Episodic Memory Network

Episodic memory depends on a widely distributed network of cortical and subcortical brain regions that overlap with and extend beyond the networks subserving other memory functions. Permanent representation of learned episodes seems to rely on distributing information into multiple neocortical areas and on the MTLs binding this information together (Squire and Zola, 1996). Normal MTL function during encoding, and for a period afterwards, appears essential for permanent declarative episodic memory formation. We will consider the contributions from the temporal and frontal lobes to episodic memory in more detail as they appear particularly important and altered activation in these areas have been demonstrated during episodic memory

processing in AD and high-risk states. The roles of the frontal and temporal lobes have been revealed by lesion and functional neuroimaging studies.

1.4.2.1 Lesion studies - the medial temporal lobes

The MTL is comprised of the hippocampal region (dentate gyrus, hippocampal cell fields C1-C3, subicular complex) and parahippocampal region (entorhinal cortex, Brodmann area (BA) 28,34; and perirhinal cortex BA 35) (Witter et al., 1989). During memory processing, information from sensory cortices pass to unimodal and

polymodal association areas in frontal, temporal and parietal lobes and then through the parahippocampal region to the hippocampal region. The hippocampal region in turn projects back to the cortex via the parahippocampal region. The entorhinal cortex in the hippocampal region receives the majority of these projections from the

parahippocampal region, via the dentate gyrus. This circuitry explains why episodic memory crucially depends on MTL structures as indicated by lesion studies. Medial temporal lesions results in dense amnesia for personal experiences whilst leaving semantic memory largely intact (Wheeler and McMillan, 2001). Selective lesions of MTL areas cause differential impairment on memory tasks and more extensive damage causes more substantial anterograde (formation of novel memory

representations) and retrograde (recollection of pre-lesion memory representations) amnesia (Squire and Zola, 1996). MTL structures therefore appear crucial in episodic memory function and this has lead to the large number of studies that have focused exclusively on this area, to the exclusion of other areas.

1.4.2.2 Lesion studies - the frontal lobes

Lesion studies in humans also demonstrated impaired episodic memory following frontal cortex damage. Such impairments are evident on tasks that are more complex and that involve high levels of interference and disruption of the temporal order of memory (Incisa della Rocchetta and Milner, 1993; Janowsky et al., 1989). The influence of such interference and disruption is not routinely tested although similar distracting circumstances occur frequently in everyday life in crowded and noisy surroundings. Frontal lesions are thought to cause failure in the control processes of memory rather than of automatic storage processes and we will look at this in detail in a later sections on attention and executive control (§1.6.1-3).

Lesion studies in animals have revealed contributions from other structures including the fornix, mammillo-thalamic tract and anterior thalamic nuclei to normal episodic memory function (reviewed by (Aggleton and Pearce, 2001)).

1.4.2.3 The functional neuroanatomy of episodic memory

Functional neuroimaging has been widely used to investigate memory processes in cognitively normal individuals and in those with memory impairment. Verbal

episodic encoding consistently activates left hemispheric prefrontal cortex (PFC) and temporal lobe structures, confirming the findings from lesion studies [for a review see (Cabeza and Nyberg, 2000)].

Left PFC activity during encoding relates to deep meaning-based semantic processing that optimises memory and involves the executive processes of generating, maintaining, selecting and organising semantically related information (Fletcher and Henson, 2001; Wagner et al., 1998). Indeed, disruption of left PFC function by transcranial magnetic stimulation results in impaired verbal encoding (Floel et al.,

2004), and disruption of semantic elaboration results in decreased left PFC activation and reduced memory performance (Fletcher et al., 1995; Grady et al., 1995). The extent of activation of the left frontal cortex and hippocampus during verbal encoding correlates positively with subsequent successful recognition (Morcom et al., 2003;

Wagner et al., 1998).

The PFC can be divided by the inferior frontal sulcus into dorsolateral (DLPFC) and ventrolateral PFC (VLPFC) with the former loosely corresponding to Brodmann areas (BA) 9 and 46, and the latter to BA 44,45 and 47 (Fletcher and Henson, 2001). Activation in the left DLPFC reflects the working memory processes of reorganising information and subvocal rehearsal of working memory content whereas activation in VLPFC relates to semantic processing during verbal encoding (for reviews see (Desgranges et al., 1998; Fletcher and Henson, 2001).

Activation in the MTL appears predominantly on the left during word encoding and left sided lateralisation of activation for non-verbal stimuli appears related to the verbalisability of the stimuli (Desgranges et al., 1998; Golby et al., 2001). Not all verbal encoding tasks activate the MTL and it appears that novelty and associative encoding (encoding pairs of items) are more likely to recruit MTL areas.

The effect of novelty and MTL activation is further illustrated by the finding that repeated exposure to initially novel face-name pairs results in deactivation in the hippocampus (Rand-Giovannetti et al., 2006; Sperling et al., 2003a; Sperling et al., 2003b).

Activation in MTL and PFC predicts subsequent recall performance and demonstrate the interaction between these two key areas in episodic encoding (Kirchhoff et al., 2000; Sperling et al., 2003a; Staresina and Davachi, 2008; Wagner et al., 1998).

Recent studies have also revealed correlations between memory performance and parietal and cerebellar activation (Brassen et al., 2006; Fliessbach et al., 2007;

Staresina and Davachi, 2006).

In summary, the PFC and MTL are key nodes in the network that underpins verbal episodic memory. This network also receives contributions from other cortical areas.

Next, we look at the neuropsychology of normal memory that depends on this functional network.