سوئیچینگ پویا بین سیستم های حافظه معنایی و اپیزودیک
|کد مقاله||سال انتشار||مقاله انگلیسی||ترجمه فارسی||تعداد کلمات|
|33633||2009||9 صفحه PDF||سفارش دهید||محاسبه نشده|
Publisher : Elsevier - Science Direct (الزویر - ساینس دایرکت)
Journal : Neuropsychologia, Volume 47, Issue 11, September 2009, Pages 2252–2260
It has been suggested that episodic and semantic long-term memory systems interact during retrieval. Here we examined the flexibility of memory retrieval in an associative task taxing memories of different strength, assumed to differentially engage episodic and semantic memory. Healthy volunteers were pre-trained on a set of 36 face–name pairs over a 6-week period. Another set of 36 items was shown only once during the same time period. About 3 months after the training period all items were presented in a randomly intermixed order in an event-related fMRI study of face–name memory. Once presented items differentially activated anterior cingulate cortex and a right prefrontal region that previously have been associated with episodic retrieval mode. High-familiar items were associated with stronger activation of posterior cortices and a left frontal region. These findings fit a model of memory retrieval by which early processes determine, on a trial-by-trial basis, if the task can be solved by the default semantic system. If not, there is a dynamic shift to cognitive control processes that guide retrieval from episodic memory.
Retrieval from long-term memory is likely to employ constant interaction between different memory systems (Poldrack et al., 2001) to ensure smooth progression towards successful recovery of stored information which is relevant to the current goals. The interaction between memory systems, as well as brain networks supporting them, is dynamically changing on basis of factors such as rehearsal of the information and level of experience with the task (Chein & Schneider, 2005; Poldrack, 2000). An everyday example of such a dynamic change is recalling the person-specific information, such as names, of persons whom we have previously encountered. Retrieving the names of familiar persons is a seemingly effortless process, whereas correctly naming persons met only once may involve effortful memory search to evoke the contextual details from the past encounter. Of course, familiar persons were once also unknown, which points to a fundamental learning-related re-organization of that information. One way of conceptualizing such a learning-related re-organization is in terms of episodic and semantic memory (Tulving, 2002). Episodic memory is concerned with encoding, storage, and retrieval of personally experienced events, whereas semantic memory refers to general knowledge about the surrounding world. Although still a matter of controversy, current theories seem to converge on the notion that repeated acquisition of the same information results in a change in how the episodic and semantic memory systems are involved (see Graham, Patterson, & Hodges, 1999; Moscovitch et al., 2005; Squire & Alvarez, 1995; Tulving, 2001). In the case of the above example, retrieving the name of a once met person would likely engage the episodic system, whereas retrieving the names of highly familiar persons can be based on the semantic memory system (i.e., there is no need to think back at a specific prior event). Support for this position comes from a cluster analysis (Dritschel, Williams, Baddeley, & Nimmo-Smith, 1992) which revealed a dissociation among retrieval of personal episodes (such as names of once met persons), personal semantic information (such as familiar names), and nonpersonal semantic information (such as names of presidents). Furthermore, a multivariate analysis of brain-imaging data (Nyberg, Forkstam, Petersson, Cabeza, & Ingvar, 2002) revealed that what could be conceived of as a test of personal semantic information (the autobiographical cued word-retrieval task) activated a similar left prefrontal network as factual retrieval, and was differentiated from activations associated with episodic memory (cued recall of experimental items). These findings suggest that one basis for a learning-related re-organization in case of associative recall, such as recalling someone's name, is that retrieval of well-learned associations engages the semantic memory system instead of the episodic memory system. Episodic memory has been found to engage a specific network of prefrontal regions, its most characteristic feature being activations in right inferior frontal and frontopolar cortices (Düzel et al., 1999; Habib, Nyberg, & Tulving, 2003; Nyberg, Cabeza, & Tulving, 1996; Tulving, Kapur, Craik, Moscovitch, & Houle, 1994). These brain regions have been suggested to support retrieval mode (REMO), a neurocognitive task set that is a necessary precondition for episodic retrieval ( Lepage, Ghaffar, Nyberg, & Tulving, 2000; Nyberg et al., 1995). REMO is hypothesized to be initiated by intention or instruction to retrieve information from one's personal past and maintained throughout the retrieval task, allowing stimuli to be treated as retrieval cues, suppressing irrelevant processing, and allowing successfully reactivated memory traces to be consciously attended to. By this definition, REMO activity should be present whenever episodic retrieval occurs, independently of retrieval success. Hence, to the extent that learning an association induces a shift from episodic to semantic memory, one would expect right prefrontal regions to be more engaged for relatively less familiar compared to more familiar items, regardless of whether the memory is actually retrieved. Conversely, strengthening of associations may induce increased activity in specific regions concerned with storage of that particular information during recall. Consistent with this notion, the recall of person-related semantic information (see Bruce & Young, 1986; Gobbini & Haxby, 2007) has been associated with lateral temporal (e.g., Elfgren et al., 2006; Gorno-Tempini & Price, 2001; Leveroni et al., 2000; Paller et al., 2003; Reinkemeier, Markowitsch, Rauch, & Kessler, 1997; Sergent & Signoret, 1992; Sugiura et al., 2006; Tsukiura, Mochizuki-Kawai, & Fujii, 2005), medial parietal (e.g., Elfgren et al., 2006, Shah et al., 2001 and Sugiura et al., 2006), as well as medial temporal regions, including more anterior sites in or near the hippocampus and more posterior sites in the fusiform gyrus (e.g., Elfgren et al., 2006 and George et al., 1999). The main purpose of the present functional magnetic resonance imaging (fMRI) study was to test the notion of a selective role of REMO-specific regions in retrieval of “episodic” associations by examining functional brain activity during memory for weak and strong face–name associations. During a 6-week pre-fMRI training period, the participants were familiarized with a set of initially unfamiliar face–name associations. Half of the stimuli were shown only once (1×) over the entire training period and the remaining half was shown six times (6×), thereby introducing two levels of strength for the face–name associations. During fMRI scanning, 1× and 6× items were presented in randomly intermixed order in an event-related design to test for dynamic switching between semantic and episodic memory systems. We assumed that right prefrontal activation would be elicited in the 1× condition but not in the 6× condition. The reduction of right prefrontal activity was predicted to be most salient for hits in the 6× condition, which would be strongly dependent on semantic memory system; whereas right prefrontal involvement in the 1× condition was expected to be unaffected by retrieval success (Nyberg et al., 1995 and Nyberg et al., 2000). A secondary purpose was to examine learning-related increases in brain activity, with a special focus on posterior association cortices, which would be consistent with increased involvement of semantic memory system with repeated learning.