ارتباط بین حافظه کاری و اختلالات حافظه اپیزودیک در فراموشی گذرای جهانی

In a previous study, we investigated the relationship between the disorders of both episodic memory and working memory in the acute phase of transient global amnesia (TGA). Since executive functions were spared, another dysfunction may be responsible for the binding and maintenance of multimodal informations and contribute to the encoding disorders observed in some patients [Quinette, P., Guillery, B., Desgranges, B., de la Sayette, V., Viader, F., & Eustache, F. (2003). Working memory and executive functions in transient global amnesia. Brain, 126, 1917–1934.]. The aim of this present study was to assess the functions of binding and maintenance of multimodal information during TGA and explore their involvement in episodic memory disorders. We therefore conducted a more thorough investigation of working memory in 16 new patients during the acute phase of TGA using two tasks designed to assess the binding process and both dimensions of the maintenance, namely the active storage and the memory load ability. We also investigated the nature of the episodic memory impairment in distinguishing between the performance of patients with preferential encoding deficits and those of patients with preferential storage disorders on the episodic memory task. This distinction was closely related to the severity of amnesia, i.e. an encoding disorder was observed rather in the early phase of TGA. The results showed that while the functions of binding and maintenance of multimodal information were intact in patients with storage disorders, they were impaired in the case of encoding deficits. These results are interpreted in the recent framework of episodic buffer proposed by Baddeley [Baddeley, A. D. (2000). The episodic buffer: A new component of working memory? Trends in Cognitive Sciences, 4, 417–423] that represents an interface between working memory and episodic memory.

Transient global amnesia (TGA) is a clinical syndrome characterized by the abrupt onset of massive anterograde amnesia, usually accompanied by variable retrograde amnesia and repetitive questioning (Guillery et al., 2000; Hodges & Warlow, 1990; Quinette et al., 2006). As has been widely demonstrated by numerous neuropsychological studies, the memory impairment mainly concerns episodic memory, sparing other components such as procedural memory, priming effects and conceptual knowledge (Guillery et al., 2001 and Hodges, 1994; Kapur, Abbott, Footitt, & Millar, 1996; Kazui & Tanabe, 1995). Concerning working memory (Baddeley, 1986), conclusions are more mixed. Most authors have made only a cursory assessment of this system, using common forward digit and visuospatial span tasks, and have concluded that the phonological loop and the visuospatial sketchpad are preserved. Even when one or two authors have used more complex tasks, involving the manipulation of information and/or executive processes, the results have been far from conclusive. Regarding the shifting process, while Goldenberg (1995) demonstrated impairment using a modified version of the Wisconsin Card Sorting Test (Berg, 1948 and Goldenberg, 1995; Grant & Berg, 1948), Hodges (1994) failed to do so with the Trail Making Test (Reitan, 1958). Similarly, authors using the same version of the Stroop task (Stroop, 1935) have come up with contradictory results concerning inhibition (Hodges, 1994; Regard & Landis, 1984; Stillhard, Landis, Schiess, Regard, & Sialer, 1990). In a specially designed study, we investigated the main components of working memory during the acute phase of TGA (Quinette et al., 2003). We administered several working memory tasks to gauge the integrity of this system. Our assessment first focused on the central executive, the phonological loop and the visuospatial sketchpad, using the Brown-Peterson paradigm and both forward and backward digit and visuospatial span tasks. Second, in the light of Miyake et al.'s (2000) study, we tested the basic executive functions (inhibition, updating, shifting) using the Stroop test, the Running span task and the Trail Making Test, respectively. We also used a dual-task paradigm (Baddeley, Della Sala, Gray, Papagno, & Spinnler, 1997) to investigate the multi-tasking component, which is the primary major function attributed to the central executive and is thought to be independent of the others (Miyake et al., 2000). Except for pathological performances observed on the Brown-Peterson task for 3/3 patients during the episode, all the results were within the normal range of the age-matched control group. Thus, these results raised the question of the nature of the cognitive mechanisms involved in the Brown-Peterson task, which may depend not only on executive processes, but also on long-term memory. Moreover, because we made only an overall assessment of phonological loop abilities, we could not exclude the possibility that the pathological performances on the Brown-Peterson paradigm might also result from a deficit in short-term verbal storage. At the same time, we investigated the anterograde amnesia by means of an episodic memory task, derived from Grober and Buschke's test (1987), differentiating the nature of the memory impairment that may refer to preferential encoding, storage and retrieval processes (see Eustache et al., 1999, for details). The encoding phase of the episodic memory task involved the semantic processing of a word list. Its effectiveness was checked by immediate cued recall performed every two words, while retrieval was assessed by means of free recall and recognition tasks. In this manner, we were able to identify two patterns of impairment, affecting either initial encoding (impaired immediate cued recall) or the storage of episodic events in long-term memory (pathological performances on both free recall and recognition tasks) previously reported in other studies of our group (Eustache et al., 1999, Guillery et al., 2000 and Guillery et al., 2001). Taken together, our results suggested that the executive functions, i.e. inhibition, updating and shifting, did not contribute to the episodic disorder. Since an episodic memory deficit may result from an impairment of an associative binding of separate components into compound episodes as demonstrated notably by Naveh-Benjamin et al. (Naveh-Benjamin, 2000; Naveh-Benjamin, Hussain, Guez, & Bar-On, 2003; Naveh-Benjamin, Guez, & Shulman, 2004), we speculated that the encoding deficit observed despite a successful deep processing resulted from this binding deficit that may operate at the initial stage of episodic learning. Indeed episodic memory is comprised of collections of different features that are combined to form a coherent representation of an event. Consequently, creating new associations between the elementary features of an experience (e.g. colours, locations, time, objects, …) is critical for establishing episodic memories. This binding process may occur during the first stage of learning, namely during encoding (Johnson & Chalfonte, 1994). These authors have suggested that encoding, which determines storage efficiency, can be defined as the process by which information is maintained, manipulated and associated in working memory (Johnson, 1992). Some studies have assessed the process of feature binding in working memory in the context of ageing, schizophrenia or amnesia. Hence, to find out whether the deleterious effect of ageing on episodic memory reflects binding deficits during initial encoding, Mitchell, Johnson, Raye, Mather, and D’Esposito (2000) examined memory for combinations of features in a working memory task. Participants were shown pictures of common objects in a 3 × 3 grid. They were asked to remember either the objects, the location of the pictures, or a combination of the two (object + location) after an unfilled 8 s interval. The authors reported that the older adults had lower difficulty in the ones than in the combination condition. In a second experiment, they demonstrated that even when older adults were influenced by the memory load (i.e. when they were tested on 2 features rather than 1), they still had an age-related binding deficit. On the basis of these results, the authors concluded that age-related deficits in the processes that mediate feature binding were manifested in the working memory task and might be responsible for long-term episodic memory disorders, namely for problems in establishing associations between episodic information, i.e. time, place and the modality of acquisition, at both the encoding and retrieval stages (Chalfonte & Johnson, 1996). The same procedure was used in a more recent study of binding processes in patients with schizophrenia, using a working memory task (Burglen et al., 2004). The results of 25 patients were compared with those of 25 controls and showed that the processes which establish coherent and temporary episodic representations in working memory were impaired in schizophrenia. The authors hypothesized that the association of distinct features of an event needed in order to build a new multi-representation requires additional processes that go beyond the simple storage of individual features themselves. They suggested that the origin of the deficit in long-term memory observed in schizophrenia might be linked to difficulty in establishing associations between the different features of events. Baddeley (Baddeley, 2000; Baddeley & Wilson, 2002; Baddeley, 2003a and Baddeley, 2003b) proposed that working memory and episodic memory were related through an “episodic buffer”. Using this component, he explained various results which did not fit his “classic model”, notably correct immediate recall of prose in densely amnesic patients contrasting with poor performances on delayed recall (Baddeley & Wilson, 2002). These authors reported amnesic patients with an immediate memory span for sentences of about 15 or 16 words, while the equivalent span for unrelated words was typically about five items. Given that the difference could not result from a contribution of long-term memory in these amnesic patients, the authors postulated a further involvement of a binding process coordinated by working memory and particularly the episodic buffer, which would allow the correct recall of prose. According to Baddeley, the episodic buffer is not only a passive storage system but is also assumed to actively combine and maintain representations from different sources (i.e. other slave systems and long-term memory) in a novel episode by drawing on executive processes. Furthermore, this binding mechanism seems to be a highly attentional process dependent on both executive functions and fluid intelligence (Baddeley & Wilson, 2002). In their study, Baddeley and Wilson confirmed in amnesic patients the relationship between these three cognitive functions in showing positive correlations between prose recall and the performances at tests of both executive and intellectual functioning. Finally, on the basis of the Baddeley's definition and by analogy with the two other slave systems, two functions might be sustained by the episodic buffer, namely the binding and the maintenance processes, i.e. the active storage and the memory load abilities. While the studies detailed above provide arguments in favour of the relationship between an episodic memory disorder and a possible dysfunction of the binding process, the episodic buffer is currently the subject of debates. Gooding, Isaac, and Mayes (2005) observed in 19 densely amnesic patients more controversial results challenging the hypothesis of the episodic buffer. First, they demonstrated a considerable degree of variation in the prose recall of their amnesic patients, some of them having pathological scores. Second, unlike Baddeley and Wilson (2002), they found no significant positive correlation between the immediate prose recall scores and the scores on measures of executive function and fluid intelligence. These authors proposed that the different pattern of data from their amnesic patients might in part be explained by the heterogeneous aetiology of amnesia. They postulated that the location and extent of brain damage might vary and disrupt the binding and maintenance of prose information in different ways. This hypothesis was supported by neuroimaging findings arguing for the involvement of a frontal-hippocampal circuit in the process of feature binding (Mitchell, Johnson, Raye, & D’Esposito, 2000; Prabhakaran, Narayanan, Zhao, & Gabrieli, 2000). However, while these structures may work in parallel and be active simultaneously during the formation of a new episodic representation, each region may be involved in a specific process. The prefrontal cortex may underlie the binding and maintenance of information over a short interval, while the hippocampal system may consolidate these associated representations in long-term memory. In TGA, the immediate recall of short stories appears to be impaired (Beauregard, Weiner, Gold, & Chertkow, 1997; Eustache et al., 1997; Gallassi, Lorusso, & Stracciari, 1986; Hodges, 1994 and Kapur et al., 1996; Kritchevsky, Zouzounis, & Squire, 1997; Stillhard et al., 1990). However, this has only been demonstrated in a few patients and needs to be confirmed. Furthermore, even if these findings support the idea of impairment of a binding process, the nature of the dysfunction still has to be specified. Consequently, irrespective of the theoretical controversies about the episodic buffer, it appears to be relevant to investigate the binding and maintenance (active storage and memory load abilities) functions since they seem to be crucial to both encoding and retrieval processes in long-term episodic memory. However, given the neuropsychological and neuroanatomical variability related to TGA (Baron et al., 1994, Eustache et al., 1997 and Guillery et al., 2002), we would expect to find different patterns of results depending directly on the mechanism underlying the episodic impairment, i.e. preferential either encoding or storage (Quinette et al., 2003). Findings from our previous study on working memory during TGA led to two major conclusions. The first concerned the relevance of a further investigation of the slave systems and particularly the phonological loop to have a closer look at the hypothesis of storage deficit resulting in pathological performances in the Brown-Peterson paradigm. The second referred to the absence of a direct link between the massive episodic memory impairment and the preserved executive functions. However, we cannot exclude the involvement of another process of working memory such as binding processing in the episodic memory disorder. Consequently, the aim of the present study was to continue the exploration of working memory in new patients during the acute phase of TGA. We focused our investigation on both the slave systems and both binding and maintenance processes. First, we carried out a detailed investigation of the verbal and visuospatial storage abilities in assessing the phonological loop and the visuospatial sketchpad with a rigorous experimental methodology. The phonological store was investigated by means of tasks assessing the phonological similarity effect and the articulatory suppression effect (Belleville et al., 1994). At the same time, we used a visuospatial span task with tapping to test the visual cache of the visuospatial sketchpad. Second, we measured the ability of patients to establish and maintain temporary links between verbal and spatial information. In this manner, we expected to assess the binding and maintenance processes and find evidence in some patients of an impairment of this function related to their episodic memory deficits. While one task composed of two conditions was designed to investigate both the binding (with an 1 s interval) and the active storage (with an 8 s interval), another one was used to test the memory load abilities, i.e. the number of multimodal episodes that can be stored. The first task was derived from the studies by Mitchell, Johnson, Raye, and D’Esposito (2000), Mitchell, Johnson, Raye, Mather, et al. (2000) and Prabhakaran et al. (2000), while the last consisted of a multimodal span task, in which verbal information was combined with spatial information.