ON EPISTEMIC RESPONSIBILITY WHILE REMEMBERING THE PAST: THE CASE OF INDIVIDUAL AND HISTORICAL MEMORIES

The notion of epistemic responsibility applied tomemory has been in general examined in the framework of the responsibilities that a collective holds for past injustices,but it has never been the object of an analysis of its own. In this article, I endeavour to isolate and explore it in detail.To this end, I start by conceptualizing the epistemic responsibility applied to individualmemories. I conclude that an epistemically responsible individual rememberer is a vigilant agent who knowswhen to engage in different kinds of mental and nonmental actions in order tomonitor and update his or her memories, and who develops and nurtures different kinds of virtuous attitudes that guide those actions.These (epistemic) virtuousattitudesareorientednot only towardsoneself but also towardsothers.Although this conception of epistemic responsibility does not pose a problem for understanding sharedmemories of family members and friends, it may seem suspicious when applied to large-scale collective memories. These memories,which I namehistoricalmemories,arememories of events that have a traumatic impact for the community, are permeated by unequal relations of power, maintain a complex relationship with historical science, and present other characteristics that distinguish them from individual memories. But despite these differences, the analysis undertaken in this work shows that the general principles that govern theepistemic responsibility of individual and (large-scale) collective rememberers are similar, andarebasedon similar grounds:pragmatic considerations about the consequences ofmisremembering or forgetting and a feeling of care. The similarities on the individual and collective scale of the epistemically vigilant attitude that is and should be taken toward our significant past may partially justify the use of the same epithet—“memory”—to refer to these different kinds of representations.


V O L U M E 1 4 N U M É R O 2 A U T O M N E / F A L L
Some of the literature on ethical issues related to collective forms of memory has focused on the notion of "taking responsibility for the past" (see, for example, Blustein, 2008). The idea of taking responsibility for the past is a quite broad notion that looks into the ways in which a community should appropriately relate to its own history. Although discussions around this issue should in principle analyze the community's relationship with its past, whether or not there was wrongdoing, these kinds of discussions are in fact mostly focused on past injustices and thus also deal with ethical questions concerning collective-memory formations. As Blustein (2008) explains, "if the question of responsibility for past wrongs is going to be taken up collectively … the truth about the past … must become part of the group's collective self-understanding" (p. 138). That is why the question about the way in which a collective must be vigilant at the moment of recollecting its past enters into play. Collectives hold a certain epistemic responsibility if they aim to take concrete measures to repair past injustices.
The idea of taking responsibility for the past presupposes the idea of an epistemically responsible collective rememberer. Although it is true that these two responsibilities are not discrete stages in the process of coming to terms with the past (Blustein, 2008), they are conceptually distinguishable. Because the notion of an epistemically responsible collective rememberer has been generally studied inside this framework, an independent conceptual analysis that exclusively focuses on this concept may be beneficial to achieve a better understanding of it. What is more, because this notion concerns not only collectives but also individual forms of memory, the analysis of the way in which rememberers may be epistemically responsible for their personal memories is certainly a fruitful start to understand the collective counterpart.
Unlike Corlett's (2008) notion of epistemic responsibility, which is applied to the beliefs that a person voluntarily and autonomously forms, my interest here is to understand in general terms what it means to be an epistemically responsible rememberer. So, my analysis does not aim to present the necessary and sufficient conditions in which a rememberer is epistemically responsible for his or her memories, nor the different levels and kinds of epistemic responsibility that apply to different situations, but rather the general attitude that rememberers (as individual and collective epistemic agents) adopt-and should adopt-in their pursuit of accurate memories about their past experiences.
Many of the ideas developed in this article, especially those concerning collective forms of memory, are not necessarily new but are widely scattered in the literature. The aim here is to try to systematize these ideas in order to offer an account of what it means to be epistemically responsible rememberers of our own history. To do this, I examine the notion of epistemic responsibility applied to individual memories, in order to better grasp what is at stake with this concept and to consider whether this notion thus formulated is sufficient for understanding the epistemic responsibility of large-scale collective rememberers. Before entering into this last topic, which is also the main issue developed in this article, I analyze the notion of collective memory and propose to apply the term "historical memories" to a subset of collective memories that are produced by large-scale collective rememberers, then I present other specific and distinctive traits that differentiate them from family memories and other cases of smallscale collective memories. Once the notion of historical memory is defined and characterized, I come back to the main purpose of this article in order to see whether the notion of epistemic responsibility previously conceptualized for individual memories is also adequate for understanding cases of historical memories. Setting apart the differences between these two types of memories, I conclude that the epistemic responsibility that founds both of them is based on similar grounds: pragmatic considerations about the consequences of misremembering or forgetting and a feeling of care. Therefore, this article aims to shed light not only on the notion of epistemic responsibility applied to individual and collective memories, but also on conceptual distinctions in the field of memory studies that sometimes are quite blurred due to an absence of specific characterizations (as it happens, for example, with the concept of historical memory).

MEMORY, RECONSTRUCTION, AND EPISTEMICALLY RESPONSIBLE REMEMBERERS
At first sight, it may be natural to believe that memory is a capacity that to a large extent is out of our control: either we remember-that is, we manage to accurately represent a past experience-or we do not. The possibility of accurately representing past experiences may certainly imply a cognitive effort at encoding and at retrieval, except, manifestly, for cases of involuntary memory. I may pay extreme attention and repeat some information many times in order to be sure of correctly encoding it, and I may also mentally move along a chain of associated ideas in order to arrive at the one I am searching for. Although people can sometimes be blamed for not having made enough effort to encode some important information, people are not generally blamed for not making enough effort to retrieve it. This difference in blameworthiness is probably grounded in the belief that we have little control over our retrieval processes, so, if we cannot retrieve something we previously learnt or experienced, it is because the idea was never encoded-or in fact it was, but later "vanished" from our mind. In conclusion, although sometimes we can be blamed for not having paid enough attention in order to retain some important information, it may be quite natural to believe that rememberers do not really have control over memory processes after the encoding stage, so in general they are not accountable when forgetting or misremembering.
Because this common-sense idea assigns a very low (almost nonexistent) degree of epistemic responsibility to rememberers, I propose to call it the minimal version of epistemic responsibility applied to memory. According to the minimal version of epistemic responsibility applied to memory, rememberers are responsible only for making a cognitive effort toward retaining valuable information, and can be blamed only for not having made enough effort at the encoding stage, but do not have any other responsibility at the retrieval stage if they misremember or forget.

242
On the other hand, the minimal version of epistemic responsibility applied to memory is based on the conceptualization of accurate memory as retrieval of an idea previously encoded. This conceptualization obeys a preservationist conception of memory: if memory preserves past perceptions and experiences, and if these past perceptions and experiences were true at the moment of encoding them, memory then ensures that the recalled representations are also true. In fact, there are two variants of preservationism: the version just mentioned is focused on the preservation of perceptions and ideas (Locke, 1690;Hume, 1732), whereas the most recent version considers that what is preserved is the memory trace of the past event from which this past true representation can be constructed or inferred by the mind (Martin and Deutscher, 1966;Bernecker, 2010). In any case, both versions suppose that memory preserves content because it guarantees a sort of linear causal link between past representations and present ones. If memory preserves content, and also truth if the past perceptual representation was true, then the only effort we need to make in an act of recollection is to try to find the representation we are looking for. If we fail to do this, and thus we cannot remember it or we misremember it, the failure can be credited only to the malfunctioning of memory mechanisms over which we have no control, but not to the epistemic agents that we are. According to this conception, memory is involuntary and thus more similar to the muscles of the heart than to those of the hand (Margalit, 2002, p. 56). The minimal version of epistemic responsibility applied to memory and preservationism about memory seem to go hand in hand.
Nonetheless, the preservationist conceptualization of memory is a very questionable idea. First of all, distortions and delusions are not exclusively pathological phenomena that arise from brain disease: healthy people are very prone to intrusions, errors, and distortions that lead to the formation of memory illusions (Kopelman, 1987). Sometimes these memory errors and illusions can be as simple as misremembering where one put the keys, and other times they imply misremembering an entire experience. They can be created by endogenous factors, such as anomalous processing of input or a vivid imagination, but they can also be the result of exogenous influences such as social coercion or suggestibility (Kopelman, 1999). The variety and frequency of memory errors and illusions cast doubt on the conceptualization of memory as a capacity that fundamentally preserves past experiences.
False memories are thus very common. But the mechanisms and processes that generate everyday false memories are not dissimilar from those that produce faithful ones. And this leads us to the second point that thwarts the preservationist conception of memory and the possibility of directly accessing the past. Memory is always reconstructive: "reconstruction in general (and content generation in particular) is not the exception but the rule" (Michaelian, 2011, p. 328). Over the last decades, empirical research has provided strong evidence for the flexible and dynamic nature of our memory systems, which extract and recombine elements of different experiences and adapt the reconstruction of past event to the specific demands of the context of remembering. While a reproductive memory would not present much benefit from an evolutionary perspective, reconstructive memory is highly adaptive: it allows individuals to flexibly use past experiences in order to simulate future events and plan their actions accordingly (Schacter, 2012). Nonetheless, there is a price to pay for this high level of adaptability: the impossibility for our memory systems to keep exact replicas and records of past experiences. If memories are not reproductions but reconstructions from a network of distributed memory traces, there cannot be a linear connection between a memory representation and a past experience or event (Robins, 2016).
Consequently, if memory cannot guarantee direct access to the past, the reliability of memory as a source of information is brought into question. Some authors have proposed that we should work with a default assumption of trust in our memories (Sutton, 2003). If we do not do this, we are forced to deny the possibility of gaining any knowledge through this source (Michaelian, 2013). And it is true that the denial of this possibility does not sound reasonable: although there may be empirical evidence that shows that our memory systems produce a significant number of errors and distortions, in everyday life in general we manage to successfully navigate the physical and social world. So, it seems that in one way or another, the information from the past that we use for pragmatic purposes is quite accurate. This can be possible only if the trust in our memories is not a simple blind trust. If we succeed at constructing or making use of accurate information gained through past experiences, it is not because we simply trust our memory systems, but because we in fact exert different kinds of control processes over it. As Sperber et al. (2010) stated, "vigilance (unlike distrust) is not the opposite of trust; it is the opposite of blind trust" (p. 363). So, what we trust is the resulting information delivered by these unconscious and conscious control processes over our memory systems, not the information delivered by our "naked" retrieval processes themselves. And these control processes make the rememberer an epistemically vigilant agent: in fact, it is epistemic vigilance that counteracts the memory's inclination to distortion or error.
Thanks to this vigilant stance, memory is thus a sufficiently reliable source of information. Some of these specific control processes are unconscious and automatic, and are part of the process of remembering itself. Metacognitive mechanisms internal to the rememberer operate during the production of a representation of a past experience, in order to monitor and control the process itself, as well as the source of the remembered information. Whereas process monitoring allows the agent to determine whether he or she is remembering or engaging in some other form of construction, such as imagination, source monitoring evaluates different features of the remembered information (and its compatibility with other experiences and beliefs) in order to determine whether its source is a personal past experience or not (Michaelian, 2012). So, recollection is in fact a two-level process that involves not only the retrieval of stored information, but also metamemory processes that monitor that information and determine endorsement or rejection (ibid.). On the other hand, both internal processes of monitoring can also become conscious, reflective, and intentional, especially when the rememberer has some doubts about the reliability of his or her memory. Metacognition is not only implicit but can be explicit as well (Frith, 2012). Furthermore, the physical and social world is full of material traces left by past events, so it can act as a sort of external memory to which the rememberer can have access when he or she needs to (Sutton, 2010). If I am at the supermarket and am not so sure if I have to buy bananas, I can try to reconstruct the moment I opened the fridge to see which fruits and vegetables were lacking, but I can also take out my shopping list and simply check the items that I previously wrote down. Therefore, the rememberer has different mechanisms at his or her disposal in order to become an epistemically vigilant agent: the rememberer can monitor the formation of his or her memories and verify the accuracy of the remembered information. Some of these mechanisms are unconscious and automatic, whereas others are intentional and voluntary. Some refer to mental acts; others to external behaviour. But all these different kinds of mechanisms allow the rememberer to generally form accurate representations of the past, despite the reconstructive and fallible nature of our memory. In conclusion, memory does not enable a sort of immediate access to past truths, but rememberers have at their disposal sufficient cognitive and technological tools to faithfully reconstruct their past experiences in accordance with the pragmatic demands of the context of remembering.
To sum up, if memory preserves past experiences, epistemic vigilance with respect to memory is incorrect not only from a descriptive point of view, but also from a normative one: because people cannot exert any control over their memories-they are preserved or they are not-they cannot be compelled to do so either. But because empirical evidence shows that preservationism is wrong and memory is essentially reconstructive, epistemic vigilance is the rule. Rememberers not only execute different kinds of control processes in an act of recollection, but also must do this in order to guarantee memory accuracy (Michaelian, 2012).
This implies that rememberers are-to different degrees of blameworthiness and praiseworthiness-epistemically responsible not only at the encoding stage, but essentially at recollection. Compared to the minimal degree of epistemic responsibility that emerges from the preservationist conceptualization of memory, the version of epistemic responsibility implied by the reconstructive nature of memory is much stronger. But, at the same time, when memory is essentially reconstructive, this version of epistemic responsibility can also be considered quite standard. Because of this, I propose to call it the basic version of epistemic responsibility applied to memory. According to this version, rememberers must act as epistemic vigilant agents, which means that they are responsible for executing different kinds of actions related to processes of control, monitoring, and updating in order to construct accurate memory representations, and can be blamed for not having done them when they misremember or forget. This constraint does not mean that on each occasion rememberers must intentionally engage in multiple kinds of actions in order to monitor their memories: it would be cognitively costly to do so in situations where there is no reason to doubt, or when misremembering has no possible negative outcome. Epistemic vigilance is context sensitive: the particularities of the context of retrieval determine whether it is necessary to engage in monitoring and control processes and orientate the possible actions that could be performed for this purpose. 1 In fact, rememberers generally know when they need to engage in further monitoring and control processes because they already are epistemically vigilant rememberers. And although epistemic vigilance may be considered to be an epistemic virtue that can (and should) be cultivated through time, it is developed partially in a spontaneous way, 2 in order to avoid the negative consequences for practical life of having too many false memories. To paraphrase Sperber et al. (2010), if we broadly trust our representations of the past, it is because we are already epistemically vigilant agents. Normally, rememberers reflectively monitor their memory processes and sources and engage in different kinds of pragmatic actions in order to verify or change the remembered information because of the pragmatic benefits of having accurate memories: false and distorted memories lead usually to poor decisions and ineffective or detrimental actions. But sometimes other kinds of less practical reasons push rememberers to adopt this vigilant stance. These reasons concern attitudes related to oneself that are epistemically virtuous, such as self-criticism, sincerity with oneself, and avoidance of self-deception. They do not spontaneously develop, are hard to adopt, and may even present some disadvantages: although too many false and distorted personal memories are certainly maladaptive, some false memories can present more pragmatic and immediate benefits for the rememberer, especially for his or her psychological well-being (Bortolotti and Sullivan-Bissett, 2018). A rapist may unconsciously prefer to remember that his victim wanted to have sexual intercourse with him or that she explicitly provoked him, in order to avoid recognizing himself as the perpetrator of a hideous action. So, these epistemic virtues that are also at the origin of epistemic vigilance with respect to memory are not naturally developed by the pressure of taking good decisions and being effective while navigating the physical and social environment. They need to be cultivated and nurtured and thus require a continuous effort on one's own part.
Consequently, epistemic vigilance about memory is motivated not only by direct pragmatic benefits of forming accurate representations of the past, but also by epistemic virtues like sincerity with oneself. It does not seem possible to be a real vigilant rememberer without adopting and developing certain epistemic virtues that actively guide the process of reconstruction of past experiences. Consequently, the epistemic responsibility applied to memory implies more than what the basic version states. For rememberers to act as epistemically vigilant agents, they must not only execute different kinds of updating, monitoring, and control processes of their memories, but also develop and nurture certain kinds of epistemically virtuous attitudes related to themselves, such as self-criticism, sincerity with oneself, and avoidance of self-deception. These epistemically virtuous attitudes orientate the kinds of mental and nonmental actions that must be performed in order to be epistemically vigilant while reconstructing the personal past. So, epistemic vigilance with respect to memory is not only context sensitive but also virtue dependent: it depends on the possession of certain epistemic virtues for its good performance.
It remains to be seen whether these epistemic virtues are sufficient for being a full vigilant rememberer. To explore this issue, I propose that we consider memories that are important not only for our own identity but also for interpersonal relationships, for the common history and identity we shape with others. Most of these important memories cannot be reduced to a fact that can be easily verified through internal monitoring processes and acts of verification performed in the outer world. They present an evaluative component because they refer to more than the material world: these memories involve other people, their actions, and the impact and meaning that other people and their actions have in our life. At least two different kinds of evaluations shape the way in which we remember past events: ethical and affective ones. The ethical component refers to evaluations concerning what is right and what is wrong. An example would be when I remember that my boss continuously takes advantage of me. This memory can also present an affective component if, for example, I feel simultaneously angry. Affective evaluations refer to the personal significance and meaning that an event has for the subject in terms of harms and benefits. Some of these evaluations are so bound up with the way we remember past events that it becomes impossible to separate our evaluative perspective from a state of affairs that would be objectively verifiable. Therefore, when individuals have different evaluative perspectives about a shared past event, it is not so easy to determine what a rememberer should do in the presence of memories divergent from his or her own. Simply ignoring others' memories does not seem to be a responsible action from a conscientious rememberer. This is for two reasons.
The first reason concerns the way in which the evaluative aspects of our memories are constructed, and the consequences that these memories have. Although we may experience the past individually, its affective and moral significance is in general determined with others, because we are relational beings who constantly "negotiate the past-both the personal past and the shared past-in company" (Sutton, 2006 p. 126): that is, through relations with other people and other narratives. What is more, many of these meaningful memories are not simply private and passive representations without any causal effect in the world: by reinforcing or resisting those of others, these memories contribute to other people's grasp of the meaning of a shared past, and even to the construction of their identity (Campbell 2006(Campbell , 2014. This is particularly remarkable in the cases of thick relationships-that is, of relationships with people with whom we have strong ties, which are essentially built on care (such as friendship, partnership, family ties, etc.). Memory is clearly constitutive of the notion of care: "If I care for someone or for something, and then I forget that person or that thing, this means that I have stopped caring for him or it" (Margalit, 2002, p. 28). We stop caring about someone with whom we have a thick relationship not only when we forget him or her, as Margalit (2002) has emphasized, but also when we fail to understand his or her perspective of a common and shared experience. For example, my husband can remember that during the fight we had last night I yelled and treated him with disdain, but I might not remember the fight that way-that is, I remember that we had a normal and peaceful argument. Because I care about my husband, it seems responsible to take into consideration his perspective of the fight while constructing a common narrative of it, and to even rethink and reshape my own memory. If I refuse to consider his perspective, especially if I frequently adopt this attitude, it means that I do not take his feelings and evaluations of what happened seriously. If I disdain his words and thoughts as not worth it, I fail to care about his feelings and points of view, and this means that I fail to care about him tout court. When we care about other people, we care also about the way we shape our own relationship together and construct our common reality. It is in fact this care, then, that generally pushes us to adopt certain kinds of attitudes towards the way we remember common experiences: to be openminded and responsive to others' memories, to the extent of being willing to remember the past differently.
Consequently, the construction of our memories cannot simply be motivated by values and virtues such as the search of accuracy and truthfulness to oneself: it must also imply being true to others. Campbell (2006Campbell ( , 2014 uses the notion of integrity to refer to this implicit accountability to others for the way in which we remember our common past: "The integrity with which we remember has to do both with how we understand our own past in ways that contribute to self-knowledge, identity, and the shape of personal responsibilities and possibilities, and also with whether others can rely on our memories not only for what they do not know but also as a contribution to a social grasp of the significance of a shared past" (Campbell, 2006, p. 374). This notion of integrity introduces the second point I wanted to make against the disregard of other people's memories about a common past. According to Campbell, integrity is an epistemic virtue closely linked to accuracy in the reconstruction of faithful memories. But the description presented above is somehow ambiguous. Integrity appears more as a moral virtue than as an epistemic one: a failure of integrity related to our memories is a failure of care, not in principle a failure of accuracy. Nonetheless, the notion of integrity does not only refer to having the right sort of relation to ourselves as well as other people with whom we have thick relationships, and their narratives. It presents a simpler meaning: the wholeness and completeness of something-in our case, memory. A memory of a shared past does not seem really complete without the integration of the different evaluative perspectives of that past made by the different participants. While talking about last night's fight, I consider the perspective of my husband not only because I care about my husband but also because I might care about getting a complete picture of what happened last night, including the subjective feelings and moral appreciations of my husband. If I refuse to consider his perspective, especially if I frequently adopt this attitude, I fail to recognize not only the evaluative components of our memories, but also the existence itself of different meanings, subjective feelings, and appreciations that can be directed towards an event. What is more, without the perspective of the other, I may also fail to be sincere with myself and to avoid self-deception: it may be the open-mindedness towards others that allows those self-focused virtuous epistemic attitudes to rise and flourish. Personal memories of shared and common past experiences thus gain in accuracy when they are integral. Therefore, integrity related to memory is not only a moral virtue that is valuable in itself for developing and maintaining good relationships of care. According to this second sense, integrity is also essential for accuracy, so it is at the same time an epistemic virtue. Although it is probably the care for others that is the major force that moves us to take into account their perspectives of a shared past, this care leads at the same time to the construction of more accurate memories.
If integrity is a constitutive part of memory accuracy-at least for memories of shared and common experiences-an epistemically vigilant rememberer is an agent who adopts an epistemically vigilant stance not only towards himself or herself, but also towards others. Consequently, the basic version of epistemic responsibility is not really sufficient; a reformulation of the notion of epistemic responsibility is thus needed. This version is clearly more demanding than the previous one; that is why I propose to call it the strong version of epistemic responsibility applied to memory. According to this strong version, epistemically responsible rememberers not only must engage in different kinds of actions in order to control, monitor, and update their memory processes, and be guided by certain kinds of self-focused epistemically virtuous attitudes, but must also develop and nurture other-focused virtuous attitudes such as being open-minded and responsive to other people's perspectives when it is a shared and common past that is remembered. So, in this strong version of the epistemic responsibility applied to memory, the moral responsibility is explicitly present. Care leads us to consider other perspectives in the construction of memories and to develop other-focused attitudes that are morally good and epistemically good: they nurture the care we owe to people with whom we have thick relationships, while at the same time enhancing the accuracy of our memories of shared experiences. Although the moral responsibility we have as rememberers is conceptually distinct from the epistemic responsibility, both of them converge in the construction of our significant memories. It seems difficult to conceive of a fully vigilant rememberer who is not simultaneously a virtuous rememberer. Epistemic vigilance with respect to memory is virtue dependent not only because it is contingent on epistemic virtues, but also because it is contingent on moral virtues. It seems thus that epistemic vigilance partially depends on the development of a virtuous character (Audi, 1991). 3 The notion of integrity certainly enhances and enriches the conception of epistemic responsibility applied to memory. However, this later becomes more complex and difficult to grasp when the kinds of memories that are at stake are not individual memories or memories of groups whose members have thick relationships with one another, but memories of large-scale collectives such as nations.
It is thus necessary to better examine the nature of these collective memoriesand the nature of collective memory in general-in order to conceptualize a suitable notion of epistemic responsibility applied to these cases.

THE NOTION OF COLLECTIVE MEMORY
It is a commonplace in the literature to highlight the vagueness and uncertainties that surround the concept of "collective memory," which was first introduced by Halbwachs (1925). As Wertsch and Roediger (2008) have stated, "collective memory is a term that is widely used, yet poorly understood in contemporary academic discourse" (p. 318). While it is difficult to deny the fact that we each have our own memories about personal past experiences, it is more difficult to accept that a certain human group has a common memory of an experience that it considers its own.
The first problem that arises is how we should conceive the nature of this shared memory and of the collective that remembers. Some authors consider that although social resources and practices shape our memories, only people can memorize and remember and not collectives (Kansteiner, 2002). Collective memory would be "shared individual memories that shape collective identity" (Hirst and Manier, 2013, p. 196). This individualist conception of collective memory is also related to the tendency to analyze collective memory in the same terms as individual memories, as if the collective were an individual (Ricoeur, 2006). 4 For other authors, there is a distinction to be made between the aggregated individual memories of members of a group and real collective memories. While the first case can be considered to be only "collected memories" (Olick, 1999) or "common memories" (Margalit, 2002), the concept of collective memories refers to public symbolic representations of a common past that are situated "in the world" and have a certain independence from individuals. But this holistic conception of collective memory is also far for being consensual. Just like individual memories, which are not representations of the personal past fixed once and for all, but which often change according to the particular context of retrieval, collective memories are not a static body of knowledge shared by a group of individuals and materialized in their culture through different social technologies. Wertsch and Roediger (2008), for example, believe that the essence of collective memory would be better characterized by the term "collective remembering," which reveals the struggles, contestations, and negotiations that characterize the accounts of a common past. This emphasis on the idea of collective memory as a complex and dynamic process that extends through time also presents some similarities with the more recent conception of individual memory not as a mental state but as a mental act.
There is also a second problem stirred up by the concept of collective memory, which is related to the relationship between the collective and the experience that this collective remembers. Michaelian and Sutton (2017) hold that a truly collective memory would require that individuals interact at the moment of encoding a common experience and also at the moment of remembering it. This strong conception of the notion of collective memory may lead us to believe that there is a single and quite unitary collective memory representation that is held by the group. 5 Nonetheless, this notion could in principle be applicable only to small-scale groups: members of large-scale groups that are affected by the same event necessarily have divergent experiences of that general event and thus remember it differently. On one hand, people who experience the same general event, such as a war or an earthquake, do not experience the same particular events that are subsumed under the general one, such as a specific battle or a specific building collapse. On the other hand, even people who experience the same event do not necessarily experience and remember it in the same way, because of the particular evaluative components that shape the way in which each individual perceives and remembers a specific state of affairs. For example, during a particular act of torture, the tortured will certainly not have the same experience and thus memory of it as the torturer. On the other hand, some individuals of a collective who actively remember a historical past episode that continues to shape their identity did not even directly experience the episode in question. In fact, Assmann (1995) has proposed that the most significant collective memories that shape the cultural identity of a community go beyond the individuals who actually experienced the event remembered, and become accessible across the centuries. 6 That is why although members of a community may not exactly remember the same specific event or experience, the collective memory about the general event that affected the whole community "belongs to them" (Blustein, 2008, p. 183), in the same way that some significant personal memories are intrinsically tied to the identity of an individual. Furthermore, as Blustein (2008) has explained, to reduce the notion of collective memory to collective beliefs or other forms of discursive production implies forgetting the variety of other forms of remembering that involve memory activities and memory places. 7 The problematic nature of collective memory and the wide variety of conceptualizations and uses of this notion are manifest and well known. I do not intend here either to deepen this debate or propose a final solution to this conceptual problem. 8 Starting from three premises-(a) that Michaelian and Sutton's (2017) notion of a truly collective memory is quite reductive and so establishes an essential difference between small-scale group memories and large-scale representations of a common and shared past; (b) that collective memory seems to involve different kind of entities (not only individuals) and practices, and is thus better conceived as a complex and dynamic process that extends through time; and (c) that it plays a key role for the internal cohesion and the identity of the collective such as individual memory does for the individual (I will say more about collective memory functions in the next section)-here I propose to use a reformulation of the "epidemiological" approach to collective memory outlined by Hirst and Manier (2008) as a working definition of "collective memory." This approach focuses on how a memory of a shared past spreads across a collective, as if it were an epidemic disease. I thus conceive collective memory as a set of complex and dynamic processes of interaction between societal memory practices, mnemonic technologies, cultural tools, institutions, and individuals, through which (not necessarily discursive) representations of a common past are constructed, reconstructed, performed, and also sometimes deconstructed or crystalized and transmitted to subsequent generations. These representations of a common and shared past refer to representations of events that were experienced in some way by the collective or by a significant proportion of its members, or by the collective's ancestors or the ancestors of a significant proportion of its members.
It is clear that interaction does not always have to take place among all these different entities; nor has it to involve a variety of practices. This is typically the case for small-scale groups, like groups of friends, who remember a significant event from the past, such as a trip. Whereas the collective memory can be constructed or reconstructed through the dynamic interaction among friends who talk together while looking at photographs of the trip, it is unlikely that institutions and more complex societal practices and technologies intervene in this process. Nonetheless, collective memories of large-scale groups, such as communities, countries, and regions, in the long run involve all the elements and entities previously mentioned. These kinds of memories endure longer than those of the small groups, and usually refer to events that are or could be studied by historians. I thus reserve the term "historical memory" for these last cases, which I conceive as a kind of collective memory.
The strong version of epistemic responsibility is certainly adequate for the understanding of memories of small-scale groups, because a relationship of care exists among their members, as is the case in families and group of friends, or because it exists not necessarily among their members, but towards the preservation and proper functioning of the group itself, as is the case in groups of business partners or coworkers. Therefore, it is possible to say that small-scale group rememberers are responsible for engaging in different kinds of actions in order to control, monitor, and update their collective memory representations, but are also responsible for having certain kinds of attitudes, such as being self-critical, open-minded, and responsive as a group to all members' perspectives on the shared past. Many of the significant memories for group identity refer more to the meaning of an event, to its evaluation, than to its physical properties. That is why if only one perspective is imposed on the group, the memory is not integral and thus not really accurate.
Nevertheless, this strong version of epistemic responsibility applied to memory does not seem adequate for understanding the case of large-scale collectives who remember a common and shared past. Large-scale collective memories are more difficult to grasp than small-scale ones. First, large-scale collectives do not have clear and established limits, so it is not always easy to trace a line between those who are part of the collective and those who are not. What is more, their members may change through time: the current members of the collective may not have directly experienced the remembered event, and all those who experienced it may already be dead. On the other hand, members of the collective occupy different positions in the social reality at the moment of experiencing the event and at the moment of reconstructing it as well. That is why large-scale human groups are crisscrossed by different relationships that individuals establish among them, which in some cases can be very lopsided in terms of power.
Because of the complexity and heterogeneity of large-scale collectives, and the thin or almost nonexistent ties among their members, it is not easy to conceive that they could have epistemic responsibilities as rememberers in the strong sense presented above. If relationships among their members are not based on mutual care or care for the group, there is no reason to be open-minded and responsive to other perspectives of the past. If nonetheless there is an epistemic responsibility towards other people when recollecting the shared and common past, it may be grounded in something other than care. On the other hand, as rememberers of historical events, we can simply contrast our representations of the past with material traces and history in order to form accurate representations. So maybe a basic version of epistemic responsibility applies better to collective representations of a shared and common past. Nonetheless, given the particularities of large-scale collectives, the relationship between history as an academic discipline and memories of historical events is neither simple nor unidirectional.
In the next section, I develop a characterization of the particularities of historical memory and exemplify it with concrete historical cases in order to better understand its nature and conceptualize it; then, in the last section, I develop a notion of epistemic responsibility applicable to historical memories.

HISTORICAL MEMORY: ITS PARTICULARITIES
The term "historical memory" is used in the literature, but not defined with precision. In fact, it is sometimes used by members of a collective to express a demand for justice, and sometimes by the state itself to compensate for past crimes. 9 The association of the notion of historical memory with memory of distant events from present times is quite common and goes back to Halbwachs (1950, especially chapter 2), for whom there exists a distinction-and opposition-between historical memory and collective memory. Nonetheless, this characterization is confused: he even admits that the expression "historical memory" is inadequate. His main interest is actually to distinguish collective memory from history. Whereas collective memory is related to the experiences of a group, history introduces conceptualizations and temporal distinctions that are absent from collective memory representations. For Halbwachs, as the temporal distance from the past event increases, the event ceases to be the object of the collective memory and becomes part of the historical memory (Barash, 2016). But temporal distance and opposition to history cannot be considered criteria for defining historical memory, since contemporary history does not deal only with the distant past (Koselleck, 2002). In fact, history as an academic discipline coexists with representations of significant past events shared by a largescale collective. I thus propose understanding historical memory as a kind of collective memory that presents certain characteristics that distinguish it from other forms of collective memories.
First, the objects of historical memories refer to events of great importance for social cohesion, because of their traumatic impact for the collective and their consequences in the present.
Second, as I just mentioned, historical memory involves large-scale collectives; that is why it is characterized by a marked heterogeneity.
Thirdly, because of this multiplicity and heterogeneity, historical memory is essentially pierced and permeated by substantial relations of power that establish unbalanced relationships among the members of the collective.
Fourthly, historical memory accomplishes different functions for the collective that are not truth oriented, and this makes its relation with truth problematic.
And finally, historical memory maintains a dynamic but complex relationship with history, which sometimes is harmonious, and at other times is tense and conflictive.
In the following subsections, I explain each of these characteristics in more detail.

The objects of historical memory
Historical memory seems to be focused on events that are part of the identity of a collective because of their traumatic impact and their emotionally charged nature. Whereas individual personal memories and significant memories of small-scale collectives can have a negative but also a positive valence (the memory of the wonderful first year of life on campus can be at the origin of the cohesion of a group of friends), historical memories are in general traumatic in nature. Positive past events persist in the present in the form of rituals, anniversaries, and festivals, all of which have in general become quite detached from the past event in question and have acquired a new meaning for the collective: a gathering moment or a nonworking day (such as it happens in some countries' commemoration of their independence day). In certain ways, memories of past positive events have metamorphosed into a sort of collective procedural memory, a traditional internalized practice that has lost its attachment to its origins and is not performed for commemorative purposes any more. That is why murders, disappearances, massacres, deportations, imprisonment in inhumane conditions, torture, repression, criminal negligence, and the like are commonly the object of historical memories. This characteristic establishes an important distinction between the standard or normal type of individual and of small-scale collective memories and the standard or normal type of historical memory.

V O L U M E 1 4 N U M É R O 2 A U T O M N E / F A L L
On the other hand, although historical memory interacts with historical knowledge, historical memories are not similar to the historical knowledge that a collective may have, such as "The Argentinian caudillo Facundo Quiroga was murdered in Barranca Yaco." This difference, besides the fact that historical memories are not reducible to simple beliefs but include memory activities and memory places, is similar to the difference, introduced in the literature on individual memory, between episodic and semantic memories (Tulving, 1972). Whereas the objects of historical memories are events experienced by the collective or by a significant proportion of its members, or by the collective's ancestors or by the ancestors of a significant proportion of its members, historical knowledge refers to historical facts, endorsed by academic history, that through education have also become part of the collective background knowledge. 10 As in the case of the individual memory, the difference is useful but not always neat, because of the complex interactions that exist between the two types of memories. In any case, I will come back to this topic in the section about history and historical memory.
Finally, historical memories refer to traumatic events that still have negative consequences in the present for the collective or for a part of it, even if the causal connection is extended through centuries, as it happens with colonization for the First Peoples. In some cases, the consequences of the event have already disappeared and material and symbolic reparations have been granted, but the event is still remembered because it represents or symbolizes similar injustices to present or more recent ones.

Heterogeneity of memories of large-scale collectives
As a result of the conjunction of these characteristics-large-scale collectives who remember an emotionally charged event from their past that still has a negative impact in the present-historical memory is much more heterogeneous and complex than other forms of collective memory. The use of the singular term should not hide the fact that it is a complex and often conflictual social construction, whose tensions are not always solved and may resurface under certain circumstances. That is why in many cases it seems more appropriate to talk about historical memories in the plural. Disputes between different historical memory constructions of the same event are quite common. These disputes can sometimes even present the characteristics of a war, whose trophy is the epithet "the true representation of the past." This is the case for the memory of the Civitella Val di Chiana Massacre in 1944 in Italy, when Nazi soldiers murdered 115 men as retaliation for a partisan attack. While the victims' relatives remember them as martyrs of a violence that they did not cause and make the partisans responsible for the massacre, the partisans vindicate the attack as a heroic act (Portelli, 2013).
In some other cases, a specific historical memory is established as if it were the only one available, by silencing other possible collective representations of the past and fixing a unique meaning to the event remembered. This is what some-

V O L U M E 1 4 N U M É R O 2 A U T O M N E / F A L L
times happens with the "official" memory: the memory held by the state-which is also a form of historical memory and uses the school apparatus as powerful instrument of transmission-might also conflict with more local memories held by different members of the community. For example, the armed struggle to establish the Argentinean national state by advancing on territories first inhabited by Indigenous peoples was named "Conquest of the Desert" in classical historiography and school manuals. Through formal education, this notion became a popular term used by Argentinians to refer to that period of time. However, the descendants of those first inhabitants have questioned the idea of a "desert," because it makes reference to an empty space when in fact that territory was inhabited by multiple tribes. Nowadays, the notion of a genocide of the Indigenous populations has come to the fore (Delrio et al., 2010).
It is common that those who occupy subaltern positions in the social reality, and are thus in a poor position to directly influence the socially accepted memory of a shared past, build a collective memory that strengthens their own identity against the hegemonic or official one (Castells, 2009). But other collectives that are not necessarily in subaltern positions also privilege the social cohesion generated by the historical memory and avoid in this way facing the challenges that may arise when confronted with its factual corroboration. An interesting example to illustrate this point is the case of the common "error" in the historical memory of the Fosse Ardeatine Massacre that was carried out in Rome in 1944, when German soldiers killed 335 people in retaliation for a partisan attack. While the German reaction was immediate, all survivors stated that several days, even months, passed between one event and another (Portelli, 2002). Portelli considers these massive fallacious memories to be unintentional, the result of the collective meaning that survivors gave to what happened: that the partisans could have avoided retaliation if they had surrendered.
The large-scale nature that characterizes historical memory diversifies in huge proportions not only the experiences that the members of the collective have of the same historical event, but also the events experienced. The degree of involvement during the event can be also extremely broad and includes simple witnesses who passively suffered as a consequence of a past event, among a wide variety of intermediate situations (LaCapra, 2016). For example, an economic crisis that causes the ruin of a family will be experienced differently by the family nucleus in comparison with the group of distant relatives. The massive difference in the degree of involvement is also present during the construction and persistence of a historical memory, which are not automatic processes but depend on the effective articulation of several factors, such as particular actors, symbolic and material resources, and practices. A particular historical memory is usually nurtured and maintained by those members of the collective who occupy prominent positions in the community, such as the governing body. But its persistence also requires the participation of other actors who actively support and promote it over time from their own specific social spheres of influence (economic, political, or cultural spheres). These actors refer to what Jelín (2002) has called "memory entrepreneurs," such as the famous Auschwitz survivor Primo Levi (Levi, 1995). At the other extreme of the spectrum of involvement and compromise, there are members of the collective who are less committed or who are even completely passive. Once a collective representation of the shared past becomes more or less stable during a certain amount of time, it is usually structured in highly ritualized forms such as ceremonies, institutions, and social practices. These rituals fix the meaning of the remembered event and, at the same time, free individuals to autonomously consent to the truthfulness of the historical memory. That is why some members of the collective limit themselves to tacitly accepting these different collective rituals without either being part of its design or worrying about its transmission or possible expansion. Finally, historical memories usually transcend the limits imposed by the life of their original bearers. Therefore, the collective who remembers an important episode of its common past can include new generations who have only indirect knowledge about the event because they have not directly experienced it.

Relations of power
Relations of power are at the centre of historical memories, and this establishes an important difference with individual memory and other kinds of small-scale collectives. Individuals sometimes struggle with themselves, their self-image, their desires, or their beliefs at the time of reconstructing personal past experiences. But cases of self-deception, where the individual seems to be split into one self who persuades other self of some biased interpretation of the personal past, are not really analogous to the relations of power that are intrinsic to the nature of historical memory. 11 Power refers to the ability of a social agent to influence through his or her actions a particular state of affairs, which can also include the actions of others. This capacity is directly related to the positions that those agents occupy in a particular social context. That is why, in general, power establishes asymmetric relations between different actors, depending on the degree of influence that each one can exercise (Giddens, 1984). Concerning historical memory, we can distinguish different kinds of power relations that permeate it. First, relations of power are present at the moments of construction and consolidation of a specific historical memory. Because, in general, events experienced and experiences are thoroughly different within a large-scale collective, it is expected then that tensions and conflict among the members of the collective will be common during the negotiations over the meaning of that past event, and that the outcome of the negotiations will depend on relations of power. Second, relations of power are also present in the disputes over the "right" interpretation of the shared past. This happens when there are different collectives (including the state) that have divergent visions about what happened. Moreover, relations of power exist when dissident voices are silenced immediately after the event through a sort of ad hominem fallacy whereby the credibility and the identity of their bearers are implicitly or explicitly cast into doubt or harmed. These power relations are also present when dissident voices cannot even be raised due to hermeneutical marginalization-that is, when members of a collective lack interpretative and conceptual resources to understand their own experience (Fricker, 2006). These last two cases (and specially the last one) are in general grounded in socioeconomic conditions of inequality between those who are marginalized-and who cannot even construct their own historical memory-and those who establish the meaning of the historical past.
In conclusion, whereas other forms of collective memories, such as family memories, may sometimes be shaped by unequal power relations, the case of historical memories is quite different: these are essentially pierced and permeated by different kinds of relations of power due mainly to their essential heterogeneity.

Functions of historical memory
All the previous sections lead to the conclusion that historical memory is more oriented toward internal cohesion and the purposes of identity than toward a faithful reconstruction of a shared past. As Wertsch (2009) has written, "this is not to say that accuracy is not important or is not assumed by those doing the remembering, but it does mean that accuracy is of secondary importance and may be sacrificed to the extent required to serve other functions" (p. 123). In this sense historical memory is similar to individual forms of memory of personal experiences. Whereas the variety of functions of autobiographical memories are the object of recent discussion, most cognitive psychologists argue that the aim of memory is not to give access to past experiences, since the fact of giving access to past experiences is only a capability but not a function of memory (Klein et al., 2009). Autobiographical memory allows individuals to plan present and future behaviour and thought (directive function), to preserve and promote continuity and development of the self (self function), and to maintain and nurture social bonds (social function) (Bluck et al., 2005). In the same vein, historical memory also presents the following functions: (a) A directive function. A community can use its past to plan actions and project itself into the future, and to draw lessons from the past and avoid repeating past mistakes (Todorov, 2001). Authors such as Szpunar and Szpunar (2016) have even stated that the collective projection into the future (or "collective future thought," in their terms) is in fact the "driving force" that guides the reconstruction of collective memories. (b) An identity function. Historical memory shapes and sustains the community's identity and its internal cohesion through time, so that the community can become aware of its unity and peculiarity (Assmann, 1995). (c) A social function. Historical memory can help to strengthen the emotional bonding among the members of a community, and also to help them cope and heal from past traumas.
These functions are neither exhaustive nor exclusive, and probably differ across cultures and according to specific features of the historical time (Wang, 2008). In any case, they give us an idea of the way in which memory functions, in both individual and collective cases, and they distance memory itself from the pursuit of truth. As Blustein has stated, historical memory "is caught in a kind of tug of war between history and myth" (2008, p. 200): 12 on one hand, as a representation of the past, it aims to be accurate, but, on the other hand, it accomplishes other important functions for the community that are truth independent. This bipolarity that characterizes historical memory is responsible for the tensions that it maintains with history.

The relationship between historical memory and history
Historical memory maintains a dynamic and complex relationship with history (understood as an academic discipline). And in this it differs from individual memories and other forms of collective memories, such as family memories. Although some historians are currently concerned about the way in which families remember (Green and Luscombe, 2019), their interest is focused on memories' structure, mechanisms, and interactions with other institutions and not on their content per se. On the other hand, history not only researches historical memory as a particular form of shared social representation, but also uses historical memory as a source for understanding the historical event in question. This happens because history and historical memory are both directed towards the same object: a particular historical event that is studied by history and remembered by a collective.
Because of this common object, interaction between history and historical memory is inevitable. Historical memory can be complemented and reinforced by history, or it can be in tension with it. In some cases, historical research can be formally convened to prove the truth of a particular historical memory. In response to social demands or legal requirements, historians act as experts to determine whether historical memories comply with their claims of accuracy or not (Rousso 2008). For example, a team not of historians but of anthropologists recently started excavations in some mass graves in the Argentinian Chaco in order to provide legal evidence for the Napalpí Massacre of the Qom people that occurred in 1924. This massacre, longtime denied and then forgotten, has survived until today only through the testimony of the elderly (who have since died, except for one woman, Rosa Grillo), so the result of scientific research is of the utmost importance for the legal case that is in process and which may lead to a truth trial (Sinay, 2019). 13 In other cases, historical research itself can detect fallacious elements in historical memories. When this happens, the collective may react in unfriendly ways to the "intrusion" made by history that puts into question and even cracks a particular formation of historical memory. This is what happened in the case of the destruction of the Jewish community in Jedwabne, Poland, in 1941. The Polish inhabitants thought that the Nazi had carried out the massacre by setting fire to a barn with 1 600 Jewish people inside. But the truth was that the Polish inhabitants themselves had been responsible for the mass murder. Nonetheless, the version given by the Polish inhabitants was consolidated as the predominant historical memory, probably because it helped to strengthen a certain social identity: that the Polish were simply victims of the foreign occupation and never active participants in the persecution of Jewish people. The true story did not become nationally known until 2000, with the publication of the book Neighbors: The Destruction of the Jewish Community in Jedwabne (Gross, 2001: the Polish version was published one year before). This book sparked a controversy in Poland: many Polish people doubted the credibility of Gross's account, and this provoked an intensive two-year public and academic debate and even an official investigation by the Institute of National Remembrance (Wolentarska-Ochman, 2006).
Because historical memory is halfway between myth and history, "[historical] memory does not entirely retreat from history, and therefore [historical] memory becomes vulnerable to correction by history" (Blustein, 2008, p. 178). Blustein's argument could be formalized as follows: because historical memory, as any other kind of memory, has truth as one if its values, and because truth is the regulative ideal of history, 14 as it is of science, historical memory is answerable to the outcomes of historical research. Therefore, although historical memory may in some cases resist historical evidence, history provides a sort of "empirical base" to which collective representations of a shared and common past can be contrasted so that they will eventually be corrected. This nevertheless does not mean that historical memory can be reduced to the sum of beliefs about the shared past, as I already explained. Whereas history belongs to the discursive domain, historical memory includes memory activities and memory places. The contrast with history thus is not as straightforward as the contrast with empirical evidence is for an individual who has formed a false memory. It certainly takes more time and may involve practices and activities that go beyond the discursive domain, such as public manifestations.

EPISTEMIC RESPONSIBILITY APPLIED TO HISTORICAL MEMORY
The previous section discloses the main characteristics of historical memory that are significant for understanding what it could mean to be an epistemically responsible collective that remembers its historical past.
Unlike individual memories and other kinds of collective memories, relations of power, which are the product of the heterogeneity that characterizes historical memory, are intrinsic to their constitution, configuration, and maintenance, so it is expected that they are much more prone to distortion and falsity. These relations of power, together with the different truth-independent functions that historical memories accomplishes, are likely to doubly distance historical memory from the pursuit of truth.
What is more, false, distorted, or partial historical memories seem to be much more harmful than individual or small-scale collective memories. When a subject produces false memories, either because his or her memory is adaptive relative to a specific function or because he or she simply failed to be an epistemically vigilant agent, the negative consequences of these distortions or delusions seem to be minimal. If the memory is false but adaptive, the positive outcomes can compensate for the negative ones. This happens, for example, when one remembers a positively biased memory of one's past self: this memory can directly benefit one's current self-image and influence future behaviours and motivations for the better (Howe, 2011). On the other hand, if the memory is false and has no adaptive value, it can certainly present some disadvantages for the rememberer, especially if he or she acts guided by it. It is true that a false memory that is adaptive for the individual can also negatively impact on the rememberer's thick relationships. From the fact that some false memories with adaptive value for the rememberer can present negative consequences for the people close to him or her, we can infer the importance of remembering the event with integrity by taking the perspectives of those to whom we owe care into consideration. Nonetheless, false individual and small-scale group memories do not in general have negative consequences for people beyond the rememberer and the exclusive circle of individuals with whom he or she maintains thick relationships.
This presents a significant difference from historical memory. Whereas distorted and biased historical memories may not have any negative consequence for me or my family, they can present negative outcomes for people who are not members of the collective or who are at its margins, but whose perspective of the past event is generally excluded from the dominant and sometimes false representation of the past. A distorted historical memory can fulfil a specific function, such as national cohesion, but this fulfilment cannot compensate for the damage done to those who are excluded. This would be the case for the construction of the fallacious historical memory (very widespread until recently) that characterizes the arrival and domination of white settlers in the American continent as a civilizing process. Together with their symbolic exclusion and the impossibility of ending the grief over the death of their ancestors, this memory that conceals the great massacres that were perpetrated in the name of the supposed civilization denies the survivors of the First Peoples their possibility of participating in the life of the contemporary community. In doing so, it causes and reinforces their economic marginalization. As the case of the Indigenous peoples of Latin America shows, the conditions of their current poverty can be seen as a consequence of the discrimination and persecution that began in the past (World Bank, 2015). Because there is always some group who suffers from a collective distortion of the past (even when they cannot find a way to express their suffering), false historical memories seem to be much more noxious than false individual memories.
Consequently, historical memories are those memories that require the greatest vigilance due to their strong tendency toward falsity and distortion, and to the meaningful ways in which falsity and distortion negatively affect other people. It remains to be seen what kind of mechanisms collectives have at their disposal in order to be epistemically vigilant and responsible rememberers of their shared history, and what may push collectives to adopt this vigilant stance.
The first possible mechanism that comes to mind is metacognition. Do collectives have an ability to monitor their own memories, as individuals do? Accord- Arango-Muñoz and Michaelian (2020), large-scale groups do not have metacognitive capacities, because any change can be explained by appealing to decision-making processes of their members. Their reasoning is nonetheless based in a more general argument that denies that collectives are capable of remembering because neither they display mental states nor their members properly interact with each other during encoding and retrieval. However, I have conceived historical memory not as a mental phenomenon, but as complex and dynamic processes of interaction among elements of different nature (individuals, practices, institutions, material things, etc.) that take place through time. Although their argument thus does not directly concern the conception of historical memory that is at stake here, it is true that metacognition is a mental phenomenon that loses its essence if it is applied to historical memory. Collectives can perform different kinds of actions and practices to monitor the truthfulness of their memories, but it would be misleading to state that these actions and practices are automatic or well-defined mechanisms that develop quite spontaneously in a collective, as it happens for individual rememberers.
The strong tendency of historical memories toward falsity and distortion is not counterbalanced by strong mechanisms of control and monitoring, but quite the contrary: these are in fact inexistent. Whereas individual memory, thanks to metacognitive mechanisms, is quite reliable at least in what concerns its factual content, historical memory is not reliable. Collectives are not naturally prone to developing a basic epistemically vigilant attitude directed toward their memories, as individuals are. Therefore, actions and practices of control and monitoring can be the outcome of a collective effort exclusively, "of a willingness to explore these matters, to air the past and to pursue the evidence wherever it may lead, and to hear the testimony of those who feel they have been wronged" (Blustein, 2008, p. 141). In this collective and purposeful effort, there are two actions and practices that are essential to controlling and monitoring the accuracy of existent historical memories and avoiding the collective amnesia of some significant historical events. First, the outcomes of historical research and other historical resources such as archives and documents (which nowadays are more accessible to the general public through the internet) are a key source for accurately informing public debate and public discourse. Although collective memory cannot be replaced by history because of the other vital functions that it accomplishes (Blustein, 2008), it can become, through its interaction with history, closer to a historical understanding of the shared and common past.
The contributions of history are certainly essential, but they do not account for all the forms that a collective effort toward controlling and monitoring historical memories can take. Among the most important ones figures the willingness "to hear the testimony of those who feel they have been wronged." This is particularly important when historical research on the event has not yet started or is still underway, so those testimonies have not been incorporated into a historical narrative and may have been only sporadically heard by some members of the collective. One example of this case is the testimonies of the Grandmothers of the Plaza de Mayo, an organization formed in Argentina during the last dicta- (1976)(1977)(1978)(1979)(1980)(1981)(1982)(1983) by grandmothers whose aim was to look for their grandchildren stolen during repression or born in prison. Their voices were ignored by many local and international institutions that they contacted to get humanitarian aid, until the newspaper the Buenos Aires Herald published a letter written by them that announced that children were stolen and people had been "disappeared." This act of giving discursive space to marginal voices contributed to disseminating a fact that until then was at most believed or suspected only by some people (Abuelas de Plaza de Mayo, 2007). So, in certain way, it acted as a sort of vaccine against an eventual collective amnesia. 15 On the other hand, the Plaza de Mayo example is also useful for showing how collectives purposefully engage in actions and practices of control and monitoring: through the division of labour (Margalit, 2002). This means that members engage in different actions and practices to different degrees of involvement in order to make the collective into a vigilant epistemic agent. For example, whereas some of them, minimally involved, may assure the transmission of information, others, more engaged, may volunteer for organizations that do memorial work or advocate for the establishment of truth commissions (Blustein, 2008, p. 215). Collective measures can also be taken by institutional agents, as happens with the implementation of educational practices that train new generations in respect, solidarity and recognition of differences, and which instill a real collective concern about past injustices and past episodes of violence and immorality.
The perspective of those who were offended and injured in the past and who were (or still are) in an unequal relation of power with respect to the collective or some of its members is also essential for correcting biased historical memories for the sake of their integrity. For example, historical memories of many nations refer to a heroic past that often hides violent or even aberrant acts. The celebrations of October 12 as the anniversary of Christopher Columbus's arrival in the Americas in 1492 masks the atrocities committed by Columbus and other Europeans against the Indigenous populations. An integral reconstruction of those outrageous events cannot be omitted from an epistemically responsible historical memory. Partial historical memories present only one side of what happened, but do not fully account for the past event; although they may not necessarily be distorted, they are fragmentary and incomplete, thus not truly accurate.
But even when those testimonies are already part of an integral collective representation of the past event-as it happens nowadays with the abduction and illegal adoption of children during the last Argentine dictatorship-in many circumstances these voices still need to be heard to allow the present collective to better grasp the evaluative aspects of the past and the particular significant ways in which that past had a direct impact on people's life. Beyond the academic interest that these testimonies can have for historical science, the openness to other voices and experiences, especially those of minorities and marginalized members of the collective who were (or still are) in a disadvantageous position, demands epistemic and moral virtues of the collective that are absent from the basic version of epistemic vigilance described above. The basic version is focused only on the need to control and monitor actions, but ignores and considers irrelevant the attitudes that rememberers must present to adequately orient and shape those actions. The strong version incorporates the need for these virtuous attitudes, so it seems more adequate for understanding the epistemic responsibility of collective rememberers. Nonetheless, it is not clear in what principle or value it may be grounded, given that the idea of relationships of mutual care among the members of large-scale collectives seems suspicious.
The consideration for the pragmatic consequences of misremembering or forgetting is a preliminary intuitive answer and is frequently endorsed in the literature, which presents epistemic responsibility regarding memory as part of a general framework of taking responsibility for past injustices. According to the consequentialist approach, the collective vigilant stance is desirable because of the state of affairs it brings about: it is necessary to compensate and repair past injustices and also to prevent similar events in the future. This last characteristic, for example, is the one that Todorov (2001) attributes to "exemplar memories," which are defined as those that actively intervene in the present with the aim of avoiding repetition in the future. But these pragmatic benefits are not at all guaranteed by a collective vigilant stance: the construction of accurate historical memories does not necessarily bring about reparative actions, and may even be useless for avoiding repetition in the future. What is more, as Blustein (2015) explains, these pragmatic benefits assume that accurate and integral memories present only "auxiliary obligations" (p. 77) and are thus not really valuable per se. This implies that testimonies and different perspectives on the past remembered have no real value beyond their eventual contribution to reparative justice and future prevention.
Although the collective belief in the practical advantages of accurate historical memories may be a motive that leads certain members of the collective or certain collectives to acquire a vigilant stance towards the common and shared past, it is worth it to explore whether there are other ways of justifying collective epistemic vigilance in nonconsequentialist terms. Blustein (2015), for example, has proposed different nonconsequentialist approaches and highlighted the importance of drawing on insights of various theoretical sources in order to avoid an oversimplified analysis of the phenomenon. But his analysis is focused on the moral obligation of remembrance and not on the notion of epistemic vigilance that is at stake here, so it simply does not translate into this case. 16 Margalit's (2002) distinction between an ethics and a morality of memory (which is also reworked by Blustein, 2008) may be more useful for thinking about the justification of collective epistemic vigilance applied to historical memories. According to Margalit (2002), an ethics of memory is related to care and thick relationships and partiality, and a morality of memory is based on the concepts of right and wrong and impersonal duty. The duty of truthfulness may explain and justify collective epistemic vigilance towards historical memory. Because its high degree of heterogeneity, its diverse functions, and its configuration of power relations that doubles the distance between historical memory and the pursuit of truth, the collective vigilant stance can be conceived as a duty valuable per se.
Nonetheless, it is very plausible that real collectives that state being concerned by the duty of truthfulness are in fact partially concerned by the beneficial consequences that accurate representations of the past bring for the collective.
On the other hand, the idea that collective epistemic vigilance applied to historical memory can be grounded on care does not seem in principle to be very plausible, and has been one of the reasons invoked above to explore what sort of epistemic responsibility could be appropriate in the case of historical memories. Margalit is of the opinion that "humanity is not a community of memory" (2002, p. 9) because it is almost impossible to generate and sustain caring relationships among its members. He is probably right regarding the collective that refers to the whole humanity, but there are other forms of care that could be integrated into this picture in order to avoid the denial of all concern and compassion for understanding the reasons and justifications that may lead large-scale collectives to be vigilant epistemic rememberers. A large-scale collective can care about the collective itself, about its preservation or change, and about the appropriate collective recognition of all of its members' understandings of their past (Blustein, 2008), in ways similar (but not identical) to the way in which groups of business partners or coworkers care. This general care for the community does not need to presuppose the existence of relationships of care among the community's members. On the other hand, relationships of "extended" care can be forged among their members without that implying the existence of thick relationships among them. The Chinese philosopher Mengzi thought that the care we have for our intimate circle can be extended to others with whom we have no direct contact when we come to see their similarity to those close to us (Schwitzgebel, 2019). Therefore, members of the collective can extend care to other people with whom they only have thin relationships because they understand the similarities that exist between those who are dear to them and those who are not. Although it may be improper to consider large-scale collective rememberers to be communities of memory based on care (as many small-scale collective rememberers can be considered), it is possible to believe that the collective vigilant stance applied to historical memory is not totally grounded on its pragmatic benefits, because it can also be partially founded in a feeling of care (care for the community or extension of care).

CONCLUSION
This analysis has shown that despite the different characteristics that individual and historical memories present, 17 the vigilant stance that makes individuals and collectives responsible epistemic rememberers is based on similar grounds: pragmatic considerations about the negative consequences of misremembering or forgetting and a feeling of care. Although the proportions and interplay between these elements may differ (care may be predominant in individual and smallscale collective memories), as well as the specific actions and practices of control and monitoring, the general principles that govern the epistemic responsibility of all these different cases of representations about significant past experiences may partially justify the use of the same epithet-"memory"-to refer to these different representations (a use that is sometimes questioned in the literature).
In conclusion, the strong version of epistemic responsibility delineated above seems to be adequate for conceptualizing the collective epistemic responsibility that applies to historical memories. Collective rememberers are not only responsible for engaging in different kinds of actions and practices (such as the contrast with history) in order to control and monitor their memory representations and avoid collective amnesia, but are also responsible for developing certain kinds of virtuous attitudes such as being responsive to dissident voices that emerge outside or on the margins of the collective, or even allowing their emergence. It is this openness and responsiveness that allows a collective to avoid self-deception and to be self-critical. Epistemic vigilance applied to historical memory, just as to individual memories, partially depends on the development of collective virtue.
Certainly, large-scale collectives may sometimes fail to be epistemically responsible rememberers, such as individual and small-scale groups also fail to be. Nonetheless, the notion under scrutiny accomplishes its task, because it intends not only to describe the phenomenon but also to be used as a regulatory ideal.

ACKNOWLEDGEMENT
I am indebted to Rosa Belvedresi (UNLP / Conicet), who provided several ideas for the writing of sections 2 and 3, and many of the historical references mentioned in the article. I also would like to thank the two editors of this special issue, the copy editor and administrative editor of the journal, the anonymous referees who provided valuable feedback, and audiences at the Third Annual Memory Studies Association Conference (Madrid, 2019), Issues in Philosophy of Memory II (Université Grenoble Alpes, 2019) and the internal seminar of the research project "La normatividad en ética y en lógica: una perspectiva evolutiva" (Universidad Nacional de La Plata, 2019).

1
For example, it does not seem necessary to engage in monitoring and control processes if I do not remember now-while at work-the exact time next Tuesday when I agreed to meet my friend. But it is necessary to do so if I am not sure to remember the exact time on the day of the meeting, because the meeting itself is at risk. If in this last scenario I do not look at the texts I exchanged with my friend or call her directly to check the time, I will probably miss the meeting and be blamed by my friend (and even blame myself) for not having engaged in any kind of action to monitor my dubious memory. 2 It is true nonetheless that epistemic vigilance with respect to memory may vary depending on individuals and contextual factors such as their culture. 3 Because epistemic responsibility differs from moral responsibility as concerns memory, it is in principle conceivable-at least from a theoretical point of view-to be an epistemically responsible rememberer without being at the same time a morally responsible rememberer. Even if such a case is very unlikely, we can certainly imagine a very rational and Kantian human being who is obsessed with truth and thus extremely concerned about getting a complete picture of what happened in the past. That is why he or she decides to take into account the perspective of another person; so this truth-obsessed individual is not really moved by the care that he or she should provide to this other person. In fact, he or she cares neither about considering the other person's perspective nor about this other person; he or she cares only about accuracy. 4 Although Ricoeur explicitly highlights the limitations of this perspective (2006, p. 124-132), he also analyzes some aspects of collective memory according to categories generally attributed to individual memory, such as psychoanalytic categories (see Lythgoe, 2007). That is why he could be considered as an advocate of the individualist conception of memory. I recognize nonetheless that this idea may be controversial for some Ricoeur scholars. 5 This position is explicitly denied by Michaelian and Sutton (2017): "whereas analyses of collective belief focus on cases in which a group can be said to believe a single, determinate proposition, it is, in many cases of collective memory, implausible to ascribe a single, determinate memory to the group" (p. 147). But on the other hand, the authors also state that "in such cases of interactive encoding/interactive retrieval, there is interaction among group members at both encoding and retrieval; individuals learn and recall together. Such cases correspond to the sorts of ongoing transactive memory systems-stable groups characterized by a division of cognitive labor, with group members responsible for remembering different aspects of events and playing different roles during encoding and retrieval … Of the four forms of collective memory distinguished here, interactive/interactive cases-which we will refer to as cases of strongly shared memory-may have the best chance of representing a truly collective phenomenon, in the sense that we may legitimately treat the group itself as the remembering subject" (p. 143). This strong notion of collective memory based on the existence of a group-level rememberer can certainly lead one to conclude that the collective memory is single and determinate without however considering that each member of the group has exactly the same memory representation. Although this conclusion certainly contradicts the authors' previous statement, it also follows from the ideas expressed in the last quotation. 6 Assmann recovers Halbwachs's (1950) conceptual distinction between collective memory and historical memory (see some paragraphs below in the main text), but he redefines it. He uses the terms "communicative memory" and "cultural memory" instead, and considers them both to be cases of collective memory but to belong to different kinds. For him, collective memory is a broader concept that includes these two different modalities. 7 The notion of "memory activities" has been taken from Sue Campbell, and "memory places" from Pierre Nora (Blustein, 2008, p. 185). 8 I clearly do not deal with the problem of the nature of the collective who remembers. 9 The case of Spain is a good example of both meanings. The "Association for the Recovery of Historical Memory" aims to identify the bodies found in the mass graves used to bury people who died due to violence during the Spanish Civil War and the subsequent Francoist regime (Asociación para la recuperación de la memoria histórica, 2015). In addition, in 2007 the Spanish government sanctioned the "Ley de la Memoria Histórica" (Historical Memory Law), which aims to recognize the rights of those who were victims of repression and acts of violence during that period (Ley 57/2007). 10 An interesting question to explore that parallels the understanding of historical and collective memories is whether there can be a sort of collective historical knowledge or whether historical knowledge can be possessed only by individuals of the collective. 11 And this is probably one important reason that makes the use a single model of analysis to understand individual and collective memory consolidation processes impossible, as Anastasio et al. (2012) envisage. 12 Blustein (2008) uses the term "collective memory" in his book. Nonetheless, it broadly corresponds to the conceptualization I have made here of the notion of historical memory. 13 "These "truth trials" (juicios por la verdad), as they became known, were an innovation in Argentine justice, and possibly in the rest of the Americas. They were unlike ordinary criminal trials in that judicial action was expressly limited to investigation and documentation, without there being a possibility either of prosecution or punishment. They were based on the right (both of the relatives and of society as a whole) to know the truth, and the right of the relatives to bury and mourn their dead (derecho a duelo) (Brett, 2001). 14 Although narrativists like Hayden White have stated that history cannot be objective or truly scientific, I prefer not to enter into this debate and suppose, as Blustein does, that history belongs to the scientific realm and has truth as its regulative ideal, and that this means that "accuracy in what is said about the past is the primary criterion for assessing the quality of historical research and the achievements of historical inquiry" (2008, p. 178). 15 For the ways in which environmental and social scaffolds can enhance collective amnesia, see Tanesini (2018). 16 According to Blustein (2015), an act of remembrance can be considered (a) virtuous when it embodies and flows from some virtue or combination of virtues (virtue ethics); (b) morally obligatory because of the reparative justice it embodies in officially sanctioned practices of commemoration, recognition, and symbolic reparation (deontology); (c) appropriate if it conforms to the adequate expression toward what happened in the past, such as collective shame (expressivism). 17 The specific characteristics of small-scale collective memories have been sporadically mentioned, but have not been an object of analysis in this work. Nonetheless, as I briefly explained, they show more similarities to significant individual memories than to historical ones.