Handbook of Episodic Memory

By Elsevier Science

Episodic memory is the name of the kind of memory that records personal experiences instead of the mere remembering of impersonal facts and rules. This type of memory is extremely sensitive to ageing and disease so an understanding of the mechanisms of episodic memory might lead to the development of therapies suited to improve memory in some patient populations. Episodic memory is unique in that it includes an aspect of self-awareness and helps us to remember who we are in terms of what we did and what we have been passed through and what we should do in the future.

This book brings together a renowned team of contributors from the fields of cognitive psychology, neuropsychology and behavioural and molecular neuroscience. It provides a detailed and comprehensive overview of recent developments in understanding human episodic memory and animal episodic-like memory in terms of concepts, methods, mechanisms, neurobiology and pathology. The work presented within this book will have a profound effect on the direction that future research in this topic will take.

- The first and most current comprehensive handbook on what we know about episodic memory, the memory of events, time, place, and emotion, and a key feature of awareness and consciousness
- Articles summarize our understanding of the mechanisms of episodic memory as well as surveying the neurobiology of epsidodic memory in patients, animal studies and functional imaging work
- Includes 34 heavily illustrated chapters in two sections by the leading scientists in the field

• Language: English
• Publisher: Elsevier Science
• Release date: Sep 4, 2008
• ISBN: 9780080932361

Book preview

Elsevier.

Chapter 1.1 Perspectives on episodic and semantic memory retrieval

Lee Ryan, Siobhan Hoscheidt, Lynn Nadel*

---

Department of Psychology and McKnight Brain Institute, University of Arizona, Tucson, AZ, USA

* Corresponding author.

E-mail address: [e-mail]

Abstract

Episodic or autobiographical recollection involves re-experiencing a past event that is specific in time and place, while semantic recollection is concerned with facts and general knowledge about the world. Several prominent memory theories posit that the hippocampus differentiates between these two types of memories, mediating episodic, but not semantic, retrieval. In this chapter we explore a different view of hippocampus, one that emphasizes a singular response of the hippocampus during memory encoding and retrieval of both episodic and semantic memories, based on an amalgam of two existing theories of hippocampal function, multiple trace theory (MTT; Nadel & Moscovitch, 1997), and cognitive map theory (O'Keefe & Nadel, 1978). We review neuropsychological and neuroimaging literature suggesting that both semantic and episodic memory retrieval engages the hippocampus, at least within the normally functioning brain. We then describe an updated version of MTT that incorporates these new findings. Finally, we explore the notion that differences in the role the hippocampus plays in these forms of memory reflect two critical factors - the nature of the information being retrieved, and the requirements of the retrieval task.

Episodic and semantic memory seem, at least phenomenologically, quite different from one another. Episodic or autobiographical recollection involves thinking about a past event – it is personal, emotional, populated with players and specific places, imbued with detail, and it often has relevance to our sense of self and the meaning of our lives. Semantic recollection, on the other hand, has to do with knowledge – it is factual, and typically devoid of emotion or reference to time, place, and self. While semantic knowledge conveys meanings, it is rarely the kind of personal meaning embodied in autobiographical and episodic memories. This distinction, as outlined by Tulving (1983), focused originally on the different types of information processed by the two systems, unique spatial-temporal contexts for episodic memory, and facts and concepts for semantic memory. More recently, Tulving (2002, 2005) has emphasized that the critical distinction is not so much the type of information being processed, but instead that episodic memory allows the rememberer to have the conscious experience, or autonoesis, of being mentally present once again within the spatial-temporal context of the original experience – the phenomenal experience of remembering. Of course, this ability presupposes that the individual can retrieve the spatial-temporal context in which the to-be-remembered event occurred. Thus, spatial-temporal context remains a critical component of episodic memory.

Given the significant differences across these two memory types, it is not unreasonable to assume that they are mediated by separate and relatively independent systems, not only at the level of cognition, but also the brain. This idea was bolstered early on by the finding that amnesia resulting from damage to the medial temporal lobe (MTL) combined severe deficits in episodic memory retrieval with an apparent ability to access previously acquired world knowledge, facts, and skills. Given the brain damage observed in H.M. and other MTL amnesics, it seemed plausible that the hippocampus differentiated between these two types of memories, mediating episodic, but not semantic, retrieval.

The hippocampal role in retrieval has traditionally been viewed as temporary, lasting only until a process of memory consolidation (cf., McGaugh, 2002) transferred responsibility for retrieval to extrahippocampal (presumably neocortical) circuits. This has become known as the standard theory of consolidation (Squire, 1992). A more recent alternative view, multiple trace theory (MTT; Nadel and Moscovitch, 1997, 1998) addressed the question of what happens to episodic memories over time and, in contrast to standard consolidation theory, postulated an ongoing role of hippocampus in autobiographical memory retrieval. Based on an extensive review of the amnesia literature, MTT proposed that episodic memory retrieval would elicit a new encoding, leading to an expanded representation of that memory within the hippocampus itself. Early computational models suggested that such an assumption could plausibly account for the facts of retrograde amnesia (e.g., Nadel et al., 2000).

The debate over whether the hippocampus is utilized in retrieval of well-consolidated episodic memories has been largely resolved. fMRI studies have shown repeatedly, and in various ways, that even very remote event memories activate the hippocampus. Neuroimaging studies have shown consistent activation in MTL structures during retrieval of recent and very remote autobiographical memories (Ryan et al., 2001; Maguire et al., 2001a; Gilboa et al., 2004; Rekkas and Constable, 2005). Even those studies reporting a time-dependent gradient have shown that the activity appears to be related to aspects of the memories such as vividness or the amount of detail retrieved, rather than how recently the event was experienced (Addis et al., 2004) Additionally, it appears that amnesics are not normal in autobiographical memory retrieval for very old events as was once assumed (Cipolotti et al., 2001; Steinvorth et al., 2005; but see Kirwan et al., 2008). Although they may be able to access the general facts of a particular past event (e.g., that their wedding happened in 1972 in Toronto), their recollections are generally sparse and lacking in detail; they are unable to reconstruct a cohesive description of these types of events that normal individuals would produce naturally (Moscovitch et al., 2006). And, consistent with Tulving's (1983, 2005) predictions, they appear equally unable to imagine themselves within the context of a future event (Klein et al., 2002; Hassabis et al., 2007). The preponderance of evidence supports the view that the hippocampus plays a lasting role in the retrieval of episodic memories (see the chapter by Moscovitch et al., this volume, for further discussion of issues surrounding episodic memory consolidation).

In contrast, debate continues regarding whether the hippocampus is critical for the retrieval of semantic memories, including personal semantics and world knowledge. Much of the evidence on both sides of this debate comes from patients with MTL damage. Squire and others (Squire and Zola, 1998; Lou and Niki, 2002; Manns et al., 2003; Squire et al., 2004) emphasize that at least some amnesics appear to have significant deficits in semantic memory retrieval, even for well-established world knowledge. However, semantic memory impairment tends to be extensive only when the damage extends beyond the hippocampus to other MTL and neocortical structures (Schmolck et al., 2002) and can reach the same level of deficit as autobiographical memory loss, or even exceed it, in some patients (Bayley et al., 2003; Bayley and Squire, 2005). Alternatively, a recent review of the patient literature (Moscovitch et al., 2006) concluded that retrograde amnesia for semantic memory is either spared completely or confined to a period of about 10 years prior to the head injury, providing that the damage is limited primarily to the hippocampal formation.

While the extent of hippocampal involvement in retrieval remains controversial, there is little doubt that acquisition of new episodic and semantic memories is impaired by damage to the hippocampus, at least when the injury is acquired in adulthood. Anterograde amnesia remains a defining feature of the amnesic disorder (Milner, 2005; Keane and Verfaellie, 2006). Cases of developmental amnesia caused by hippocampal damage early in life are interesting because of the remarkable amount of semantic information that these individuals acquire despite profound deficits in episodic memory (de Haan et al., 2006), although their acquisition of new knowledge is not completely normal (Vicari et al., 2007). Adult-onset amnesics too can learn new semantic information, but it is clearly a very inefficient learning process and the resulting knowledge is inflexible and does not easily generalize to other contexts (Baddeley and Wilson, 1986; Glisky et al., 1986; Wilson and Baddeley, 1988). Thus, the hippocampus appears to play an important role in the acquisition, but not retrieval, of semantic memories, while participating in both the acquisition of new episodes and their subsequent retrieval throughout the lifetime of the rememberer.

In this chapter we explore a different view of hippocampus, one that emphasizes a singular response of the hippocampus during memory encoding and retrieval of both episodic and semantic memories. This view is an amalgam of fundamental assumptions drawn from cognitive map theory (CMT; O’Keefe and Nadel, 1978), and MTT (Nadel and Moscovitch, 1997). CMT assumes that the hippocampus is preferentially involved in the processing of spatial contexts and spatial relations. MTT assumes that inputs to the MTL automatically engage hippocampal networks, whether the information involved is semantic or episodic, resulting in activation in cortical networks related to the input. While CMT placed special emphasis on the role of the hippocampus in episodic memory, here we will assume a preferential role of the hippocampus in the processing of all spatial content, whether episodic or semantic.

In what follows we review some of the literature suggesting that semantic and episodic memory are interactive and that retrieval of either engages the hippocampus, at least within the intact brain. We describe an updated version of MTT that incorporates these new findings. Finally, we explore the notion that differences in the role the hippocampus plays in these forms of memory reflect two critical factors – the nature of the information being retrieved, and the requirements of the retrieval task.

I MTT: Episodic retrieval and beyond

MTT posits that information representing the spatial context of an event is encoded in an ensemble of neurons within the hippocampus, and that this ensemble trace acts as a pointer to the various features of the event represented in cortical regions. Contrary to the standard view of consolidation (e.g., Squire and Alvarez, 1995), MTT assumes that the hippocampal trace remains relevant over time, as it represents a critical component of the memory for any episode – namely, the precise details of the spatial-temporal configural context within which the event transpired. Expansion of this hippocampal trace over repeated reactivations increases the likelihood that the complete episode can be successfully retrieved given only a partial cue from the original event.

Assuming that retrieval of even quite old episodic memories engages the hippocampus, it is worth pursuing the idea that the same might be true for semantic knowledge gained at various times in the past, particularly when the information refers to spatial locations or spatial relations. We, along with Moscovitch (1995), have suggested that the hippocampus automatically binds portions of an event to related information that has been previously stored. This automatic encoding at the level of the hippocampus for all incoming information that is deemed novel (Nadel et al., 2007a; see below) would account for observed deficits in learning both episodic and semantic materials in amnesics. MTT assumed, based on the amnesia literature, that the hippocampus is not involved in semantic memory retrieval. It agreed with standard theory that during memory consolidation semantic knowledge would be substantially established within extrahippocampal circuits, thereby allowing some, but not all, learned materials to be retrieved without the use of the hippocampus. However, to say that some semantic knowledge can be retrieved in the absence of the hippocampus is not to say that when the hippocampus is present it plays no role in such retrievals. Perhaps, in the intact brain, both episodic and semantic retrievals engage the hippocampus.

This conception of semantic and episodic memory as interactive in the normal brain is supported by cognitive approaches to memory as well as by the analysis of overlapping brain networks involved in the retrieval of semantic and episodic memories. Barsalou and colleagues (Barsalou, 1983, 1988; Barsalou and Sewell, 1985; Barsalou et al., 1998) have suggested that semantic memory is contextually bound to autobiographical information, such that episodic memory is frequently used to generate semantic information. For example, Vallée-Tourangeau et al. (1998) asked participants to produce category exemplars to such common categories as kitchen utensils or food items, and then asked them to describe the strategies they used to produce the category items. In over 70% of the common categories, participants reported using a strategy involving a personally familiar context, such as imagining their own kitchen, or walking through the aisles of their neighborhood grocery story (see also Williams and Hollan, 1981; Walker and Kintsch, 1985); we discuss these findings and their implications in greater detail below. Category production tasks are used widely in neuropsychological evaluations to assess the integrity of semantic memory (Lezak et al., 2004). If episodic memory is preferentially used by neurologically normal individuals to generate semantic information in such classic semantic tasks as category production, then patients with hippocampal damage might show impairment on this task. Patients with MTL amnesia can perform normally on category production tasks (Schmolck et al., 2002), but at least one study (Gleissner and Elger, 2001) has reported that patients with lesions restricted to the hippocampal complex produce fewer exemplars during category production than normal individuals or patients with nonhippocampal temporal lobe lesions.

A study by Westmacott and Moscovitch (2003) demonstrated that episodic memory can facilitate access to semantic knowledge. Participants were asked to name the professions of famous people. Reading speeds and categorization by profession were faster and more accurate when the name of the famous person was associated with a personally significant recollection. For example, the name Elvis Presley might be associated with a visit to Graceland, whereas Frank Sinatra holds no such personal association. Performance favors Elvis Presley, though both people are equally famous. For another individual, the opposite may hold true, with enhanced performance for Frank but not Elvis. The results suggest that episodic experiences can influence semantic judgments. Studies of semantic dementia suggest that this influence goes both ways, that is, semantic memory can equally influence episodic memory. On the Autobiographical Memory Interview (AMI; Kopelman et al., 1990), semantic dementia patients retrieve memories from recent time periods significantly better than memories from childhood and early adulthood (Snowden et al., 1996; Graham and Hodges, 1997; Hodges and Graham, 1998). One proposed explanation of this finding is that semantic knowledge is a critical component of episodic memory that provides common information about an event (Moscovitch and Nadel, 1999). Impairments in semantic memory like those seen in semantic dementia will thus impair retrieval of event memories, particularly remote memories for which some loss of spatial-temporal detail has occurred over time (Moscovitch et al., 2006).

While behavioral studies in normal individuals and patients provide evidence for an interaction between episodic memory and semantic knowledge, recent neuroimaging studies suggest that these two types of memory share common neural substrates. It is well established that MTL structures are activated during episodic memory retrieval. A growing number of studies have also reported MTL activity during tasks that require access to semantic knowledge, although the activation is sometimes in hippocampus, sometimes in parahippocampal gyrus, and sometimes in both. For example, hippocampal activation has been observed during retrieval of public events (Maguire and Mummery, 1999) and famous faces (Kapur et al., 1995; Leveroni et al., 2000; Bernard et al., 2004), while parahippocampal gyrus activation occurs during recognition for famous faces (Haist et al., 2001) and famous names (Douville et al., 2005). Whatmough and Chertkow (2007) reported a PET study showing that cerebral blood flow (CBF) in the hippocampus covaried with superior performance during the retrieval of general knowledge. Increased left hippocampal CBF was associated with faster word meaning retrieval and increased right hippocampal CBF with better picture naming.

A handful of neuroimaging studies focusing on semantic spatial knowledge have also found activation in MTL structures. For example, Maguire et al. (1997) reported activation in parahippocampal gyrus when experienced London taxi drivers were required to find novel routes from one location to another after familiar routes were blocked. While it might appear that parahippocampal cortex, but not hippocampus, is critical for spatial memory, Maguire et al. (2006a) reported on the patient T.T. with bilateral hippocampal damage who was a London taxi driver before his head injury. T.T. had preserved knowledge of the major routes around London, but he was significantly impaired on tests requiring navigation through nonmajor routes, known as B-routes, which require more detailed representation of London roads. This finding suggests that schematic or sparse representations of space may be mediated by extrahippocampal structures, while the hippocampus itself provides detailed knowledge that can be used in a flexible and goal-directed manner (see also Rosenbaum et al., 2007).

Few neuroimaging studies, however, have made a direct comparison between episodic and semantic retrieval tasks that are well matched in other respects, including the type of stimuli presented, the familiarity of the stimuli, and the responses made by the participant, while varying only the source of the retrieved information. In one such study, Maguire and Mummery (1999) compared yes/no recognition for autobiographical events and public events and found hippocampal activation during both semantic and episodic retrieval, although the level of activation was greater for episodic events. Duzel et al. (1999) also matched conditions carefully in a PET study comparing semantic living/nonliving judgments for novel words with old/new recognition judgments for previously studied words. They found significantly greater activation in MTL when comparing recognition to semantic judgments. However, semantic judgments were not compared to a baseline condition, so it could have been the case that both conditions elicited hippocampal activation but to varying degrees. Interestingly, applying structural equation modeling to the same dataset as Duzel et al. (1999), Rajah and McIntosh (2005) compared the networks mediating the two tasks. They found that separate episodic and semantic models failed to differentiate one task from the other, with similar patterns of path coefficients emerging for the two retrieval models across tasks. The authors concluded that the same memory network was engaged across tasks, and that differences between episodic and semantic retrieval likely reflected variation along a continuum of processing during task performance, rather than the output of two completely independent memory systems. A similar outcome was recently reported by Burianova and Grady (2007), showing overlapping networks of activation during semantic and episodic retrieval, which included regions of left MTL.

In our laboratory, using several matched semantic and episodic tasks, we have demonstrated that the hippocampus may indeed be involved in semantic retrieval, to varying degrees depending upon the specifics of the task. For example, we recently compared a classic semantic retrieval task, category production, to an episodic version of the same task, category cued recall (Ryan et al., in press). On the day prior to fMRI scanning, participants learned a set of seven relatively unusual exemplars for each of 15 categories. The following day while in the scanner, participants recalled the lists of items from the same category cues, and also produced exemplars belonging to categories that were not studied previously. Thus, the two tasks were well matched in terms of the cues presented and the number and quality of the responses generated by the subject, but differed in the source of the information to be retrieved. In both the episodic and semantic tasks, compared to a control condition, we observed bilateral hippocampal activation, although other regions of the brain clearly distinguished the two tasks. Interestingly, participants reported using retrieval strategies during the semantic retrieval task that relied on autobiographical and personally relevant spatial information in order to generate category exemplars; for example, visualizing themselves standing in their kitchen looking through the cupboards and drawers while producing items for the category kitchen utensils. Others reported thinking about the various teams they played on during high school for the category sports events. In a follow-up experiment, we considered whether the use of these spatial and autobiographical retrieval strategies could have accounted for the hippocampal activation observed earlier. Categories were presented that were carefully normed to elicit one of three retrieval strategy types, autobiographical and spatial, autobiographical and nonspatial, and neither autobiographical nor spatial. Contrary to our expectations, similar hippocampal activation was observed bilaterally for all three category types, regardless of the inclusion of spatial or autobiographical content. Interestingly, the categories that elicited personally relevant spatial contexts showed significantly greater activation in bilateral posterior parahippocampal cortex extending into the fusiform gyrus, along with greater activation in lateral and superior parietal cortical regions, all regions that have been associated with processing complex scenes and spatial relations.

In another study, we investigated the hippocampal response to spatial relational judgments using closely matched episodic and semantic tasks. Based on a paradigm devised by Maguire and Mummery (1999), we interviewed subjects one week prior to their scanning session in an open-ended fashion about significant events in their past. True or false sentences regarding these events were created where the last word in the sentence contained the critical information required for the participant to determine the validity of the sentence. Half of the sentences cued subjects to retrieve specific spatial information, while half of them cued nonspatial details of the reported event memories. We compared this task to semantic knowledge sentences which required the retrieval of spatial and nonspatial facts. Examples of the four types of sentences are listed below. Subjects read the sentences as they were presented in the scanner and pressed a button indicating yes or no to each of the sentences. Control items were ungrammatical sentences composed of prepositions and conjunctions (Maguire and Mummery, 1999) that participants read and then pressed a mouse button when they were done.

• Episodic spatial: At your wedding, your father sat on your left.

• Episodic nonspatial: At your wedding, your bridesmaid's dresses were pink.

• Semantic spatial: The city of New York is located north of New Orleans.

• Semantic nonspatial: Komodo dragons are a large species of lizard.

Consistent with Maguire and Mummery (1999), retrieval in all four conditions elicited hippocampal activation compared to the control task. Of critical importance here is the comparison of the episodic and semantic conditions when participants were focusing on spatial information. As depicted in Fig. 1 below, significantly greater activation was observed in the hippocampus and adjacent parahippocampal cortex for spatial sentences compared to nonspatial sentences in both episodic and semantic conditions.

Fig. 1 Episodic and semantic memory judgments, comparing sentences with and without spatial content. Data were analyzed in SPM99 within two regions of interest, hippocampus and parahippocampal gyrus, using regional masks obtained from MarsBar (Brett et al., 2002), using a cluster threshold of p<.05. (See Color Plate 1.1.1 in Color Plate Section.)

Plate 1.1.1 Episodic and semantic memory judgments, comparing sentences with and without spatial content. Data were analyzed in SPM99 within two regions of interest, hippocampus and parahippocampal gyrus, using regional masks obtained from MarsBar (Brett et al., 2002), using a cluster threshold of p<.05.

These results are consistent with the notion that, at least in the intact brain, the hippocampus is engaged by tasks that emphasize spatial relational information, whether classified as episodic or semantic. The findings support one of the central tenets of both CMT and MTT that the hippocampus plays a critical role in representing spatial relational information. However, it is important to note that even when nonspatial information was retrieved, the hippocampus still showed significant activation during both episodic and semantic retrieval tasks. These effects cannot be explained fully by reference only to spatial contextual information, suggesting that the hippocampus plays an important general role in both episodic and semantic retrieval (Eichenbaum, 2004).

How do we account for the finding that hippocampus is involved in retrieval of semantic information, particularly when that information includes spatial content?Barsalou (1988) has long championed the idea that semantic knowledge of the world is embedded in a network of episodes. By this view, episodes are represented as single events and are connected, in the memory system, to other related episodes. Semantic memory is essentially derived from these event memories, which can be convolved to emphasize common information that is experienced across contexts. This information can then be retrieved separately from a specific context if necessary. However, more often, the act of retrieving one component of these convolved memories results in the retrieval of autobiographical information as well. As noted earlier, when producing exemplars belonging to a category such as kitchen utensils, virtually every person reports that they visualize themselves standing in a kitchen, opening drawers, looking systematically around the room and naming the objects they see. Importantly, it is not just any kitchen they report visualizing, but their own kitchen, the one in which they would most likely find themselves. It could be that the cue kitchen is simply a salient cue for personal experience and thus one's own kitchen comes to mind. It could be, however, that semantic knowledge, as Barsalou has suggested, is actually embedded or integrated into the structure of episodic memory.

The category production study by Ryan et al. (in press) highlights one of the complications in this domain of research; that is the tendency to label a task simply as episodic or semantic, depending upon whether or not it requires the retrieval of a specific spatial-temporal context. While it is true that most semantic tasks, such as generating a verb that relates to a noun (bat –swing), do not require the retrieval of this kind of contextual information, it does not follow that such tasks would fail to benefit from accessing such information as a route to retrieving the desired semantic knowledge. In this example, recalling an episode in which one held a bat in one's hand, or observed someone else holding a bat, might facilitate retrieval of the relevant action verb, consistent with Westmacott and Moscovitch (2003). Consider, also, a more complex example, such as being asked to recall information about the death of President Lincoln. Clearly this is not itself an episodic memory, because the rememberer was not personally present. However, in order to generate the fact that he was shot, the name of his assailant, the approximate date, and so on, we might visualize in our mind's eye a theatre, a play, a gunman, and perhaps imagine too the aftermath as Lincoln collapsed in the theatre. Though not an episode memory, by definition, this retrieval includes many of the components necessary for an episode memory, such as a specific and unique context, a time, actors, and a series of events that are connected linearly in time. The line between episodic memory and semantic knowledge is fine indeed.

According to Barsalou, "there are no invariant knowledge structures in memory. Instead, people continually construct unique representations from loosely organized generic and episodic knowledge to meet the constraints of particular contexts" (Barsalou, 1988, p. 236). Instead of focusing on abstracted concepts, Barsalou emphasizes the critical role of personally relevant instances for generating semantic knowledge. This is an interesting idea, because it suggests that semantic memory is not simply a stable and accurate record of past learning, but something that is generative, flexible, contextually bound, and subject to revision through novel experience. By this view, semantic memory is generated anew each time it is required, in much the same way as Bartlett (1932) and others (e.g., Bergman and Roediger, 1999; Nadel et al., 2007a) have noted that episodic memories are reconstructed and revised over time and through multiple retrievals. This stands in contrast to the classic distinction between episodic and semantic memories and the assumption that semantic memory, at least, is a faithful record of prior learning.

Perhaps the role of the hippocampus in memory retrieval can best be understood by taking into consideration the goal of the retrieval task engaged in by the individual, and what information is relevant to the attainment of that goal. Depending upon what we, as rememberers, are trying to accomplish, we seek to retrieve either an entire event, or just some knowledge obtained during an event, or other relevant aspects of an event such as the sequence of behaviors we or others normally engage in during similar events. These considerations suggest that the hippocampus is activated by all kinds of inputs, and that it does not distinguish, in the first instance, between categories or qualities of information. A cue is just a cue. However, the hippocampus necessarily mediates the retrieval of episodic memories because of the critical importance of spatial and temporal context in defining singular events, and the essential role the hippocampus plays in representing this kind of contextual information. Since semantic knowledge often does not necessarily include such contextual knowledge it can be retrieved without involving hippocampal circuits. But, when the hippocampus is available, even inputs in search of semantic knowledge are likely to activate this structure, because semantic retrieval can frequently be enhanced by taking advantage of routes to information connected to specific episodes in the rememberer's past.

II Extending MTT from episodic to semantic memory

The basic tenets of MTT are that the hippocampus binds information within an event and that this binding reflects its preferential engagement by some types of information, particularly spatial and contextual detail. Each time an event is recollected, an updated trace is created that incorporates information from the old trace, but now includes elements of the new retrieval episode, resulting in traces that are both strengthened and expanded. This process is primarily initiated by active retrieval, although the off-line reactivation that occurs during sleep, and indirect reminder-induced reactivation can also trigger it (e.g., Wilson and McNaughton, 1994; Hupbach et al., 2007; Nadel et al., 2007b).

Although MTT focuses on the role of the hippocampus (and other MTL structures), memory traces involve a network of regions throughout the brain that are also strengthened and expanded with retrieval, reactivation, and re-encoding. The hippocampus is always necessary for the efficient reinstatement of this network and, as already noted, continues to provide information relevant to the recollection of spatial and contextual detail, regardless of the age of the memory.

The fundamental conclusion deriving from this analysis is that every act of encoding engages processes akin to retrieval, and every act of retrieval engages processes akin to encoding. At the level of behavior, therefore, encoding and retrieval are virtually indistinguishable from one another. However, at the neurophysiological level, it should be possible to distinguish encoding and retrieval processes. For example, there is good evidence that the hippocampus can enter into distinct states corresponding to either encoding or retrieval (e.g., Hasselmo et al., 1995). Inputs to the hippocampus can act as a cue for the retrieval of related relevant information, and in a subsequent cycle within the same circuits initiate the creation of a new or expanded trace. The contents of consciousness will always include, to varying degrees, the experience that is currently happening, recollections of prior similar or related events, and relevant semantic knowledge.

The pattern separation capacity of the hippocampus engaged by encoding guarantees that each time a trace is laid down, it also retains its separate identity, while allowing information to be extracted statistically over multiple similar experiences, giving rise to what we call semantic memory. This idea is the basis of latent semantic analysis models (e.g., Landauer and Dumais, 1997). By this view, semantic information may be indistinguishable from episodic memory at the level of brain when it is first acquired, and only later becomes differentiated as similar experiences accumulate. Importantly, cues that are sufficiently distinctive retain the ability to extract a specific trace. Within this view, one can see how memories can readily become confused, particularly with respect to the details of context etc., when they are retrieved after the passage of time. In this circumstance most memories are connected to other similar traces, and other contexts that were experienced during subsequent retrievals. We have argued elsewhere (Nadel et al., 2007a) that this may be one basis for the inclusion of erroneous information and confusions into episodic memories.

The extent to which encoding and retrieval processes are engaged will primarily be determined by two things: (a) the requirements of the task or goal, and (b) the novelty of the information that is presented.

II.A Requirements of the task

Differences observed between episodic and semantic memory tasks may be best understood in terms of the information required by any particular task, given that an individual will use whatever means are available to solve it. Episodic memory tasks, by definition, require the retrieval of contextual information about time and place. Many recognition tasks require that the individual determine not only whether an item has been experienced before (which would include all words in a list, targets, and distractors alike) but also whether the particular word occurred in a particular list. No such requirement is inherent in most semantic memory tasks, which often require a judgment about the semantic qualities of words (such as in a living/nonliving judgment task). Engagement of the hippocampus in such cases may depend on the strategy adopted by the rememberer. Some tasks clearly emphasize world knowledge (e.g., solving a mathematical problem), while others clearly require more specific event knowledge (e.g., describing what you did last Saturday). Some, however, can be solved in several ways by accessing a mix of episodic, autobiographic, and semantic knowledge. For example, if a subject is asked to retrieve items that are typically found in a restaurant, they may think about restaurants in general or a restaurant that they go to frequently, but they may also visualize the specific restaurant that they were at the previous night. These considerations suggest, as we have argued above, that the hippocampus is best viewed as a system that automatically uses inputs (in this case, a category cue) to generate pattern completions, thereby retrieving any appropriate related information. This idea connects directly, as we have seen, to proposals by Barsalou that semantic knowledge necessarily interacts with autobiographical events and experiences.

II.B Novelty

Depending on the task situation, novelty can play an important role in determining what aspects of an environment will be attended to and encoded. In some situations, for example, threatening ones, novelty might well be ignored as known features of the environment signaling safety are sought out. But in other situations, where information gathering is more important, detecting and reacting to novelty can be critical. But how is novelty assessed? One possibility, long suspected (e.g., Sokolov, 1960; Vinogradova, 1970), is that the hippocampus acts as a comparator device, whereby incoming inputs are matched against existing traces. To the extent to which a match occurs, an input is judged familiar. A failure to match indicates novelty. Recent evidence from fMRI studies support this old idea that the hippocampus generates a novelty signal when new inputs are presented (e.g., Bunzeck and Duzel, 2006). Based on this kind of activation, the hippocampus will emphasize the encoding processes that serve to bind new information to retrieved similar information and/or to the context in which it is now experienced. In the absence of a sufficiently strong novelty signal the hippocampus will emphasize retrieval processes that use pattern completion to seek out relevant information in semantic stores. As Nadel et al. (1985) suggested some time ago, the hippocampus functions to both bind and find information.

One important implication of this formulation is that it suggests that connections between hippocampal (contextual) representations and extrahippocampal stores of featural information are not lost altogether over time. Rather, at least some of these connections must remain, and can be used in the normal case to connect semantic knowledge to specific contexts. The fact that there are alternative routes for accessing semantic memory is not surprising. It remains to be seen how flexible these alternative routes are, and under what circumstances they may or may not be sufficient for a task. This formulation is consistent with recent work exploring the fate of contextual coding in animals. A number of studies, many of which used context fear conditioning, have shown that shortly after training performance (e.g., fear) is relatively restricted to the original training context (Wiltgen and Silva, 2007; Winocur et al., 2007). With the passage of time, and no further training, fear is evinced in a wider range of contexts. This loss of context-specificity has typically been viewed as involving the loss of some hippocampal trace that would resist generalizing fear across contexts, but such an interpretation is not mandated by the data. In fact, it is equally possible that a separate, semantic trace is created (outside the hippocampus) that connects fear to generic features of the original training context, such as grid floors, or a small box with four walls. This trace would support fear in multiple contexts, but its presence need not involve the loss of the more specific trace supporting fear in the original context. Much as in the human case, rats would have multiple ways to access fear memory, some via the original training episode, others via semantic knowledge about that experience.

The original formulation of MTT made the simple assumption, based on the amnesia literature, that the hippocampus need not play an important role in semantic memory retrieval. Rather, the retrieval of semantic memory was assumed to depend on cortical structures alone. The hippocampus played a role only during encoding of new semantic information, but not during retrieval. Our discussion above leads to a revised conclusion: namely, that the hippocampus can play a very important role in the retrieval of semantic information. The finding of hippocampal activation during many semantic tasks requires such a revision. But, this new view now must account for something the old view easily explained – how amnesics can readily retrieve semantic knowledge while being significantly impaired in episodic memory.

We suggest that such knowledge is, for the most part, truly independent of spatial and contextual detail, and is also of a kind that is unlikely to benefit from retrieval routes activated by episodic memories. For example, one's understanding of the meaning of abstract words, such as truth, or integrity, is unlikely to benefit from retrieving a specific episode in which such things were previously defined. It is not surprising that this kind of semantic knowledge can be readily retrieved by amnesic patients with damage to the hippocampus but sparing of other medial temporal and lateral temporal cortices.

However, semantic memory retrieval in amnesics is not always intact. There appears to be greater impairment in semantic memory when damage extends into other MTL structures and beyond that, into the lateral temporal lobe (e.g., Schmolck et al., 2002), consistent with the notion that these cortical regions are critical to the representation of certain aspects of semantic knowledge, as many studies have shown. Even when amnesic patients can retrieve semantic knowledge there may be important qualitative differences between their retrievals and those of intact individuals. By this view, we would predict that amnesics may have greater difficulty applying the semantic knowledge to novel situations, while cognitively normal individuals, presumably benefiting from connections to episodic representations, are more flexible and adaptive. Further, while amnesics can access components of episodes (personal facts), these facts apparently cannot enable them to reinstate complete and cohesive episodic memories. One indication that this might be the case comes from studies of the ability of amnesics to imagine future events. Intact individuals use their knowledge of prior events to construct possible future scenarios, but amnesic individuals are grossly impaired at this task (Hassabis and Maguire, 2007). Importantly, they can access specific features of an event, but cannot construct a spatially coherent scene or integrate the self into the scenario. Much as their failure to retrieve spatial contextual information dooms their attempts at retrieving an episodic memory, their inability to generate a spatial context dooms their attempts to use semantic knowledge as the basis for a future event. The problem is not that they fail to access semantics, but that they simply cannot make much use of this knowledge flexibly in response to the requirements of the task.

If it turns out that even the semantic memories of amnesic patients are abnormal in some way, as we now know episodic memories are, then the entire basis of systems-level consolidation must be called into question. We would no longer to be able to assert that memory representations somehow transfer from one neural system to another over time. Instead, we might assume that different kinds of knowledge are always represented in distinct systems, and that changes in retrieval dynamics over time reflect shifts in how specific tasks are approached. Retrieving old semantic knowledge, for example, could be accomplished with or without the hippocampus.

III Conclusions

The broader implication of this way of thinking concerns the notion of a memory system. Perhaps it is no longer sensible to talk about self-contained memory systems, either in the singular or plural. Rather, we should be talking about the acquisition of various kinds of knowledge (cf., Nadel, 2008), and the subsequent deployment of particular aspects of that knowledge in the service of memory tasks, or more broadly, problem solving. Within this view, all forms of knowledge would be subject to transformation (via consolidation) and updating (via reactivation and reconsolidation), and memory retrieval would involve accessing the appropriate knowledge to fit the task demands. There is neither one nor multiple memory systems, rather a variety of knowledge systems that both process and store information, to be deployed as required. The hippocampus plays a critical role in accessing and updating some, but not all, of these systems. The challenge now is to think about hippocampal function in a different way and to reevaluate previous dichotomies, such as encoding and retrieval or episodic and semantic, in flexible ways.

Acknowledgment

We gratefully acknowledge support from NINDS (to L. Ryan and L. Nadel, Grant number: RO1 NS044107); State of Arizona Alzheimer's Research Center, McDonnell-Pew Cognitive Neuroscience Program, and the Flinn Foundation Program in Cognitive Science. We thank Elizabeth Glisky for her insightful comments on an early draft.

References

D.R. Addis, M. Moscovitch, A.P. Crawley, M.P. McAndrews. Recollective qualities modulate hippocampel activation during autobiographical memory retrieval. Hippocampus. 2004;14:752–762.

A. Baddeley, B. Wilson. Amnesia, autobiographical memory, and confabulation. In: D.C. Rubin, ed. Autobiographical Memory. Cambridge, UK: Cambridge University Press; 1986:225–252.

L.W. Barsalou. Ad hoc categories. Mem. Cognit.. 1983;11:211–227.

L.W. Barsalou. The content and organization of autobiographical memories. In: U. Neisser, E. Winograd, eds. Remembering Reconsidered: Ecological and Traditional Approaches to the Study of Memory. Cambridge University Press; 1988:193–243.

L.W. Barsalou, J. Huttenlocher, K. Lamberts. Basing categorization on individuals and events. Cogn. Psychol.. 1998;36:203–272.

L.W. Barsalou, D.R. Sewell. Contrasting the representation of scripts and categories. J. Mem. Lang.. 1985;24:646–665.

F.C. Bartlett. Remembering. Cambridge: Cambridge University Press; 1932.

P.J. Bayley, R.O. Hopkins, L.R. Squire. Successful recollection of remote autobiographical memories by amnesic patients with medial temporal lobe lesions. Neuron. 2003;38:135–144.

P.J. Bayley, L.R. Squire. Failure to acquire new semantic knowledge in patients with large medial temporal lobe lesions. Hippocampus. 2005;15:273–280.

E.T. Bergman, H.L. Roediger. Can Bartlett's repeated reproduction experiments be replicated? Mem. Cogn.. 1999;27:937–947.

F.A. Bernard, E.T. Bullmore, K.S. Graham, S.A. Thompson, J.R. Hodges, P.C. Fletcher. The hippocampal region involved in successful recognition of both remote and recent famous faces. Neuroimage. 2004;22:1704–1714.

Brett, M., Anton, J.L., Valabregue, R. and Poline, J.B. (2002) Region of interest analysis using an SPM toolbox [abstract]. Presented at the Eighth International Conference on Functional Mapping of the Human Brain, June 26, 2002, Sendai, Japan. NeuroImage, 16.

N. Bunzeck, E. Duzel. Category-specific organization of prefrontal response-facilitation during priming. Neuropsychologia. 2006;44:1765–1776.

H. Burianova, C.L. Grady. Common and unique neural activations in autobiographical, episodic, and semantic retrieval. J. Cogn. Neurosci.. 2007;19:1520–1534.

L. Cipolotti, T. Shallice, D. Chan, N. Fox, R. Scahill, G. Harrison, J. Stevens, P. Rudge. Long-term retrograde amnesia: the crucial role of the hippocampus. Neuropsychologia. 2001;39:151–172.

K. Douville, J.L. Woodard, M. Seidenberg, S.K. Miller, C.L. Leveroni, K.A. Nielson, et al. Medial temporal lobe activity for recognition of recent and remote famous names: an event-related fMRI study. Neuropsychologia. 2005;43:693–703.

E. Duzel, R. Cabeza, T.W. Picton, A.P. Yonelinas, H. Scheich, H.J. Heinze, et al. Task-related and item-related brain processes of memory retrieval. Proc. Natl. Acad. Sci. U. S. A.. 1999;96:1794–1799.

H. Eichenbaum. Hippocampus: cognitive processes and neural representations that underlie declarative memory. Neuron. 2004;44:109–120.

A. Gilboa, G. Winocur, C.L. Grady, S.J. Hevenor, M. Moscovitch. Remembering our past: functional neuroanatomy of recollection of recent and very remote personal events. Cereb. Cortex. 2004;14:1214–1225.

U. Gleissner, C.E. Elger. The hippocampal contribution to verbal fluency in patients with temporal lobe epilepsy. Cortex. 2001;37:55–63.

E. Glisky, D.L. Schacter, E. Tulving. Learning and retention of computer-related vocabulary in memory-impaired patients: method of vanishing cues. J. Clin. Exp. Neuropsychol.. 1986;8:292–312.

K. Graham, J. Hodges. Differentiating the roles of the hippocampal complex and the neocortex in long-term memory storage: evidence from the study of semantic dementia and Alzheimer's disease. Neuropsychology. 1997;11:77–89.

M. de Haan, M. Mishkin, T. Baldeweg, F. Vargha-Khadem. Human memory development and its dysfunction after early hippocampal injury. Trends Neurosci.. 2006;29:374–381.

F. Haist, J. Bowden Gore, H. Mao. Consolidation of human memory over decades revealed by functional magnetic resonance imaging. Nat. Neurosci.. 2001;4:1139–1145.

D. Hassabis, D. Kumaran, S.D. Vann, E.A. Maguire. Patients with hippocampal amnesia cannot imagine new experiences. Proc. Natl. Acad. Sci. U. S. A.. 2007;104(5):1726–1731.

D. Hassabis, E. Maguire. Deconstructing episodic memory with construction. Trends Cogn. Sci.. 2007;11:299–306.

M.E. Hasselmo, E. Schnell, E. Barkai. Dynamics of learning and recall at excitatory recurrent synapses and cholinergic modulation in rat hippocampal region CA3. J. Neurosci.. 1995;15:5249–5262.

J. Hodges, K. Graham. A reversal of the temporal gradient for famous person knowledge in semantic dementia: implications for the neural organization of long-term memory. Neuropsychologia. 1998;36:803–825.

A. Hupbach, R. Gomez, O. Hardt, L. Nadel. Reconsolidation of episodic memories: a subtle reminder triggers integration of new information. Learn. Mem.. 2007;14:47–53.

N. Kapur, K.J. Friston, A. Young, C.D. Frith, R.S. Frackowiak. Activation of human hippocampal formation during memory for faces: a PET study. Cortex. 1995;31:99–108.

Keane, M.M. and Verfaellie, M. (2006) Amnesia II: Cognitive issues. In: Patient-based approaches to cognitive neuroscience, 2nd edn. The MIT Press, Cambridge, UK, pp. 303–314.

C.B. Kirwan, P.J. Bayley, V.V. Galvan, L.R. Squire. Detailed recollection of remote autobiographical memory after damage to the medial temporal lobe. Proc. Natl. Acad. Sci. U. S. A.. 2008;105:2676–2680.

S.B. Klein, J. Loftus, J.F. Kihlstrom. Memory and temporal experience: the effects of episodic memory loss on an amnesic patient's ability to remember the past and imagine the future. Soc. Cogn.. 2002;20:353–379.

M.D. Kopelman, B. Wilson, A. Baddeley. The Autobiographical Memory Interview. Bury St Edmunds: Thames Valley Test Company; 1990.

T.K. Landauer, S.T. Dumais. A solution to Plato's problem: the latent semantic analysis theory of acquisition, induction, and representation of knowledge. Psychol. Rev.. 1997;104:211–240.

C.L. Leveroni, M. Seidenberg, A.R. Mayer, L.A. Mead, J.R. Binder, S.M. Rao. Neural systems underlying the recognition of familiar and newly learned faces. J. Neurosci.. 2000;20:878–886.

M.D. Lezak, D.B. Howieson, D.W. Loring. Neuropsychological Assessment, 4 edn. New York: Oxford University Press; 2004.

J. Lou, K. Niki. Role of medial temporal lobe in extensive retrieval of task-related knowledge. Hippocampus. 2002;12:487–494.

E.A. Maguire, R.S. Frackowiak, C.D. Frith. Recalling routes around London: activation of the right hippocampus in taxi drivers. J. Neurosci.. 1997;17:7103–7110.

E.A. Maguire, R.N. Henson, C. Mummery. Activity in prefrontal cortex, not hippocampus, varies parametrically with the increasing remoteness of memory. Neuroreport Rapid Comm. Neurosci. Res.. 2001;12:441–444.

E.A. Maguire, C.J. Mummery. Differential modulation of a common memory retrieval network revealed by positron emission tomography. Hippocampus. 1999;9:54–61.

E.A. Maguire, R. Nannery, H.J. Spiers. Navigation around London by a taxi driver with bilateral hippocampal lesions. Brain. 2006;129(11):2894–2907.

J.R. Manns, R.O. Hopkins, L.R. Squire. Semantic memory and the human hippocampus. Neuron. 2003;38:127–133.

J. McGaugh. Memory: a century of consolidation. Science. 2002;287:248–251.

B. Milner. The medial temporal-lobe amnesic syndrome. Psychiatr. Clin. North Am.. 2005;28:599–611.

M. Moscovitch. Recovered consciousness: a hypothesis concerning modularity and episodic memory. J. Clin. Exp. Neuropsychol.. 1995;17:276–290.

M. Moscovitch, L. Nadel. Multiple-trace theory and semantic dementia. Trends Cogn. Sci.. 1999;3:87–89.

M. Moscovitch, L. Nadel, G. Winocur, A. Gilboa, S. Rosenbaum. The cognitive neuroscience of remote episodic, semantic and spatial memory. Curr. Opin. Neurobiol.. 2006;16:179–190.

Nadel, L. (2008) Multiple memory systems: a new view. In: Roedieger, H.L. III (Ed.), Cognitive Psychology of Memory. Vol. 1 of Learning and Memory: A Comprehensive Reference. J. Byrne (Ed.), Oxford: Elsevier, Vol. 4, pp. 41–52.

Nadel, L., Campbell, J.L. and Ryan, L. (2007a) Autobiographical memory retrieval and hippocampal activation as a function of repetition and the passage of time. Neural Plasticity, Online http://www.hindawi.com Article 90472.

L. Nadel, M. Moscovitch. Memory consolidation, retrograde amnesia and the hippocampal complex. Curr. Opin. Neurobiol.. 1997;7:217–227.

L. Nadel, M. Moscovitch. Hippocampal contributions to cortical plasticity. Neuropharmacology. 1998;37:431–439.

L. Nadel, A. Samsonovich, L. Ryan, M. Moscovitch. Multiple trace theory of human memory: computational, neuroimaging, and neuropsychological results. Hippocampus. 2000;10:352–368.

L. Nadel, J. Willner, E.M. Kurz. Cognitive maps and environmental context. In: P. Balsam, M. Tomie, eds. Context and Learning. Hillsdale: Lawrence Erlbaum Associates; 1985:385–406.

Nadel, L., Winocur, G., Ryan, L. and Moscovitch, M. (2007b) Systems consolidation and hippocampus: two views. Debates Neurosci., online.

J. O’Keefe, L. Nadel. The Hippocampus as a Cognitive Map. Oxford: Oxford University Press; 1978.

M.N. Rajah, A.R. McIntosh. Overlap in the functional neural systems involved in semantic and episodic memory retrieval. J. Cogn. Neurosci.. 2005;17:470–482.

P.V. Rekkas, R.T. Constable. Evidence that autobiographic memory retrieval does not become independent of the hippocampus: an fMRI study contrasting very recent with remote events. J. Cogn. Neurosci.. 2005;17:1950–1961.

R.S. Rosenbaum, G. Winocur, C. Grady, M. Ziegler, M. Moscovitch. Memory for familiar environments learned in the remote past: fMRI studies of healthy people and an amnesic person with extensive bilateral hippocampal lesions. Hippocampus. 2007;17:1241–1251.

Ryan, L., Cox, C., Hayes, S. and Nadel, L. (in press) Hippocampal activation during episodic and semantic memory retrieval: comparing category production and category cued recall. Neuropsychologia.

L. Ryan, L. Nadel, K. Keil, K. Putnam, D. Schnyer, T. Trouard, M. Moscovitch. Hippocampal complex and retrieval of recent and very remote autobiographical memories: evidence from functional magnetic resonance imaging in neurologically intact people. Hippocampus. 2001;11:707–714,.

H. Schmolck, E.A. Kensinger, S. Corkin, L.R. Squire. Semantic knowledge in patient H.M. and other patients with bilateral medial and lateral temporal lobe lesions. Hippocampus. 2002;12:520–533.

J. Snowden, H. Griffiths, D. Neary. Semantic-episodic memory interactions in semantic dementia: implications for retrograde memory function. Cogn. Neuropsychol.. 1996;13:1101–1137.

E.N. Sokolov. Neuronal models and the orienting reflex. In: M.A.B. Brazier, ed. The Central Nervous System and Behaviour. New York: Josiah Macy Jr. Foundation; 1960:187–276.

L.R. Squire. Memory and the hippocampus: a synthesis from findings with rats, monkeys, and humans. Psychol. Rev.. 1992;99:195–231.

L.R. Squire, P. Alvarez. Retrograde amnesia and memory consolidation: a neurobiological perspective. Curr. Opin. Neurobiol.. 1995;5:169–177.

L.R. Squire, C.E. Stark, R.E. Clark. The medial temporal lobe. Annu. Rev. Neurosci.. 2004;27:279–306.

L.R. Squire, S.M. Zola. Episodic memory, semantic memory, and amnesia. Hippocampus. 1998;8:205–211.

S. Steinvorth, B. Levine, S. Corkin. Medial temporal lobe structures are needed to re-experience remote autobiographical memories: evidence form H.M. & W.R. Neuropsychologia. 2005;42:479–496.

E. Tulving. Elements of Episodic Memory. Oxford: Clarendon Press; 1983.

E. Tulving. Episodic memory: from mind to brain. Annu. Rev. Psychol.. 2002;53:1–25.

E. Tulving. Episodic memory and autonoesis: uniquely human? In: H.S. Terrace, J. Metcalfe, eds. The Missing Link in Cognition: Origins of Self-Reflective Consciousness. Oxford University Press; 2005:3–56.

F. Vallée-Tourangeau, S.H. Anthony, N.G. Austin. Strategies for generating multiple instances of common and ad hoc categories. Memory. 1998;6:555–592.

S. Vicari, D. Menghini, M. Di Paola. Acquired amnesia in childhood: a single case study. Neuropsychologia. 2007;45:704–715.

O. Vinogradova. Registration of information and the limbic system. In: G. Horn, R.A. Hinde, eds. Short-Term Term Changes in Neural Activity and Behaviour. Cambridge: Cambridge University Press; 1970:94–140.

W.H. Walker, W. Kintsch. Automatic and strategic aspects of knowledge retrieval. Cogn. Sci.. 1985;9:261–283.

R. Westmacott, M. Moscovitch. The contribution of autobiographical significance to semantic memory. Mem. Cogn.. 2003;31:761–774.

C. Whatmough, H. Chertkow. rCBF to the hippocampal complex covaries with superior semantic memory retrieval. Behav. Brain Res.. 2007;181:262–269.

M.D. Williams, J.D. Hollan. The process of retrieval from very long-term memory. Cogn. Sci.. 1981;5:87–119.

B. Wilson, A. Baddeley. Semantic, episodic, and autobiographical memory in a postmeningitic amnesic patient. Brain Cogn.. 1988;8:31–46.

M. Wilson, B. McNaughton. Reactivation of hippocampal ensemble memories during sleep. Science. 1994;265:676–679.

B.J. Wiltgen, A.J. Silva. Memory for context becomes less specific with time. Learn. Mem.. 2007;14:313–317.

G. Winocur, M. Moscovitch, M. Sekeres. Memory consolidation or transformation: context manipulation and hippocampal representations of memory. Nat. Neurosci.. 2007;10:555–557.

Chapter 1.2 Exploring episodic memory

Martin A. Conway*

---

The Leeds Memory Group, Institute of Psychological Sciences, University of Leeds, Leeds, UK

* Corresponding author.

E-mail address: [e-mail]

Abstract

Fourteen characteristics of episodic memory are outlined and explored in detail. Episodic memories are conceived as sensory-perceptual-conceptual-affective summary records of experience. It is proposed that they are represented separately from other autobiographical knowledge in a phylogenetically older memory system located in posterior brain regions. Autobiographical knowledge provides a conceptual context for episodic memories and episodic memories in turn provide the basis for conceptual knowledge.

Keywords

• episodic memory • autobiographical memory • temporal lobes • phylogenetic • ontogenetic • memory development • recollective experience • memory awareness

It is a curious thing remembering specific moments-fragments of the past. Why do we do it? What possible use can it be? Especially when the cognitive system appears remarkably well adapted to detecting regularities in experience and retaining these over long periods of time. Surely, in terms of survival that is all that is required: an ability to detect regularities and co-occurrences, retain these, access them at appropriate times, i.e., when cued, and respond to them. Retaining records of one-off experiences would seem to be pointless given that exactly the same experience is unlikely to be repeated and if it or a similar experience were repeated then it becomes a regularity and better represented in memory by some more abstract structure, for example, a schema. (It is interesting to note that the great champion of the schema concept, Bartlett (1932), had little interest in episodic memory, EM.) So why do we have EMs? And, equally fundamentally, what are they? In what follows I try to arrive at some answers to these questions. Throughout I am guided by Tulving's original insights into EM, although I also introduce some new lines of thought.

I Characteristics of episodic memory

Tulving in his seminal 1972 paper defined EM by contrasting it with semantic memory. One of his goals in doing this was to simplify and reduce the number of putative memory system proposals which at that time proliferated. For example, different memory systems were proposed for different types of experimental materials, e.g., action memory, picture memory, odor memory, etc. Very few of these fledgling memory systems survived much beyond their initial formulation and even the concept of semantic memory has fallen into disuse. EM, however, survives and has proved an increasingly useful concept in understanding disruptions of memory following brain damage and in psychological illness, and in understanding human autobiographical memory and animal memory. In the development of the EM concept there has been a move away from a definitional approach to an approach that focuses more on unique features or sets of features (Tulving, 1983, 2002a,b; Wheeler et al., 1997; Conway, 2001, 2005). Table 1 lists EM characteristics grouped under the headings: content, function, phenomenology, brain, and development. In what follows, these groups of characteristics of EM are considered in detail.

Table 1 Fourteen characteristics of episodic memories

II The content of episodic memories

A major, possibly even defining feature of EM is that it contains information dating to a unique moment in time and not just any time or time in general but rather to the time of an individual experiencing self. But what exactly is this information? Studies of autobiographical memory often distinguish between what we have termed autobiographical knowledge and EM (Conway, 1992, 2001, 2005; Conway and Williams, 2008, and see Conway and Pleydell-Pearce, 2000, for a review). Consider the following memory description:

I am on our swing, which is tied to one of the pine trees in the glade at the back of the lodge. I see Frances coming through the trees, white cotton dress and black ringlets – she is struggling with a heavy bucket of water on her left-hand side. Me, so happy to see her, say ‘Hello Frances’ – my voice piping and hopeful but her eyes are oddly emotionless. She comes closer and tips a full bucket of ice-cold water over me.

This EM has all the features listed in Table 1 under the heading Content: it contains sensory-perceptual details, clearly some of these are highly available and some, unrecalled here, are less available, it is a highly visual EM with other sensory-perceptual details too, and has a perspective, in this case approximating to what may have been the original perspective (from the seat of the swing). These are key features of an EM and I suggest that taken together they are defining. However, this particular presentation would not according to our thinking be classified as an autobiographical memory and that is because it lacks much in the way of contextualizing autobiographical knowledge. Consider the full memory description, given incidentally in response to a web survey (conducted in 2006 by the author) of self-defining memories (see Singer, 2005):

I lived in a small lodge house by a pine wood with my mother, father and older half-brother. My two half-sisters were up in Scotland for rare visit. David, Harriet and Frances were hiding in an orange tent on Sweet Hillocks, the field behind the house. They were obviously discussing why the three of them had been split up and were now living 600 miles apart from each other; how unfair it was (which it was). My mother had come and split their family up and I was the result, the living sign – a strawberry haired, chap-lipped mistake. I wasn’t allowed out of the garden but was grateful to be allowed somewhere near these exotic creatures who had landed and made our family suddenly huge. Frances has black, curly hair; wore dainty dresses and was closest in age to me. I guess she was also the angriest and easily egged on by Harriet. It's summer. I am on our swing, which is tied to one of the pine trees in the glade at the back of the lodge. I see Frances coming through the trees, white cotton dress and black ringlets – she is struggling with a heavy bucket of water on her left-hand side. Me, so happy to see her, say ‘Hello Frances’ – my voice piping and hopeful but her eyes are oddly emotionless. She comes closer and tips a full bucket of ice-cold water over me. I have one second of shock, long enough to see her turn and walk away. She seems tired. Then I feel the chill.

By our view then EMs represent short time slices of experience and these are either actively constructed into, or in their representation in long-term memory, embedded in contextualizing autobiographical knowledge structures (see Fig. 1 in Conway and Williams, 2008). Note, however, that it is not suggested that the content of EMs correspond directly to experience. Rather we argue that EM content is experience-near. EM content typically contains sensory-perceptual-conceptual-affective summary features rather than literal records of experience that could be relived. It may be that occasionally EM features are literal records of experience and can, because of this specificity, support some degree of reliving. But as this is most