Episodic recollection is conscious time travel into the subjective past1. This is considered by many as the most sophisticated faculty of memory, and the one most characteristic of humans. Further, a minority of authors even use the term 'memory and 'episodic memory' as synonyms. These authors would usually concede that the "declarative memory of impersonal facts is also 'memory', but refuse to consider many other types of use-dependent change in behaviour, such as "classical conditioning, "habit and "skill, as bona fide memory. This is evidently not the stand taken in this book ("memory).
The term 'episode' is from the Greek epi-, on, at, + eisodios, entering; epeisodion meant a parenthetic narrative. Indeed an episodic memory item is only one scene in the individual's accumulative personal narrative2. That there is memory which is episodic, as opposed to automatic responses, or to the recall of impersonal facts ("taxonomy), was discussed already by Greek philosophers if not earlier. A clear description of this distinct type of memory was given by Agusutine (400; "classic):'... I encounter myself and recall myself, and what, and when, and where I did some did, and how I was affected when I did it'. The term 'episodic memory' was, however, introduced into the scientific discourse much later, by Tulving (1972). This was on the background of intensive developments in cognitive research on the different categories of human knowledge. The new computer era opened new vistas for understanding the storage and "retrieval of knowledge, which infiltrated human memory research. In the mid-1960s, Quillian proposed a method by which meaning could be stored optimally in a computer program, and his postulates were adapted by psychologists in their attempt to account for the storage of knowledge in real brains (Collins and Quillian 1969; Tulving and Donaldson 1972). This type of knowledge encoding was termed 'semantic memory' ("declarative memory, "taxonomy). Tulving (1972, 1983) went on to distinguish semantic memory, which he dubbed 'mental thesaurus', from another type of knowledge that humans have about the world, which is about temporally dated events, describable in terms of their perceptible attributes, and the spatiotemporal relations of these events to each other. This memory he termed 'episodic'. The term is now used, depending on the context of discussion, to denote a type of mental act or experience (definitions 1,2), a type of memory items (definition 3), or a type of memory system (definition 4)3.
Here are a few things to remember about episodic memory:
1. It is not only about what, how and where, but also about when; the ability to place unique experienced events on the temporal axis of personal history, be it distorted and poorly resolved, is characteristic of episodic memory.
2. It refers to unique experiences, hence each episodic item is the outcome of'single trial' learning ("acquisition). Recalled "real-life episodes, however, refer each to different time windows of experience. Further recollection effort may dissect amalgamated episodes into elementary events.
3. In humans, it involves autonoetic awareness, i.e., the conscious reflexive experience of private phe-nomena4. Whether non-human species are also capable of experiencing such 'mental time travel', and if so, what kind of awareness is involved, is an open question (see below). Definition 2 does not posit awareness and therefore fits particularly to be used in animal research; this definition, however, is considered by many authors to refer to 'episodic-like' rather than to genuine episodic memory.
4. Whereas retrieval of memory is commonly oriented toward the present or the immediate future, i.e. application of knowledge for ongoing needs, retrieval of episodic memory is oriented toward the past, i.e. the past events are recognized as such (Tulving and Markowitsch 1998). The retrieved information, however, might then be harnessed for identifying and attaining future goals (Conway and Pleydell-Pearce 2000).
5. The retrieval of episodes is expected to involve multiple steps, or components, which ultimately differ in the specificity of the information retrieved, such as the recollection of specific information about the target item ('what') and about its source ('where', 'when', as well as the sequence in which the elements of the episode unrolled), and the recognition of the familiarity of each of these items (e.g., Yonelinas 2001 ; see also 'remembering' vs. 'knowing' in "recognition).
6. Many types of tests could be used to tap into episodic memory, including free recall, cued recall, target and source recognition5, etc. But not all the tests referred to in the literature as paradigmatic tests of episodic memory, are indicative that such memory has indeed been encoded ("criterion). Paired associates learning is but one example. In this type of learning, the subject is presented in training with a pair of stimuli, and then, in the test, presented with one member of the pair and requested to retrieve the memory of the other. Word pairs are commonly used in humans (e.g. Winocur and Weiskrantz 1976; Shallice et al. 1994), non-verbal stimuli in other species (e.g. Saunders and Weiskrantz 1989; Sakai and Miyashita 1991; see also pp. 74-75). The task is sensitive to mediotemporal damage ("amnesia), substantiating its declarative nature. Paired associates are considered 'episodic' tests because, when the pairs used are unique, their recall unveils the memory formed in a single episode of experience. Yet, depending on the protocol used and particularly when the memory of 'when' is not taxed, success on such tasks might be construed as the outcome ofthe formation of a semantic association, including sometimes "priming, independent of mental time travel.
Where is it in the brain? Being a declarative type of memory, episodic memory is subserved by "cortical circuits, including dependence on the mediotemporal lobe for acquisition and "consolidation, and on the frontal lobe, among others, for retrieval (e.g., Fletcher et al. 1997; Eldridge et al. 2000; Kirchhoff et al. 2000; Lepage et al. 2000; for more on functional differentiations in these areas by types of tasks and memory "phases, see "acquisition, "declarative memory, "consolidation, "retrieval). A prominent question is whether the brain circuits that subserve episodic memory differ from those that subserve semantic memory6.
No consensus answer is available to this question. The data that feed the debate stem from the study of brain lesions and from "functional neuroimaging. Whereas functional neuroimaging can unveil correlation of activity in identified brain areas with the performance on episodic tasks, currently, only the study of brain lesions could pinpoint candidate obligatory circuits. In some types of "dementia, the deterioration of semantic and episodic memory is differential, suggesting that the biological substrates of these two types of memory differ (e.g. Hodges et al. 1999; it is noteworthy that normal ageing is often associated as well with preferential impairment of episodic memory, e.g. Nilsson et al. 1997). Studies of amnesic patients provide mixed evidence (reviewed in Mishkin et al. 1997; see Tulving et al. 1988 for a case of'semantic' amnesia; also ZolaMorgan et al. 1983 for some methodological caveats). A particularly interesting type of information has been provided by the study of patients with amnesia due to hippocampal damage sustained very early in life (Vargha-Khadem et al. 1997). These patients were capable of developing normal language and social competence, and acquire new factual knowledge, in spite of displaying severe loss in episodic memory. Based on these data, a suggestion was made that the "hippocampus and the subhippocampal cortici form a hierarchically organized system for the registration of declarative knowledge. Episodic memory, so goes the idea, which depends on rich associations, is encoded in the hippocampus itself, the top processing level in the hierarchy, whereas semantic memory, less dependent on intricate associations, can be supported by the subhippocampal cortici, lower on the hierarchy (Mishkin et al. 1997, 1998). Alternative interpretations of the data, which do not support such a qualitative distinction between the semantic and the episodic systems, were also raised (Squire and Zola 1998).
Do animals have it? If one posits autonoetic awareness as a critical criterion for episodic memory, the issue of whether animals go on mental time travels becomes complicated indeed and by some accounts insolvable. Investigators who address the problem in non-human species almost always set a more modest, yet still admirable, goal: to prove that the animal has 'episodic-like' memory, which does not presuppose conscious awareness as a defining criterion. An interesting strategy, which is actually recommended as a general strategy in memory research, is to look for the behavioural ecology of the species, and search for natural situations which could benefit from episodic-like memory. Clayton et al. (2001) list a number of potential candidate systems. One is brood parasitism: brood parasitic birds such as the cuckoo deposit eggs in the nests of other species and the young are later cared for by the host species. A successful brood parasite must remember where potential victims have started their nests, and return to that place at the right time to add the egg to those already laid in the nest. This could involve where, what, and when information of a single event. Another potential candidate is a polygynous mating system, such as in the meadow vole, in which a male mates with multiple females that are distributed over a wide area and come into estrus at different times. Successful mating in this case could greatly benefit from remembering who, where, and when. These candidate systems are somewhat difficult to investigate systematically in the laboratory. But another type of suitable behaviour was found amenable to systematic, controlled analysis: food caching in the scrub jay (Clayton and Dickinson 1998; Clayton et al. 2001). Scrub jays cache perishable insects and seeds in multiple locations, and it may be adaptive for them to remember what has been cashed where and when. Clayton et al. (1998) have trained scrub jays to appreciate that worms, which the birds like a lot, are perishable and therefore degrade over time. They then allowed the birds to recover the perishable worms, or non-perishable peanuts, which the birds had previously cached in distinct sites. The birds searched preferentially for fresh worms, their favorite food, when allowed to recover them shortly after caching, but learned to prefer searching for the less-favorable (but still tasty) peanuts and avoid searching for the worms after longer intervals in which these worms decayed. (Birds that did not have an opportunity to learn that worms degrade over time, continued to search for the cached worms even long after caching; and needless to say, measures were taken in the experiment to prevent odor cues of the degrading food.) This use-dependent modification in the food search strategy was taken to imply that the jays remembered not only what was deposited where, but also when it was deposited there—hence fulfilling the behavioural criteria for episodic-like memory (definition 2).
The story of the scrub jays shows that the demonstration of episodic-like memory in non-human species is possible, given the investigators are smart enough to match the right species with the right *assay (for an example of a different kind of approach, which attempts to identify primitives of episodic memory in the rat, see Fortin et al. 2002). But is episodic-like memory a rudimentary form of what we humans experience as episodic memory? And will it ever be possible for us to identify the conscious awareness of mental time travel, if it ever exists, in other species? Can we ever come to know how is it like to be a bat (Nagel 1974), particularly, a nostalgic bat? Some say that we will never be able to do this. But there might be another, though admittedly remote and less satisfactory solution. Suppose we identify in the human brain a characteristic functional tag of mental travel, e.g. a distinct activity pattern unveiled by functional neuroimaging of the behaving subject. This might be analogous, say, to the identification of candidate dream states by recording brain waves (e.g. Dave and Margoliash 2000). Suppose then that we identify this unique functional tag in the brain of animals when they perform a task that involves episodic retrieval in humans. We may then be able to say that there is a reasonable probability that the non-human subject performs a mental time-travel. Not enough to firmly conclude that this animal feels the same as we do when we recall our private past, but still, a hint that this might indeed be the case.
Selected associations: Amnesia, Conscious awareness, Declarative memory, False memory, Recall
1 This past is assumed by the subject to be veridical, but is not necessarily accurately recalled (see 'false memory). 'Confabulations and hallucinations are excluded.
2 How many scenes are there all together is an interesting question; probably less then most of us would predict intuitively ('capacity; Dudai 1997a).
3 'Autobiographical memory' is often used interchangeably with 'episodic memory'. It might be preferable to distinguish between the two.Whereas 'episodic' refers to distinct individual episodes, 'autobiographical' connotes the narrative formed from the combination of such episodes (see also Conway 2001).
4 Autonoetic ('self-comprehending') awareness is regarded in the scientific 'culture as the highest form of conscious awareness; certain Eastern philosophies use other 'taxonomies, with higher resolutions (e.g., Harvey 1990; Keown 2000).
5 'Source information' is a term used in human memory research to refer to the time and place in which the target item was acquired; see also 'context.
6 The question can be posed at various 'levels of analysis: is the functional system monolithic? Are the 'algorithms the same? And is the hardware implementation identical? The question as posed here refers to the hardware implementation at the circuit level but clearly reflects on the algorithmic and computational levels as well.
Was this article helpful?