studies will allow us to determine the ultimate use- ulness of this kind of approach to understanding the mechanisms nderlying our updating effect.

Here, we would like to take another approach that is less about he mechanisms and more about the underlying systems-level pro- esses we discussed above. Consider what is going on in the various epresentations and linkages when our subjects acquire a mem- ry for List 1. We imagine that a number of things are happening n parallel: (1) representations of the objects in List 1, which pre- umably already exist, are activated in neocortical structures; (2) epresentations of all the other elements in the situation – the xperimenter, the features of the room, etc. are also activated in eocortical structures; (3) linkages are formed between these var-

ous activated cortical traces and elements in the hippocampus; 4) the particular hippocampal elements activated become linked ogether to form a trace that encodes the spatial relations among he entities – this neural ensemble is the hippocampal-context rep- esentation; (5) linkages, however weak, are initiated between the epresentations of the entities in cortex (forming the distributed- ortex representation).

We assume that in the short run there are cellular changes

hat modulate the strength of these traces, a process typically abeled cellular consolidation (e.g., Mednick et al., 2011). In the ong run, more substantial changes take place at the systems evel, and as suggested above, many of these seem to take place

he original session. Subjects in the No reminder condition go to a different location

during sleep. In the course of a normal night of sleep, individuals experience periods of slow-wave sleep (SWS), followed by peri- ods of rapid eye-movement sleep (REM). There are several notions about what happens during these two stages of sleep. According to the “active system consolidation” view (Diekelmann & Born, 2010), slow wave sleep repeatedly reactivates the hippocampal and neocortical traces resulting from an episode, leading to the strengthening of neocortical representations and the weakening of hippocampal representations. Such reactivations also act to incor- porate information from the new episode into pre-existing cortical stores. Diekelmann and Born hypothesize that REM sleep may then play a role in stabilizing neocortical stores through synaptic con- solidation.

Two other positions have been expressed concerning what hap- pens to memory traces during SWS and REM. According to Tonini and Cirelli (2003, 2006), a process of synaptic rescaling renormal- izes the hippocampal system during slow wave sleep without the loss of the hippocampal representation, a position more in line with our views than with the active system consolidation view. Walker and Stickgold (2010) further propose that REM is involved in memory integration, a process that goes beyond the initial con- solidation mechanism Diekelmann & Born suppose happens during SWS.

These latter considerations suggest to us the following hypothe- sis: (1) in slow wave sleep, some selection process chooses amongst traces that have been formed during the day just those that deserve to persist; (2) following this, synaptic rescaling (cf. Tonini and Cirelli, 2003) renormalizes the hippocampal system, leaving only the “chosen” traces above threshold; (3) the stage having been cleared of extraneous actors, a second phase can now begin. In this phase the remaining above-threshold “traces” are replayed to the cortex; (4) we speculate that during this phase constraints on cor- tical plasticity are transiently lowered, facilitating the formation of new associations as well as the strengthening of those ensembles activated as a result of hippocampal replay; (5) in this way, infor-

mation gathered during the episode in question can be integrated into existing cortical circuits; and finally (6) assimilation of connec- tions among the different elements of the cortical representation occurs in REM.

behav

o s p t l fi m j L t c a s a s 1 r r L i a e c p c p s n T g d fi i t i

t s s p

t m l m s s s i o

a i a e p t t a a e T t

L. Nadel et al. / Neuroscience and Bio

Consider how this hypothesized sequence of events would play ut in our memory updating paradigm: first, we assume that repre- entations formed during the learning of List 1 survive the rescaling rocess. What is more, we assume that one result of rescaling hese memory traces is their transformation into a form that allows ater updating. We make this assumption because of some recent ndings in our lab exploring the effects of sleep deprivation on emory updating. In this study (Newman-Smith et al., 2011) sub-

ects learned List 1 and then were sleep deprived before learning ist 2. In this case, subjects who were reminded of List 1 prior o learning List 2 did not misattribute List 2 items to List 1 (in ontrast to what we find when subjects had a chance to sleep fter List 1 encoding). By the time of this retrieval subjects had lept, so fatigue could not have been the cause of this seemingly nomalous result. There are two rather different ways of under- tanding this result. One possibility follows from the fact that Lists

and 2 are learned in very different mental contexts – one well ested, the other sleep deprived. List 1 is then recalled in a well- ested state, which matches the context in which List 1, but not ist 2, was learned. It is possible that this context mismatch makes ntrusions highly unlikely. Another possibility follows from the nalysis offered above: during sleep a process unfolds that low- rs constraints on plasticity and facilitates the formation of new onnections. Until one sleeps, the traces established that day are rotected from such change, and appear subject only to decay and onsequent forgetting. Consistent with this idea is a study com- aring paired-associate recall after 48 h during which one group of ubjects slept during the first 24-h interval and a second group did ot sleep until the second night after learning (Gais et al., 2007). he group that was sleep deprived forgot more associates than the roup that slept and also showed less hippocampal brain activity uring recall as measured by fMRI. This result suggests two things: rst, sleep can both strengthen memory traces (thereby diminish-

ng forgetting), and, perhaps as a function of hippocampal activity, ransform memories into a format that permits subsequent updat- ng.

In the account above we have attributed this effect to some- hing that occurs during either slow wave or REM sleep. In our tudy subjects were deprived of both slow wave and REM sleep, o we cannot pin down which sleep stages are responsible at resent.

To summarize, we believe memory traces are transformed over ime in a number of ways, but that understanding these transfor-

ations requires careful analysis of the various representations and inkages that result from an experience. Many of these transfor-

ations occur in the normal course of events: (1) the selective trengthening of only some, but not all, traces as a function of ynaptic rescaling, or some other process that can result in selective urvival of some traces; (2) the integration (or assimilation) of new nformation into existing knowledge stores; (3) the establishment f new linkages within existing knowledge stores.

One class of transformations, however, only follows upon re- ctivation of previously established memory traces. These all nvolve the up-dating of an existing episodic trace that has been ctivated (cf., Lewis, 1979). It remains unclear what aspects of the pisodic trace are actually changed as a function of updating. In rinciple there seem to be three possibilities: the hippocampal race could be expanded to incorporate links to new (contex- ual) elements; cortical traces could be changed to incorporate ssociations between old and new elements; and, finally, the link-

ges between hippocampal and cortical trace systems could be xpanded to include links to the newly incorporated elements. hese three are by no means mutually exclusive; there are reasons o imagine that all are taking place.

ioral Reviews 36 (2012) 1640–1645 1645

As we indicated at the outset, this speculative analysis of the various processes involved in memory formation and reformation is offered with the view that we must, as a field, move away from current static understandings of memory to embrace the dynamics we now know to characterize this psychological capacity. We cer- tainly do not expect our analysis to be right in all particulars, but rather hope that it will prove sufficiently interesting to propel new research and theories in the future.

Acknowledgement

We gratefully acknowledge support from the National Science Foundation (BCS-0743988) for the research described herein.

References

Barense, M.D., Ngo, J.K.W., Hung, L.H.T., Peterson, M.A., 2011. Interactions of memory and perception in amnesia: The Figure-Ground Perspective. Cerebral Cortex, doi:10.1093/cercor/bhr347.

Bartlett, F., 1932. Remembering: A Study in Experimental and Social Psychology. Cambridge University Press, Cambridge, UK.

Diekelmann, S., Born, J., 2010. The memory function of sleep. Nature Reviews. Neu- roscience 11, 114–126.

Gais, S., Albouy, G., Boly, M., Thien Thanh, D.-V., Darsaud, A., et al., 2007. Sleep trans- forms the cerebral trace of declarative memories. Proceedings of the National Academy of Sciences of the United States of America 104, 18778–18783.

Hupbach, A., Gomez, R.G., Nadel, L., 2011. Episodic memory reconsolidation: the role of context familiarity. Psychological Bulletin and Review 18, 787–797.

Hupbach, A., Gómez, R., Hardt, O., Nadel, U.L., 2007. Reconsolidation of episodic memories: a subtle reminder triggers integration of new information. Learning & Memory 14, 47–53.

Hupbach, A., Hardt, O., Gomez, R., Nadel, L., 2008. The Dynamics of Memory: Context- Dependent Updating. Learning and Memory 15, 574–579.

Hupbach, A., Gomez, R., Nadel, L., 2009. Episodic memory reconsolidation: Updating or source confusion? Memory 17, 502–510.

Hupbach, A., Weiss, M., Gomez, R., & Nadel, L. How does memory strength affect memory updating?, unpublished data.

Lewis, D.J., 1979. Psychobiology of active and inactive memory. Psychological Bul- letin 86, 1054–1083.

Lewis, Misanin, D.J., Miller, J.R., 1968. Recovery of memory following amnesia. Nature 220, 704–705.

Mednick, S.C., Cai, D.J., Shuman, T., Anagnostaras, S., Wixted, J.T., 2011. An oppor- tunistic theory of cellular and systems consolidation. Trends in Neurosciences 34, 504–514.

Misanin, J.R., Miller, R.R., Lewis, D.J., 1968. Retrograde amnesia produced by elec- troconvulsive shock after reactivation of a consolidated memory trace. Science 160, 554–555.

Nadel, L., 2008. The hippocampus and context revisited. In: Mizumori, S.J.Y. (Ed.), Hippocampal Place Fields: Relevance to Learning and Memory. Oxford Univer- sity Press, New York, pp. 3–15.

Nadel, L., Hardt, O., 2011. An update on memory systems. Neuropsychopharmacol- ogy Reviews 36, 251–273.

Nadel, L., Moscovitch, M., 1997. Consolidation, retrograde amnesia and the hip- pocampal formation. Current Opinion in Neurobiology 7, 217–227.

Nader, K., Schafe, G.E., Le Doux, J.E., 2000. Fear memories require protein synthesis in the amygdala for reconsolidation after retrieval. Nature 406, 722–726.

Newman-Smith, K., Gómez, R.L., Bootzin, R.R., Nadel, L., 2011. Presentation given at the Associated Profession Sleep Societies Meeting, Minneapolis, June. Sleep deprivation modifies the susceptibility of memories.

Sederberg, P.B., Gershman, S.J., Polyn, S.M., Norman, K.A., 2011. Human memory reconsolidation can be explained using the temporal context model. Psycho- nomic Bulletion & Review 18, 455–468.

Squire, L.R., Alvarez, P., 1995. Retrograde amnesia and memory consolidation: a neurobiological perspective. Current Biology 5, 169–177.

Tonini, G., Cirelli, C., 2003. Sleep and synaptic homeostasis: a hypothesis. Brain Research Bulletin 62, 143–150.

Tonini, G., Cirelli, C., 2006. Sleep function and synaptic homeostasis. Sleep Medicine Reviews 10, 49–62.

Tulving, E., 1983. Elements of Episodic Memory. Clarendon Press, Oxford, UK. Walker, M., & Stickgold, R., 2010. Overnight alchemy: sleep-dependent mem-

ory evolution. Nature Reviews. Neuroscience 2010;11(March (3)):218, doi:10.1038/nrn2762-c1.

Walker, M.P., Brakefield, T., Hobson, J.A., Stickgold, R., 2003. Dissociable stages of

human memory consolidation and reconsolidation. Nature 425, 616–620.

Winocur, G., Moscovitch, M., Bontempi, B., 2010. Memory formation and long-term retention in humans and animals: convergence towards a transforma- tion account of hippocampal-neocortical interactions. Neuropsychologia 48, 2339–2356.