A bilateral archi*cortical structure that extends as a ridge into the lateral ventricle, interconnects with multiple subcortical and cortical areas, and considered to play a critical part in certain types and *phases of memory.

In 1587 the Italian anatomist Arantius introduced the term 'hippocampus' (seahorse) to describe 'an elevation of white substance' that rises at the base of the lateral ventricles (Lewis 1923). He wondered, however, whether 'a white silk-worm' (Bombyx) might provide a better description. Two centuries later, a spark of dubious imagination led to a new suggestion, 'hippopotamus'. At about the same time, a structure in that same brain region also came to be called 'ram's horn', or, 'calculated to impress whoever hears it for the first time' (ibid.), the 'horns of Ammon' (after the Egyptian deity Ammon, which was represented as a ram or a human form with a ram's head). The road of'hippocampus' to glory was hence rather shaky, and the fact that we do not talk nowadays about 'bombycal "long-term potentiation' (LTP) or 'hippopotamal place cells' (see below) is due either to mere chance, or to the convoluted undercurrents of scientific etymology, or both. The seahorse won, and established itself quite firmly as a recurrent protagonist in the dreams or nightmares of brain scientists world-wide. Egyptian theology did leave a mark, nevertheless, in contemporary scientific literature: the hippocampus proper is still frequently referred to in neuroanatomy books and papers as 'Ammon's horn', and Cornua Ammonis is the source of the acronym 'CA' used to denote hippocampal subfields (Figure 34) (de No 1934; for additional historical notes, Meyer 1971).

Multiple terms are used in the hippocampal literature to refer to structures in the hippocampal region. The 'hippocampal formation' consists of the hippocampus proper, the dentate gyrus, and the subicu-lum. Occasionally, authors use the term 'hippocampus' to refer to the hippocampus proper + the dentate gyrus. The 'hippocampal system' includes in addition to the hippocampal formation the parahippocampal region, which contains the entorhinal, perirhinal, and parahip-pocampal cortices (Witter et al. 1989). The issue is not merely nomenclature. When authors report that they have performed an 'hippocampal lesion', it is crucial to note what is it exactly that they lesioned. The hip-pocampal formation interconnects either directly or indirectly with multiple subcortical and cortical areas. Without delving too much into neuroanatomy, it is useful to remember that the major cortical input to the hippocampus flows from the entorhinal cortex via the perforant path (abbreviated PP), and that major output to the cortex leaves via the subiculum. The intrinsic circuitry consisting of the successive projections from the entorhinal cortex to the dentate gyrus (via the PP), from the dentate gyrus to CA3 (via the mossy fibre pathway, MF), and from CA3 to CA1 (via the Schaffer collateral pathway, SC), is traditionally termed the 'trisynaptic pathway'. Major interconnections of the hippocampus with subcortical regions are mediated via the fornix.

The hippocampus is very popular in the neurosciences in general and in memory research in particular. In 2000, no less than 5640 papers included 'hippocampus' or 'hippocampal' in their title, abstract or keywords. This means on the average more than 15 papers per day, about 0.6% of the total papers published in science throughout that period, double the popularity of the cerebellum and the "amygdala combined, and an almost a fivefold increase over 1989

Hippocampus Cortical

Other cortical areas

Parahippocampal region

Hippocampal formation

Fig. 34 The hippocampus, the hippocampal formation, and adjacent cortical areas. (a) A drawing of the rabbit hippocampus in situ, as well as of an hippocampal slice. Only part of the pathways are indicated. CA1, CA3, pyramidal cell fields of the hippocampus; DG, dentate gyrus; MF, mossy fibre pathway; PP, perforant path; SC, Schaffer collateral pathway; Sub, subiculum. (Modified from Andersen et al. 1971.) (b) Outline of a horizontal rat brain section (right) illustrating the flow of information (schematically depicted on the left) between the hippocampal formation, the parahippocampal region, and some adjacent cortical areas. EC, entorhinal cortex; FF, fimbria-fornix (mediating information to subcortical areas); Hipp, hippocampus; OFC, orbito-frontal cortex; Pir, piriform cortex; PRC, preirhinal cortex; DG, dentate gyrus. Discussion of the role of the hippocampus in learning and memory must take into account not only the hippocampus or the hippocampal formation, but also the interconnections with the adjacent cortex. The computations performed by the hip-pocampal system are not yet known. (Adapted from Eichenbaum et al. 1996.)

(Science Citation Index 2001). Furthermore, about 25% of these 'hippocampal' papers explicitly mentioned learning or memory or LTP in their title or abstract or keywords. The majority of these papers, as did many thousands before them, reach the conclusion that the hippocampus is involved in learning and memory. But many of them fail to agree on what it does, and how and when it does it. The experimental evidence stems from multiple "methods and experimental "systems. Some data demonstrate correlation of hippocam-pal function with learning, others indicate an obligatory role ("criteria). Here is a brief illustration of the evidence.

1. Evidence from lesions. In human pathology it was noted already 100 years ago that damage to the medial temporal lobe, including the hippocampus, impairs memory (reviewed in Zola-Morgan et al. 1986). This was substantiated in patient H.M., who underwent bilateral removal of pieces of the medial temporal polar cortex, most of the amygdaloid complex, the entorhinal cortex, and a substantial part of the hippocampal formation, and as a result became densely "amnestic (Fig. 2, p. 11; Scoville and Milner 1957; Corkin et al.

1997). But rather amazingly, confirmation of whether hippocampal damage is indeed sufficient to induce amnesia proved difficult (e.g. Aggleton and Shaw 1996; Aggleton and Brown 1999). Human amnesia is a consequence of an unfortunate pathology that is seldom restricted to a single region. Accumulative evidence from a number of cases in which hippocampal damage was a major feature was therefore needed to reach the conclusion that "recognition memory is subserved by the hippocampal formation (e.g. Zola-Morgan et al. 1986; Kartsounis et al. 1995; Reed and Squire 1997,

1998); however, lesions in additional temporal cortex must be present to produce a severe memory loss (Reed and Squire 1998). A few points deserve special notice. First, the deficits are not confined to "declarative information (Chun and Phelps 1999). Second, the role of the hippocampus is important in long-term memory, but very remote memories, acquired before the damage had occurred, are spared (e.g. Teng and Squire 1999). This finding has led to the proposal that the hippocampus is required for a prolonged process of memory "consolidation, which in humans may require months or even years; once consolidated, memories become independent of hippocampal function (McClelland et al. 1995; Knowlton and Fanselow 1998; see also Haist et al. 2001). This view, supported by animal studies (e.g. Winocur 1990), does have opponents, who propose that the role of the hippocampus in some types of memory is not time limited (Moscovitch and Nadel 1998). And third, the hippocampus may be specifically important in event rather than fact memory (Vargha-Khadem et al. 1997); again, this is still debated (Mishkin et al. 1998; Squire and Zola 1998; Tulving and Markowitsch 1998; "taxonomy).

The major incentive for studying hippocampal lesions in the *monkey was to develop an animal "model for amnesia (reviewed in Squire and Zola-Morgan 1991; Eichenbaum et al. 1994; Murray and Mishkin

1998). Many of the studies have focused on "delay tasks, mainly delayed-nonmatching-to-sample (DNMTS), which were considered sensitive to human amnesia. Whereas early results have suggested that hippocampal lesions cause severe memory impairments, later studies unveiled complications. This was because the early lesions were not in fact confined to the hippocampus and included adjacent cortical areas. It was later found that lesions of the perirhinal and parahippocampal cortex that spared the hippocampus produced severe memory impairment on recognition tasks, whereas lesions restricted to the hippocampus yielded only minor deficits if at all (e.g. Zola-Morgan et al. 1989b; Murray and Mishkin 1998; "amygdala). To complicate life even further, it was also noted that versions of DNMTS as administered to monkeys may not be sensitive to recognition memory impairment in humans after all, because the extensive training of the monkey on that task involves acquisition of hippocampal-independent rules that facilitate "performance (Aggleton and Shaw 1996; Reed and Squire 1997). But all in all, the data from the monkey lesion studies do suggest that the hippocampus contributes to some aspects of recognition and inter"stimulus-association memory. It also plays a part in encoding spatial information (e.g. Gaffan 1998).

Hippocampal lesions in other mammals, such as in rodents and in the rabbit, impair performance on tasks involving spatial memory, "contextual memory, "working memory, and a variety of stimulus-stimulus and stimulus-action associations (e.g. Olton and Feustle 1981; Winocur 1990; Bunsey and Eichenbaum 1996; Clark and Squire 1998; Maren et al. 1998; Moser and Moser 1998a,b; Steckler et al. 1998b; Corbit and Balleine 2000). On some tasks, lesions of dorsal hippocampus were found to be more damaging than those in the ventral hippocampus (Hock and Bunsey 1998; Moser and Moser 1998a,b). In recent years, in addition to the "classical anatomical lesions, several other types of lesions have been used. These include transient pharmacological lesions, i.e. inhibition of targeted enzymes and receptors in the hippocampus (e.g. Riedel et al.

1999); "neurogenetic lesions, i.e. mutations in identified genes (e.g. Tsien et al. 1996a,b); and artificial saturation of hippocampal cellular "plasticity mechanisms (LTP; Moser et al. 1998).

2. Evidence from neurogenesis. There is evidence that neurons continue to be produced in the hippocampus throughout adult life (but see Rakic 2002). Training on learning tasks that depend on the hippocampus, but not on those that do not, was reported to enhance adult neurogenesis (Gould et al. 1999a).

3. Evidence from *functional neuroimaging. Despite some early failures to identify activation of the human hippocampus in "acquisition and "retrieval, many later functional neuroimaging studies of human subjects have delivered the goods (e.g. Lepage et al. 1998; Schacter and Wagner 1999; Haist et al. 2001). An imaginative example is provided by studies of navigation in virtual reality, in which successful performance was associated with activation of the right hippocampus in volunteers ranging from London taxi drivers (Maguire et al. 1997) to their clients (Maguire et al. 1998). In these studies, the left hippocampus was found active in nonspatial aspects of the task. Other functional neuroimaging studies of human subjects have used a variety of sensory and verbal tasks, and have identified hippocampal involvement in a number of declarative functions (e.g.Lepage et al. 1998;Dolan and Fletcher 1999; Schacter and Wagner 1999; Eldridge et al. 2000). Still, in some reports, the focus of activity in learning and memory tasks was identified in parahip-pocampal regions rather than the hippocampus proper (e.g. Brewer et al. 1998). Some neuroimaging studies also suggested differential involvement of anterior vs. posterior hippocampus (as well as other mediotemporal lobe structures) in acquisition (encoding) vs. retrieval, although the nature of the postulated functional segregation remains unclear (e.g. Dolan and Fletcher 1999).

4. Evidence from cellular physiology.

a. 'Place cells'. A breakthrough in understanding the relevance of hippocampal activity to behaviour was attained by O'Keefe and his coworkers (O'Keefe and Dostrovsky 1971; O'Keefe and Conway 1978). They have reported that the firing rate of certain hippocampal neurons is correlated with the animal's location in space. These cells were aptly termed 'place units', or 'place cells'. Their identification was used to support the hypothesis that the hippocampus forms and maintains spatial "maps, or, more generally, 'cognitive maps' of the world (on the concept of 'cognitive map', see Tolman 1948; on its history, see Best and White 1999). Over the years, place cells and their role in encoding experience-dependent spatial maps have become a central topic in hippocampal physiology, modelling, neurogenetics, and molecular neurobiology (e.g. Mehta et al. 1997; Wilson and Tonegawa 1997; Kentros et al. 1998; O'Keefe 1999). Electrodes mounted into rats hippocampal place cells even made it into outer space (Knierim et al. 2000). For our purpose, suffice it to say that in both laboratory and "real-life situations, location in allocentric space1 could be a major "dimension in world models encoded in the hippocampus (O'Keefe 1999). Physical space is not, however, the sole representational parameter of hippocampal units. Although place cells became quite popular from the outset of their identification, they were never claimed to exploit the response repertoire of hippocampal neurons (e.g. Ranck 1973). More recent studies have demonstrated that the activity of many hippocampal neurons is related to perceptual and behavioural events as well as their interactions, regardless of the location of where these events occurred (Wood et al. 1999).

b. "LTP. All the pathways in the trisynaptic circuit, the PP, MF, and SC (Figure 34), sustain LTP (Bliss and Collingridge 1993), a "synaptic plasticity mechanism that is assumed to subserve learning. If we indeed assume that LTP contributes to learning, then the hippocampus is surely equipped with the right cellular machinery.

5. Evidence from biochemistry and molecular biology. A number of studies have reported changes in hippocampal enzymes, growth factors, "neurotransmit-ters, and gene expression, that were correlated with learning and memory (e.g. Sunayashiki-Kusuzaki et al. 1993; Meyer et al. 1996; Cavallaro et al. 1997; Atkins et al. 1998; Hall et al.2000).

So what is the role of hippocampus in learning and memory? Despite thousands of man-years, smart paradigms, fascinating data, stimulating models (e.g. Treves and Rolls 1994), and the apparent neuroanatomical simplicity that has always attracted anatomists and physiologists, the truth is that we do not yet know for sure what the hippocampus does. One way of looking at the issue is within the framework of the expectation that there should be differentiation in "engrams along the following lines: some parts of the trace should reside in brain regions that deal with modality-, task- or content-specific information (e.g. spatial maps in the hippocampus—Gaffan 1998; see also "cerebral cortex). Other parts could depend on regions that subserve many types of engrams by performing 'global' operations (Hirsh 1974, 1980; Teyler and DiScenna 1986; Wallenstein et al. 1998; Holland and Bouton 1999). Judged by its connectivity to other brain areas, the hippocampus does fit to execute global operations. These could be of two types, which possibly overlap. First, computations necessary for the generation and processing of internal representations that are stored elsewhere. Examples are temporary "binding of multiple representations to promote the formation of new ones, or ad-hoc evaluation of the "contextual importance of on-line input or of retrieved information.

Second, operations that involve activation of highorder representations that are themselves stored, at least partially, in the hippocampus; these could involve representations of specific times and settings of episodes ("episodic memory), or of sets of other representations catalogued by yet unknown attributes. This latter view considers the hippocampus as some sort of a mental index that is used in the storage and retrieval of other representations.

With all this wealth of data, interpretations, and hypotheses, an attempt to encapsulate the proposed functions of the hippocampus in a catchy phrase, even if speculative, partial, and "metaphoric, is worth a try. So let's try this one. In that same moment in a mental time travel in which we "recall the specific contents, time, and setting of an episode all together, and suddenly experience that familiar sense of 'I remember' (Buckner 2000; Eldridge et al. 2000), it is the hippocam-pal system that is ticking in the background.

Selected associations: Amnesia, Consolidation, Declarative memory, Maze, Recognition

1'Allocentric' means independent of the location of the observer, as opposed to 'egocentric', in which spatial position is determined relative to the observer. An example of 'egocentric' units is provided by 'head-direction cells' in the "limbic cortex and subiculum, which fire as a function of the animal's head direction in the horizontal plane (Taube et al. 1990).

0 0

Post a comment