Memory for Public Events in Mild Cognitive Impairment and Alzheimer’s Disease: The Importance of Rehearsal

INTRODUCTION

Retrograde memory refers to memories from the past, as opposed to anterograde memory, which is the ability to learn new information. A unique and interesting way to assess retrograde memory is to use semantic memory tests that assess knowledge of public events. Indeed, memories of public events are unique in that they concern specific information and details associated with an event (semantic memory), in addition to being associated with a specific spatiotemporal context (the time and place the event occurred).

The two classical models on consolidation of long term memory, the Standard model [1, 2] and the Multiple trace theory [3–5], both assume that semantic retrograde memories are mediated by the medial temporal lobe (MTL) only for a limited period of time, after which they become increasingly reliant on neocortical structures; eventually, they can be retrieved without recruitment of the MTL. More specifically, the standard model of consolidation suggests that the MTL initially connects different neocortical sites in order to activate memory representations. As part of the normal brain activity, repeated co-activation gradually produces over time a long-lasting strengthening of the connectivity between neocortical regions. Eventually, memories thus become progressively independent from the MTL and rely exclusively on cortico-cortical connections [1, 2]. Similarly, the Multiple traces theory [3, 4] proposes that each new retrieval promotes re-encoding and produces the formation of multiple MTL-neocortical memory traces, with semantic information being mediated by the neocortex and unique autobiographical episodes being represented by the hippocampal complex [3]. Over time, remote memories are more often retrieved, become supported by more or stronger traces, thus acquiring more resistance to damage than recent memories. Thus, in the case of MTL damage, both models predict that remotely acquired semantic memories will be better preserved than more recent ones. Accordingly, in Alzheimer’s disease (AD) and in amnestic mild cognitive impairment (aMCI), which is considered by many to be the prodromal stage of AD in which pathology is essentially limited to the MTL [6–8], semantic memory impairment should be more pronounced for recently acquired than for remotely acquired information.

Deficits in semantic memory have been extensively documented in AD using tests of famous people [9–13] and public events [13–23]. Studies have also demonstrated semantic impairment in aMCI using tests requiring the identification of famous people and famous buildings [11, 24–29], as well as public events [15, 30–32]. However, the temporal gradient of these retrograde memory deficits has been much less investigated (i.e., does time passing promote better preservation of remote memories?), especially in aMCI, and the studies that have investigated the temporal gradient have produced mixed results.

The Remote Memory Battery [33] is a test combining naming of famous faces and a public events questionnaire that was developed to investigate the temporal gradient. Using this test, a Ribot-type temporal gradient (TG) was found in AD, with better memory for remote compared with recent periods [21, 34]. Similarly, the same Ribot TG was obtained in AD using tests of famous people [10] or public events only; however, this temporal gradient was very slight [16, 35], and probably due to items selection, since it was found in both AD and control groups [36]. To our knowledge, only two studies found a Ribot TG for memory about public events in aMCI patients specifically, and results were subtle. Indeed, in one study, this pattern was found only in half of the patients [37]. In the other study, extensive retrograde amnesia covering four decades before the diagnosis was reported, whereas memories were better preserved for the most remote periods, i.e., 60 to 40 years before the diagnosis [31]. In contrast, other studies found a TG in AD patients limited to recent periods using tests of famous people [13] or tasks combining famous people and public events [38], pointing to the anterograde amnesia as a plausible explanation for the results. Moreover, completely opposite patterns have also been found in AD when combining famous people and famous public events, whereby better performance was observed for recent rather than for remote periods [14]. Finally, no particular TG was found in a few other studies assessing memory for public events in AD and aMCI [15, 39]. Overall, although retrograde memory has been widely investigated in the past fifty years, studies have yielded mixed findings concerning the temporal extent of the retrograde amnesia in AD and aMCI.

Recent studies offer an alternative explanation that may be helpful in conceptualizing differently consolidation, as well as the processes underlying retrograde amnesia in AD. That is, rather than being consolidated over time, memories may be strengthened through rehearsal [15, 20]. Indeed, beneficial effects of repetition on consolidation processes have been found in some studies assessing memory for public events in healthy subjects. For example, Koppel and colleagues showed a positive relationship between the accuracy of memories concerning the emergency landing of US Airways Flight 1549 off the coast of Manhattan in 2009 and the frequency with which participants reported having thought or spoken, as well as watched, read, and listened to the media coverage about this event [40]. In another study, prior retrieval caused by repeated testing seemed to promote better recall for September 11th 2001 events [41].

In aMCI, Seidenberg and coworkers [28] found a better recall of semantic information concerning famous persons who became popular in the 1950s and were still well-known (enduring famous names) compared with those who rapidly went out of the public eye (transient famous names). In addition, no Ribot TG was found between recent and remote periods in regards to these famous names. Also, in a recent study, Müller and colleagues [20] specifically studied the effect of retrieval frequency on memory for public events in AD and aMCI patients. They found more detailed recall for contextual memories about events as well as more accurate dating of events and more vivid memories when retrieval frequency was judged as being high as opposed to medium or low. This favorable effect of retrieval frequency was observed across all time periods, yet a Ribot TG was also found in the recall of contextual memories. However, in these studies, aMCI and AD patients were pooled together, without examining performance of subgroups separately; therefore potential differences between patient groups are not known.

Thus, the general aim of this study was to compare the performance of aMCI patients with that of AD patients and healthy older controls (HC) on a test of memory for famous public events. More specifically, the aims of the present study were: i) to investigate semantic retrograde memory using a test of public events in order to clarify the pattern of performance across different time periods (temporal gradient); ii) to investigate the impact of repetition of the events. We hypothesized that none of the three groups would show a temporal gradient, but rather that repetition of events would have an impact on performance. More precisely, we predicted that repetition would favor a better preservation of memories for public events.

MATERIALS AND METHODS

RESULTS

DISCUSSION

In the present study, we investigated semantic knowledge about public events in aMCI and AD patients groups as a function of time period and nature of events (transient versus enduring). Specifically, we wanted to test the hypothesis that repetition rather than time course allowed a better preservation of semantic retrograde memory. In order to test this assumption, we used a test including questions about public events that occurred in the recent and distant past and that either quickly fell out of the public’s eye (transient events) or were repeatedly recalled by the media (enduring events).

First of all, our general finding that both aMCI and AD groups showed impaired memory for public events compared with HC, and that AD patients performed worse than MCI patients, is consistent with previous studies [13–24, 31]. Results of the present study suggest that the severity of the semantic memory impairment may be linked to the extent of temporal lobe damage, since the neurodegenerative process is thought to be more restricted to this area in aMCI than in early AD [62–66].

Secondly, we found that all three groups (HC, aMCI, and AD) showed a similar pattern of performance across the four time periods. Surprisingly, the three groups’ overall performance was lower for questions relating to the most remote time period. We found only one other study in which this pattern was found; this study used a test combining questions about public events and famous persons [14]. As in our own study, lower performance on remote items was also observed in HC, which leads us to believe that this may reflect inherent properties of the items. Indeed, despite the attempts made during the elaboration of the PUB-40 to have equivalent level of difficulty across all time periods, items for the 1960–1975 period may still be harder.

Thirdly, all three groups showed a better performance for enduring events than for transient ones, suggesting that frequently repeated memories are better preserved, regardless of the time period at which they were acquired. This finding is consistent with the beneficial effect of repetition demonstrated in other studies assessing semantic memory in young adults [40, 41] and aMCI [28]. This result is also similar to results reported by Müller et al. in aMCI and AD patients which showed that better performance was associated with greater self-reported retrieval frequency of historic events [20].

Furthermore, whereas AD patients were impaired on both transient and enduring events, aMCI patients showed preserved memory for frequently rehearsed public events (enduring), but impaired memory for transient events. Also, results of additional exploratory analyses suggest that aMCI patients may show a semantic memory loss that is temporally graded for transient public events that did not benefit from frequent rehearsal.

Finally, we carefully quantified participants’ media exposure to ensure that all groups had equivalent and substantial degree of media exposure to the stimuli in the PUB-40. Therefore, we are confident that the patterns of retrograde memory observed reflect true patterns of memory loss, rather than a lack of knowledge about these events.

Among the two main theories on long term consolidation, the Multiple traces theory (MTT) [4, 5] appears to be more in line with our findings, although not entirely. Indeed, according to this theory, frequently retrieved memories are supported by multiple memory traces and thus are more resistant to brain damage. This view is consistent with our global finding that all groups showed a better performance on frequently rehearsed enduring events compared with rarely repeated transient events. Also, the MTT considers that remote semantic memories are supported by more or stronger memory traces, and thus are more resistant to brain damage than recent ones. In the current study, Mann-Whitney U tests for each time period showed that memory for remote transient events was comparable in aMCI and controls. However, this effect is unlikely to be due to the fact that the most distant events were more likely to be rehearsed than recent ones, since the underlying selection criteria of transient events was that they quickly went out of the public eye, remained in the media for a very short period of time and were not later recalled, at least not on a large scale. Therefore, they had a very low probability of being repeated after their initial occurrence. Consequently, these results may be more difficult to explain according to the MTT. In sum, results of this study suggest that a slight temporal gradient may characterize aMCI’s retrograde semantic memory loss, but only for transient events. Therefore, the passage of time, rather than rehearsal per se, may play a distinct yet mild role on long-term consolidation processes. However, more studies will be needed in order to clarify the role of time passing on rarely rehearsed memories related to public event.

Furthermore, the MTT refers to a process of semantization in which semantic knowledge about events is separated from the episode where it has been acquired [3]. Indeed, in each new situation, common re-experienced information is extracted and integrated into pre-existing semantic knowledge mediated by the neocortex, whereas memory traces associated with unique autobiographical episodes are distributed in the hippocampal complex. Accordingly, memories about rarely repeated transient public events may not be semantized, relying on the hippocampal complex. On the other hand, enduring events benefit from semantization through frequent rehearsal and may thus gradually rely upon neocortical sites; at some point, knowledge becomes independent from the hippocampal complex. This is consistent with the recently proposed hypothesis that frequently retrieved memories become independent of the hippocampal complex [15, 67], as well as with our finding that aMCI patients, who are presumed to have damage more confined to the MTL [6–8, 65], showed impaired memory for transient events but spared memory for enduring events.

There are some limitations to our study that need to be mentioned. First, it would have been interesting to know when the memory impairment began for each patient and to explore whether performance varied as a function of when the public event occurred, prior or after the onset of memory deficits. This data would help to contextualize performance and clarify if results for the most recent time periods were reflecting retrograde amnesia, rather than anterograde amnesia or a combination of both retrograde and anterograde amnesia. But, it is very difficult to determine at which specific point in time the memory decline began, making this endeavor difficult. In the current study, however, the absence of correlation between patients’ performance on the PUB-40 and standard measures of anterograde memory, more specifically for the most recent time period, leads us to believe that the memory deficits observed for the most recent period is not associated with the anterograde memory impairment. Likewise, the lack of biomarkers for aMCI patients should be underlined. Indeed, we do not have biomarker evidence indicating that our aMCI patients are in fact in the prodromal stage of AD. Lastly, the status of enduring and transient events remains at least partly subjective, even if it was defined following pre-existing guidelines such as those proposed by Mayes and colleagues [61]. Also, especially for recent events, we cannot rule out the possibility that an event classified as transient may become enduring, for example if a movie based on this event is eventually made.

A better understanding of retrograde memory is of outstanding importance in the early detection of AD, since anterograde memory deficits that characterize aMCI cannot be used to distinguish patients who will develop AD from those who will not. Indeed, anterograde memory deficits can be found in treatable conditions such as depressive state [68] or nutritional deficiencies [69]. However, some recent studies suggest that semantic memory deficits could help to identify which aMCI patients will eventually progress to AD [12, 27]. Hence, sensitive retrograde memory tests assessing memories that did not benefit from rehearsal, such as transient events, may be useful in a clinical setting to target mild semantic deficits.

In summary, our study provides strong evidence that repetition allows better preservation of memories for past publics events, no matter the remoteness of these memories. This finding suggests that rehearsal may strengthen the memory trace more efficiently than the simple passage of time. This study challenges the widely accepted view that semantic retrograde memory is characterized by a Ribot-type temporal gradient, although future studies should focus on further understanding memory for transient events in aMCI. Moreover, our findings, based on the separate analysis of aMCI and AD patients, suggest that memories of transient events are first altered in aMCI, followed by memories of both transient and enduring events later in the disease. Our study, however, is cross-sectional in nature, and future longitudinal studies will allow the further elucidation of this question. Future studies using structural or functional neuroimaging methods should also investigate the neural correlates ofmemories for enduring and transient public events. In these studies, it would also be interesting to determine, for instance, if neural correlates of deficits for famous public events are more rightly-lateralized in aMCI and AD, as suggested by patients with the right temporal variant frontotemporal lobar degeneration who may be more particularly prone to being impaired on tests of unique entities [70, 71].