The term 'declarative memory', depending on the "context of the discussion, refers to a faculty and experience of memory (definition 1), the material stored and "retrieved (definition 2), and the relevant brain system(s) (definition 3). That part of our memory is directly accessible to conscious recollection but part is not, was first explicitly stated by philosophers (e.g. de Biran 1804; Bergson 1908).1 Ryle (1949) formulated it as the distinction between 'knowing that' and 'knowing how'. 'Knowing how' refers to "skills and procedures, 'knowing that' to information that can be 'declared', i.e. declarative memory.2 Declarative memory is conventionally further subdivided into memory for facts ('semantic'), and memory for episodes ('episodal', 'autobiographical'; Tulving 1983; "episodic memory). Episodic memory, and sometimes semantic memory, single-trial learning, yet could be modified over time, either by additional facts, new experiences, or retrieval in new contexts. Episodic memory includes information about an experience locked to a particular time and place, whereas semantic memory is not locked to specific coordinates in these "dimensions. Some authors classify 'episodic' apart from 'declarative' (Tulving and Markowitsch 1998). Similarly, in episte-mology, 'knowledge by acquaintance', i.e. of people, places, and things, is distinguished from propositional or factual knowledge (Bernecker and Dretske 2000). The term 'cognitive memory' is also occasionally used for recollection with conscious awareness (Mishkin et al. 1997). 'Explicit' and 'implicit' (Graf and Schacter 1985; Schacter 1987) are sometimes used in the literature instead of 'declarative' and 'nondeclarative', respectively. Others, however, prefer to consider the explicit/ implicit dichotomy to be used in the "taxonomy oflearning tasks and memory tests ("assay) rather than for that of memory "systems and mechanisms (e.g. Johnson and Hasher 1987; N.J. Cohen et al. 1997).3
Support in favour of the declarative/nondeclarative distinction has surfaced over the years not only via introspection but also in "controlled experiments in normal individuals (e.g. McDougall 1923; Eriksen 1960; Richardson-Klavehn and Bjork 1988).Yet the evidence that the brain indeed honours this distinction was ultimately provided by the neuropsychological investigation of "amnesia in humans and its "models in the "monkey (Cohen and Squire 1980; Squire and Zola 1996). It has been noted for years that the memory deficits in 'global' amnesics are not really global (e.g. Corkin 1968; Warrington and Weiskrantz 1968). A "classic study illustrates this point. Cohen and Squire (1980) subjected amnesic patients to a mirror-reading skill test, involving presentation of mirror-reflected words over consecutive sessions. Some words were presented only once and some were repeated. The reading time of the unique words was used to evaluate the ability to "acquire the procedure of mirror-reading, while the reading time of the repeated words reflected, in addition, the ability to remember specific data. Amnesic patients learned the mirror-reading skill, as indicated by the decrease with training of the time required to read the unique words. Their performance, however, did not improve further on tests of repeated words as it did in controls. Furthermore, none of the amnesics reported that the repeated words were indeed encountered beforehand. Hence, the nonamnesics learned both 'how' and 'that'; the amnesics only 'how'. Similar conclusions were obtained in different types of 'global' amnesics, using a variety of tasks that assay 'rule' ('how') vs. 'data' ('that') knowledge (Squire and Zola 1996).
The aforementioned data have been taken to indicate that amnesics can acquire nondeclarative information, and that mediotemporal brain structures damaged in 'global' amnesics ("hippocampus, "limbic system) are necessary for declarative but not for nondeclarative tasks. But is this indeed due to the existence of different brain circuits for declarative and nondeclarative information, respectively? The argument could still be raised that declarative and nondeclarative memory are subserved by the same circuits, but there is a general impairment in the processing of information in the brain, that affects only declarative memory.4 Compelling evidence against this single-system hypothesis was provided by double-dissociation experiments. 'Double-dissociation' refers to a protocol in which the effect of two different circumscribed lesions, in areas A and B, is tested on two different phenotypes, X and Y (Teuber 1955; "control). If lesion A yields a defect in X but not in Y, while lesion B yields a defect in Y but not in X, then the defects in X and Y are not due to some across-the-board damage but rather to specific damage in the dissociable functions of areas A and B, respectively. The 'global' amnesics mentioned above showed only a 'single dissociation', i.e. lesion A yielded a defect in X but not in Y. For double dissociation, patients were sought that fail on nondeclarative but not on declarative tasks. Such patients were indeed identified (Gabrieli et al. 1995; Knowlton et al. 1996). For example, Parkinson's patients failed on a task that involved the acquisition of a mental "habit, i.e. nondeclarative knowledge, in spite ofbeing able to acquire and "retrieve new declarative knowledge (Knowlton et al. 1996). Furthermore, this identifies the neostriatum, which is damaged in Parkinson disease, as critically involved in nondeclarative, but not declarative, memory.
"Functional neuroimaging "methods have further reinforced the conclusion that in humans the mediotemporal lobe subserves the acquisition and retrieval of declarative memory, in concert with the prefrontal cortex (e.g. Nyberg et al. 1996; Wagner et al. 1998a; Kirchhoff et al. 2000). There is, however, an ongoing debate in the literature whether the hippocampal formation contributes equally to both episodic and semantic memory, or primarily to episodic memory (Mishkin et al. 1997; Squire and Zola 1998; Tulving and Markowitsch 1998).
As a hallmark of the expression of declarativeness in humans is language, the task of identifying declarative memory in subhuman species could become tricky.5 There are three basic approaches to the problem.
1. Investigation in animals of the performance and brain mechanisms of tasks that are declarative in humans. These could be either tasks that normal "subjects report as involving awareness, or tasks sensitive to human amnesia (e.g. Zola-Morgan and Squire 1985; Clark and Squire 1998; Manns et al. 2000; "monkey). Some pure "recognition tasks (see "delay task) could be useful in this respect. The main problem is that different species, even different individuals and the same individual under different circumstances, can solve the same problem using different strategies.
2. Investigation in animals of the effect on learning and memory of circumscribed lesions in mediotemporal and diencephalic brain regions that are assumed to subserve declarative memory in humans (e.g. Zola-Morgan and Squire 1985; Squire and Zola 1996; Mishkin et al. 1997; "hippocampus, "monkey). Here the difficulties are, first, it is unlikely that brain circuits that subserve declarative tasks in humans do only that; the hippocampus, for example, is involved in memory with or without conscious awareness (e.g. Chun and Phelps 1999). Second, the same brain regions may fulfil different functions in different species. Further, the rationale risks circularity: a task may be deemed declarative in humans because it depends on a brain region that is postulated to subserve declarativeness, and then is labelled declarative in animals because it depends on this same region.
3. A different approach is to attempt to identify primitives of declarative memory, which are shared by prehuman manifestations of declarativity. The assumption is that declarative systems have emerged in evolution prior to the ability of "Homo sapiens to experience and express declarativity the human way. This assumption is taken by the followers of approaches 1 and 2 above as well, but here it is bolder and more speculative, because it searches for the deep structure of the computational theory ("level) of the memory system. It is further assumed that 'declarativess' is not necessarily the decisive distinctive attribute of declarative memory systems. It has been argued, for example, that the elementary characteristic of declarative memory is the ability to encode internal representations according to relationships among specific items, and flexibly "generalize and integrate this information in novel situations (N.J. Cohen et al. 1997; Eichenbaum 1997a). This, so goes the claim, differs from nondeclarative memory, which is more rigid (for opposition to this view, see Willingham 1998). Guided by this line of argument, 'paired associates tasks' were employed to tap into inferential, flexible memory. In paired associates tasks, the subject learns a list of discrete pairs of associations, denoted generically A-B, where typically A is to serve as a "cue for the recall of B or for a response that stems from the recall of B. Paired-associate learning involves multiple cognitive processes (e.g. McGuire 1961), including acquisition of associations among discrete "stimuli and of stimulus-response rules. In humans, A-B could be arbitrary verbal stimuli, whereas in laboratory animals, A-B are nonverbal sensory cues. Hippocampally lesioned animals fail on some versions of paired associates tasks that tax flexibility of response, for example, on the ability to associate paired elements presented in the reverse of training order (Eichenbaum 1997a). This was taken to support the idea that such tasks can be used to test declarative memory in nonhuman species.
A nice spin-off of the flexibility + generalization = declarativeness hypothesis is that it is in line with the classification of "priming as an intermediate stage between nondeclarative and declarative capabilities, with repetition priming being less flexible and less generaliz-able, conceptual priming being more flexible and more generalizable, yet both still independent of conscious awareness.6 But identification of the flexibility of representational reconstruction as the core attribute of declarative memory system is only a working hypothesis. Note that if primitives of declarative memory independent of conscious awareness were to be identified, the definitions of declarative memory above will have to be modified, or become a subtype of a more comprehensive definition.
All this leads us to the question whether the term 'declarative' is appropriate to be used in animal studies. The distinction 'explicit'/'implicit', despite the aforementioned occasional claim that it fits tests better than systems, may be more useful in animal studies, especially because it does not explicitly refer to 'declara-tiveness' with its linguistic roots and connotations. It also blunts the need for conscious awareness, which is problematic, especially in "simple systems. We do not know at which stage of evolution conscious awareness has entered the scene (Tulving 1983; Moscovitch 1996; Eichenbaum 1999). This is definitely an issue in which the prudent use of "Ockham's razor to eliminate superfluous "anthropomorphism is advisable.
Selected associations: Amnesia, Episodic memory, Hippocampus, Learning, Taxonomy
''Recollection' refers here to both 'recall and *recognition, see there.
2In linguistics, a 'declarative sentence' makes a statement, i.e. informs someone of something (Winograd 1972; Lyons 1977).
3See also on tasks of implicit vs. explicit learning, under 'learning. In many cases, information learned implicitly can later be used either in a declarative or in a nondeclarative mode, and the same is true for information learned explicitly.
4For example, it could be argued that amnesics are incapable of processing information at a 'level that permits proper 'acquisition and 'retrieval of declarative information, but still capable of processing at a 'shallower' level, presumably sufficient for nondeclarative memory. On levels of processing in acquisition and retrieval, see also 'dimension.
5For the sake of simplicity, no distinction is made here between memory of facts and memory of episodes. It is clearly possible that some nonhuman species have declarative memory of one type but not the other. See also 'episodic memory.
6On the hypothetical place of priming in the evolution of memory systems, see also Tulving (1983).
Was this article helpful?