حافظه اپیزودیک در حیوانات:یادآوری کدام مناسبت

Episodic memory, the recollection of past events in one's life, has often been considered a memory specific to humans. Recent work in a variety of species has challenged this view, and has raised important questions about the nature of episodic memory itself. We present a review of the types of task proposed as episodic-like in animals and consider that these tasks require animals to demonstrate memory for specific occasions in the past. We propose that identifying episodic memory as the memory for what happened where on a specific occasion is a more encompassing definition than one that relies on information about when an event occurred. These episodic-like tasks in animals support the view that the hippocampal system is necessary for episodic memory, and that the neural substrates of episodic memory can be dissociated from those of other forms of declarative memory.

Episodic memory is the conscious recollection of past events in our lives (Tulving, 1983). This type of declarative memory is particularly sensitive to the effects of ageing (e.g. Mitchell, Brown, & Murphy, 1990), disease (e.g. Graham, Emery, & Hodges, 2004) and brain injury (e.g. Aggleton & Brown, 1999). There has been much debate in the literature as to whether this is because episodic memory measures are more sensitive to a general impairment in declarative memory (Squire, Stark, & Clark, 2004) than other measures of memory, or because it relies on a separate, dissociable and presumably more vulnerable, neural system (Aggleton & Brown, 1999). Resolving this debate and understanding the neural processes underlying episodic memory is crucial to developing therapies for memory loss in a variety of conditions, and yet the understanding gained from human neuropsychology remains limited through the non-specific nature of damage within the crucial brain regions in patients. Therefore, developing animal models of episodic memory is crucial to developing our understanding of these neural mechanisms. However, as episodic memory has been intrinsically tied to conscious recall, demonstrating this in animals that do not have language to express their subjective experience is difficult, if not impossible. In recent years much progress has been made in developing animal models of episodic memory, and these models are helping us to understand the nature of the neural systems underlying episodic memory, and may also lead us to better understand the nature of episodic memory in humans more clearly by removing the phenomenological aspects of the memory. 1. What would episodic memory in animals look like? Episodic memory has been explicitly linked to autonoetic consciousness (Tulving, 2002), the awareness that an event happened to oneself. In humans such distinctions in types of awareness can be assessed (partially) through, for example, asking participants to report whether they ‘remember’ something they previously experienced, or simply ‘know’ it to be true (Yonelinas, Kroll, Dobbins, Lazzara, & Knight, 1998). However, such distinctions are often hard to convey to participants and rely entirely on the subjective experience of the participant. Moreover, use of this definition presents further problems. First, very few studies of human episodic memory ask about the subjective experience of the participant to ensure the memory is truly episodic in nature and so set an unrealistically high threshold for identifying episodic memory in non-human species. Moreover, in humans it is possible to have the subjective phenomena of an episodic memory without having actually experienced the remembered event (Roediger & McDermott, 1995), which brings into question the value of such phenomena. Second, whilst the difficulties of probing a form of subjective experience in humans with language are acknowledged, it becomes virtually impossible to study in species that do not have language. Therefore, some authors have claimed that this definition of episodic memory ties it specifically to humans, as autonoetic awareness cannot be demonstrated in other species (Clayton, Bussey, Emery, & Dickinson, 2003). This definition also ties us to the idea that the subjective experience is crucial to the memory as a whole. To some extent this is like suggesting we cannot understand, for example, visual processing of motion in animals without being certain that they have the same form of subjective experience when they see motion. Clearly the same evolutionary advantages we gain from episodic memory could be advantageous to other species too, and we therefore have to consider whether other animals have some form of episodic memory. To do this in the first instance, some researchers have proposed we look for ‘episodic-like’ memory (Clayton & Dickinson, 1998) which will have many of the features of episodic memory, but without the need to demonstrate any conscious experience. One approach is to consider the possible purpose of episodic memory. There are few evolutionary advantages to remembering the past unless it can influence future action. In the specific case of episodic memory where one remembers specific past episodes, this could be especially useful in imagining possible future events and planning for them on the basis of past experiences (Suddendorf and Corballis, 2007 and Suddendorf and Corballis, 2008a). This view is certainly supported by recent work which indicates that patients with impairments in episodic memory for past events also have problems in imagining future events (Hassabis, Kumaran, Vann, & Maguire, 2007). However, this approach does not specifically address issues about the nature of episodic memory content, other than to establish that such future planning must be behaviourally dissociated from other types of future directed behaviours, such as instinct (Suddendorf & Corballis, 2008a). Other approaches have considered very specifically the nature of the content of the episodic memory. In using Tulving's initial description of episodic memory as memory which “receives and stores information about temporally dated episodes or events, and temporal–spatial relations between them.” ( Tulving, 1983), Clayton and Dickinson identified that memories which carry information about what happened, where and when (what–where–when memory, Clayton & Dickinson, 1998) might fulfil the criteria for being ‘episodic-like’. Content alone cannot define an episodic memory however, as we can remember, for example, the date and place of our own birth without having an episodic memory of it. Therefore, work from this group has proposed that the what–where–when memory content should also be integrated into a single unified memory rather than three separate elements. Moreover, the integrated memory should be flexible; that is able to be accessed and used in novel situations not predicted at the point of encoding ( Clayton, Bussey, & Dickinson, 2003).

Whilst there is continuing argument about whether episodic memory is demonstrable in animals (Clayton et al., 2003a, Clayton et al., 2003c, Suddendorf and Busby, 2003 and Suddendorf and Corballis, 2007) it is clear that animals can display forms of memory that resemble episodic memory in many ways. The key component of this episodic memory in animals appears to be in demonstrating memory for a unique occasion, ideally with information about what happened and where it happened as well (Clayton et al., 2003c, Eacott and Easton, 2007 and Easton and Eacott, 2008). One crucial issue raised by these animal studies is the criteria set for evidence of episodic memory in animals are harder to achieve than those for humans as human tests of episodic memory rarely ask what happened where on a particular occasion and there are questions about the role of the subjective experience of remembering which has been claimed as being a necessity for the episodic memory (Tulving, 2002). Nonetheless, being forced to address such stringent criteria in animal tests of episodic memory is useful in forcing us to question the nature of the memory itself. For example, temporal information has been considered crucial to episodic memory, but work in animals has raised the question of what type of temporal information we use (Roberts et al., 2008) and prompted the question of whether temporal information is necessary at all (Easton and Eacott, 2008 and Friedman, 2007). Rather, as we propose here, episodic memory allows us to differentiate events which in many cases have high levels of similarity. Although temporal cues are one way of allowing us to identify unique occasions, other information can equally well serve to define those events (Easton & Eacott, 2008), whilst temporal cues can sometimes not differentiate them very well (Friedman, 2005). The crucial difference is that we claim that the content of the episodic memory need not contain any explicitly temporal information but that it must contain information which is sufficient to identify a specific past occasion and it is the occasion which has, by definition, temporal qualities. Finally, by generating models of episodic memory in animals (however simplistic those models might be in terms of the human experience) we are able to better understand the neural mechanisms of episodic memory, and the wider development of these tasks in common laboratory species such as rats and mice allow us to better understand these neural systems in a way that allows us to better translate that information to human disorders such as Alzheimer's disease.