In The Elementary Forms of Religious Life, originally published in 1912, Durkheim (1960) described the main ritual attitudes, and more particularly commemorative rites, at some length. For example, Durkheim described a ceremony performed by the Warramunga, an aboriginal people of northern Australia, as commemorating and representing the mythical history of the ancestor. Durkheim emphasized that the officiant was regarded not as the incarnation of the divinity, but as an actor playing a role, who participated in a process of communication whose function was to make the past available to the present and thus to ensure the continuity of traditions. Songs, dances and mimes were important adjuncts to communication, lending rhythm to the narration of the mythical story. Durkheim also highlighted the mythical places where the events in the ancestor’s history took place, as these were represented in drawings on the ground or body paintings, and served as a communication medium, acting as topographical memories.

Without explicitly naming it, Durkheim therefore laid the foundations for the concept of social memory, in other words, the ability of a social system (group, clan, community, etc.) to remember in an autonomous way (independently of what the individual members remember), and manufacture memories (symbolic and topographical) that resist the passing of generations and guarantee the clan, tribe, or group a certain moral and semantic continuity. This sociological innovation in the study of memory was part of a visionary movement guided by an original intuition according to which society is a sui generis (i.e., self-produced) reality. It was in The Rules of Sociological Method that Durkheim established the concept of social fact as a reality that was autonomous of psychic and organic processes (Durkheim, 1950, p. 5).

The concept of collective memory was introduced by Halbwachs in the 1920s and 1930s. This French sociologist based his sociological study of memory on his original intuition that one never remembers alone: “it is in this sense that there would exist a collective memory and social frameworks of memory, and it is insofar as our individual thought places itself in these frameworks and participates in this memory that it would be capable of remembering” (Halbwachs, 1925, Foreword). In this excerpt, a distinction is made between collective memory and social frameworks of memory: collective memory is the memory in which individuals participate, while social frameworks are the cognitive and normative structures of various social systems (religious, family, school, professional institutions, etc.). It is worth noting Halbwachs’ reference to individual thought, and above all to the junction between collective and individual memory, and more particularly recollection (which has since come to be known as episodic memory). Individual memory is based on a memory that transcends the individual, and is even made possible by it. Halbwachs’ study of collective memory among musicians emphasized the importance of social frameworks of memory, without which the formation of musical memories (individual and collective) would be impossible (Box 1).

The social frameworks of memory must not be confused with the (individual and collective) memory of musicians. If people remember more or less the same facts when they share a common experience, it is because a social system that is external to them gives them the opportunity to do so (Halbwachs, 1947). Here, this is the institution of musical meaning: a social memory that is very specific and largely independent of what the musicians remember or how they organize or refresh their memories. It is interesting to note that Halbwachs (1950) himself favored the notion of social memory over that of collective memory, despite what posterity has retained (Sabourin, 1997). Halbwachs’ distinction between collective memory and social memory may have lacked precision, but it was no less essential on a historical and analytical level. The way was now open for authors to explore the distinctions between internal and external memory, and between autobiographical and historical memory in greater depth, although these concepts remained highly unstable.

This lack of precision seems to have been the hallmark of research in the human and social sciences. Although Halbwachs’ definitions of collective memory quickly fell into oblivion (Apfelbaum, 2010), it was nonetheless a foundational concept in memory studies, being to memory studies what the apple was to Newtonian physics!

Since the 1980s, the term collective memory has been used by social scientists to refer to phenomena as varied as memories and shared narratives (Knapp, 1989; Smith, 2004), commemorations (Schwartz, 1982, 1997; Ghoshal, 2013), myths and cultural scripts (Green, 2004), and culture (in the broadest sense). Serving more as a metaphor than as a clearly defined concept, it refers both to collective representations (worldviews, patterns or schemas, scripts, ideas, belief models) that organize the making of sense of individual memories (coding, semanticization), and to collective practices (social, cultural, community, etc.) that aim to externalize (inscribe, share, discuss, etc.) individual memories in an interindividual framework (Olick, 1999).

In the field of cognitive neuroscience, the infatuation with Halbwachs’ work has inspired a true social turn (Coman et al., 2009; Rajaram and Pereira-Pasarin, 2010; Saxena and Morris, 2016; Fischer and O'Mara, 2022). The concept of collective memory lies at the heart of the most heated debates (Laikhuram, 2022): Metaphor or reality? Are individuals the only ones who can remember? Do groups have a consciousness, a mind, a specific memory? Can we reasonably support the hypothesis of an extended mind that goes beyond the individual (Wilson, 2005)? According to Bachleitner (2022, p. 167), “in its collective form, memory thus is lifted out of the minds and brains of individuals and instead is socially generated in societal frameworks.” According to this externalist perspective, the concept of collective memory is similar to that of joint memory, understood as the property of a social group to actualize its own past transformations (Thierry et al., 1995). It can result from the coupling of individual memories (shared or interindividual memories) and/or from couplings with the social or cultural environment (social or cultural memories).

In the wake of this new enthusiasm for a rereading of Halbwachs’ work, some people questioned the actual added value of the concept of collective memory, and criticized its metaphorical use in what were very vague conceptual and theoretical frameworks. According to Kansteiner (2002), this burgeoning field of research had made virtually no theoretical progress, and had instead accumulated methodological issues linked to the metaphorical use of psychological and neurological vocabulary to explain collective memory as an extension of individual memory. Wass metaphor therefore just a catch-all category, with a high potential for scientific emulation around the idea of transdisciplinarity, when even within individual disciplines there was no real agreement on what the concept should cover (Confino, 1997)? Why multiply related notions (shared, common, discursive, collaborative memories, etc.) in the absence of any mapping and delimitation? For Olick and Robbins (1998), social memory studies were a “nonparadigmatic, transdisciplinary, and centerless” (p. 106) research enterprise that navigated in troubled waters “between chaos and diversity” (Olick, 2009). Gedi and Elam (1996) even called for the concept of collective memory to be abandoned altogether, arguing that it was simply a passing fad based on mystification.

Our proposal is not to abandon the concept, but to clarify and articulate the concepts of collective memory and social memory, reserving the concept of social memory for the communication operations of social systems, and the concept of collective memory for the operations of individual systems of consciousness. This may seem counterintuitive, but it allows us to extract ourselves from the metaphor of collective memory as the memory of a group, given that it is actually the memory of individuals as members of a collective. To justify this proposal, we will draw on recent research in psychology and neuroscience, as well as contemporary sociology.

Although the concept of collective memory has been the subject of much debate-and we can see from the above that it has multiple meanings, there seems to be unanimity over its operationalization through the concept of shared memory (Box 2).

The notion of shared narrative is also used by social scientists (e.g., Knapp, 1989). It is undoubtedly more accurate and exhaustive than the terms shared memories and shared recollections, which refer to singular and emblematic moments and do not constitute a full and complete memory. Moreover, even if these narratives are shared, their often polyphonic (or multivocal) character allows individuals to simultaneously retain multiple and contradictory points of view (Smith, 2004). In the case of collective trauma, Hirschberger (2018) showed that the dissonance between historical crimes and the need to uphold a positive image of the group can be resolved not only through disidentification from the group, but also through “the creation of a new group narrative that acknowledges the crime and uses it as a backdrop to accentuate the current positive actions of the group” (p. 1). For both victims and perpetrators, making sense of trauma is an ongoing process that is continuously negotiated within and between groups (see, for example, Kosicki and Jasińska-Kania, 2007).

While the question of self (or identity construction) is central to work on individual memory, particularly around the concepts of autobiographical memory and projection into the future (e.g., Tulving, 1985; Conway, 2005; Eustache et al., 2016), it is also central to research on collective memory (Box 3).

This research on the psychology of collective memory is consistent with theses defended in the social sciences, including Booth (2008)’s work on how memory contributes to the construction of identities, and Assmann and Czaplicka (1995)’s research distinguishing cultural memory (directly linked to identity) from history. This distinction between history and memory, which was discussed by Halbwachs, is a central one in the literature (Crane, 1997; Assmann, 2008; Wertsch, 2008a). Whereas historical science claims to be universal, exhaustive and impartial, collective memory always exists in the plural and is highly selective. History separates the past from the present and the future, whereas memory links them together. Memory always operates in the present: it is a continual rewriting of the past in the present for future use. Both sociologists and psychologists agree on this point (Schwartz, 1991; Coser, 1992; Wertsch, 2008b; Borer, 2010; French, 2012; Adams and Edy, 2021; Bachleitner, 2022).

At the intersection of research on shared memory and collaborative memory, a number of recent studies have focused on the role of communication in collective memory, where even conversation is regarded as an act of memory (Cyr and Hirst, 2019). Hirst and Echterhoff (2008, 2012), for instance, studied the role of conversations in the formation of collective memories, while other laboratory experiments have investigated how the temporal dynamics of conversations shape the formation of collective memories (Momennejad et al., 2019). Adopting the bottom-up approach defined earlier, Hirst and Coman (2018) showed that through communicative acts of memorization, memories can be shared within small groups and gradually expanded to the level of complex social networks to form a collective memory. As Hirst et al. (2018) aptly noted, the bottom-up approach treats the study of collective memory as an epidemiological project, seeking to understand why some memories spread throughout a group or community and others do not.

Collective memory is also operationalized through the concept of collaborative memory, with the focus on the social transmission and social contagion of memory (Roediger et al., 2001; Box 4).

All these studies highlighted an essential characteristic of collective memory, namely its malleability (Caron-Diotte et al., 2022). Collective memory is imperfect and necessarily imprecise: it is a process of reconstruction made up of distortions and glaring inaccuracies. Despite the drawbacks associated with this malleability, Brown et al. (2012) argued that the reconstructive nature of memory allows for greater cognitive flexibility and supports the construction and maintenance of identities (individual and collective). The malleability of memory thus supports the formation of shared memories.

Another benefit of this empirical research in social psychology is that it focuses on a key function of collective memory: collective selective forgetting (Choi et al., 2017; Hirst and Coman, 2018) or amnesia in collectives (Anastasio, 2022). This function has also been extensively described in sociology (Bucholc, 2013). Vinitzky-Seroussi and Teeger (2010) emphasized the importance of using collective silences (overt or latent) to reinforce memory and forgetting within the group, in order to manage one’s past. In psychology, collective silences have also been studied in the case of traumatic memories (Kevers et al., 2016).

Another important point to note is the role of metacognition in collective memory (Lund and Russell, 2022), namely, the capacity for self-reflection required for social interactions, through which collective memories are formed. From this perspective, Adams and Edy (2021)’s detailed analysis of a process of temporal negotiation revealed that changing the end of a story entails rewriting it in its entirety (see also Olick and Levy, 1997). This research is not dissimilar to the interactionist project in sociology (Fine and Beim, 2007) and the study of negotiations of meaning about the past (and future) by interactors (Borer, 2010). However, there is a key difference at the methodological level: sociologists observe interactions in situ (in the field), whereas psychologists observe them in vitro (in the laboratory). This type of decontextualized observation has many practical and analytical advantages for research, but there is inevitably a gap between the cognitive mechanisms that have been identified and the sociocultural environment in which these mechanisms develop.

As Rajaram (2022) aptly summarizes, one of the main issues for research in this bottom-up approach is the impact of culture, which constitutes a blind spot in some ways. It is here that the complementarity of the bottom-up and top-down approaches becomes clear (Box 5).

Despite the undeniable value of these innovative studies and the complementarity of their approaches, they have remained firmly within the confines of the laboratory. The cultural context has, of course, been taken into account, but often reduced to a few quantitative variables, for the purposes of the investigation. Moreover, on a more theoretical level, although collective memory has been operationalized through the concepts of shared memory and collaborative memory, the same cannot be said for social framework, which has been insufficiently problematized. The articulation between what takes place at the level of systems of consciousness and what takes place at the level of social systems clearly remains fragile.

It is on this point, it seems to us, that sociology today has the most to contribute to the scientific debate. It can draw on a range of theoretical resources (of varied origins) to support the founding intuitions of Durkheim and Halbwachs, thus obviating the need for metaphors or reifications. This is the theory of social systems (Parsons, 1951), which Niklas Luhmann brilliantly formalized and updated at the end of the 1990s. Curiously enough, whereas Halbwachs’ writings are at the heart of current debates, Luhmann’s writings continue to be totally ignored in the literature on memory studies (including sociology). We will therefore try to make up for this oversight.

Systems theory allows us to think of society as a sui generis reality: an autopoietic system [as defined by Varela (2017)] that produces and reproduces itself through communication. Social systems are communication systems, composed exclusively of communication operations (and not of individuals or human beings). For example, a group is not a gathering of human beings, from this perspective, but an intense sequence of communication events (self-referential and self-produced): a group is not defined by the people of its members but by the scope of its norms (Teubner, 1996). For Luhmann (2021), human beings must be considered part of the environment of society. They participate in the operations of communication, but do not communicate. Only communication (and thus society) communicates. We will use his new theoretical foundations to clarify the concepts of social memory and collective memory.

In Luhmann’s sociology, the concept of social memory refers very precisely to a specific operation of social systems: the constant distinction that is made between forgetting and remembering within communication processes (Luhmann, 2012, p. 137). Forgetting is essential, as it frees up the system’s information processing capacities and opens it up to new operations, preventing the system from becoming clogged up “by coagulating the results of its previous observations” (Luhmann, 2021, p. 396), as psychologists and psychiatrists have observed in the case of posttraumatic stress disorder (Mary et al., 2020).

This continuous discrimination, characterized by both great rigidity and great flexibility, allows the social system to “constantly say what is what” (Boltanski, 2009). For the social system, “memory consists in the fact that, in any communication, certain statements about reality can be presupposed as known without having to introduce them into the communication and justify them. Memory is at work in all operations of the system of society, that is, in all communications” (Luhmann, 2012, pp. 91–92). It is the primary role of institutions within social systems (Searle, 1995; De Munck, 1999), comparable to re-entry phenomena within the central nervous system (Edelman and Tononi, 2000). Social memory operates continuously, selectively re-imprinting the system’s own states. Repetition produces forgetting and remembering, always in the present.

How does the memory of society and social subsystems function? Social memory is not collective memory (as defined by Halbwachs), where individuals remember roughly the same facts when they are exposed to the same social conditions: “social memory is by no means constituted by the traces left by the communication in the individual systems of consciousness. It is, on the contrary, a proper performance of social systems, of communicative operations, with their own recursivity” (Luhmann, 2021, pp. 398–399). A social memory is formed by the simple fact that communication actualizes meanings, which are therefore already known to some extent. Repetition of the same references results in this continuous re-imprinting of usable meaning at the level of communication. Social memory does, of course, require the cooperation of systems of consciousness (as with any communication): social memory cannot function without individual systems of consciousness with memory. However, as Luhmann pointed out, social memory does not depend on what people remember or how they refresh their memory by participating in communication.

For Luhmann, every society depends on its own, self-produced memory. As oral communication is characterized by many communications taking place simultaneously, which cannot therefore be coordinated, ancient societies without writing necessarily used functional equivalents of writing (places, objects, sacred buildings) to refer to communication and provide memory for the system’s operations (preparing, supporting speech and listening). More specifically, in order to allow for the recognition of sameness or repetition, without having to rely on neurophysiological and psychological mechanisms, which are far too labile (e.g., Fanta et al., 2019), tribal or segmental societies (without writing) based social memories on sacred places, known spaces (topographical forms), and (sacred) objects and quasi-objects (symbolic forms), including rites, festivals, myths, and legends (Luhmann, 2021, pp. 439–440). The commemorative rites of the Warramunga, described by Durkheim, are a good example of this (see Section 2.1).

As the anthropologist Goody (2018) has eloquently demonstrated, the introduction of writing profoundly transformed the organization of society and allowed for the gradual formation of the main functional subsystems (religion, law, economy, politics). The introduction of writing thus constituted a key stage in the evolution of social memory: communication could henceforth be preserved in an autonomous way, independently of living memory and social interactions. It could reach absent ones. The social dimension gained in autonomy, compared with the material dimension. Until the modern era, as Luhmann noted, writing was above all regarded as a memory aid and a means of transport (transporting signs without having to move things): writing was a mnemonic device that profoundly modified the meaning of memory. The invention of writing allowed for the differentiation of a specifically social memory: writing was henceforth a memory that constituted itself, and people could now remember and forget more things than before. “Writing modifies the possibilities of developing a social memory independent of the neurophysiological and psychological mechanisms of individuals” (Luhmann, 2021, p. 211).

Printing emerged and developed as a technical infrastructure that could maintain and update a memory of society, separate from what individuals more or less remembered, and above all independent of the generational changing of the guard. In order to make this memory more widely available, public libraries, museums, and documentation centers were created. As Blouin (1999) showed, archivists play an essential role in the construction of a social memory: they ensure a complex function of mediation between what has survived and what we know.

Back in 1937, Talcott Parsons had already come to consider the cultural system as the maintenance (or memory) subsystem of society (Parsons, 1937). Modern society invented the concept of culture to designate its memory, and the cultural system was differentiated from the social system. As Luhmann suggested, the concept of culture allows modern society to self-adapt its memory to the requirements of the functional differentiation of society into autonomized subsystems. Culture is social memory: the filtering between forgetting and remembering, always in the present, and the link between past and future. As Pierre Bourdieu (1979) skillfully showed, the concept of cultural capital designates this accumulated past, which can be used in the present as a resource for the future. Schwartz (1982, 1991, 1997)’s work provides a very good illustration of this point (Box 6).

While the advent of writing and the development of printing constituted two important stages in the development of an autonomous social memory of the living, we are now embarking on a third phase, with the emergence of new communication technologies. Digital technologies are emerging as the new social conditions for memory (Hui, 2017; Box 7)). Within modern society, the mass media perform an important selection and sorting function on a global scale, guiding and framing the archiving of cultural products and memories (Figure 1).

Modern society can be described as a functionally differentiated society (Luhmann, 1999), within which each subsystem (law, science, arts, politics, economy, etc.) takes on part of the social complexity, processing it according to its own indecipherable code. This functional differentiation of society gives rise to increasingly autonomous subsystems with no central steering body to supervise the subsystems. Each communication subsystem is therefore regulated by a code and has its own memory and specific basis for sorting between forgetting and remembering: law has a jurisprudence that makes it possible to forget the details of each case¹; science has repertoires of publications that make it possible to forget the wanderings of each research study; education has selection criteria that make it possible to forget students’ social origins; and the monetary economy has an abstract mode of counting value designed to forget the origin of payments (which the legal system must sometimes correct), in order to facilitate transactions.

The mass media constitute a specific functional subsystem of modern society, an important institution of meaning at the global level that serves as a frame of reference and fulfills an essential function of sorting between forgetting and remembering. According to Luhmann (2006), each system is defined by the difference between itself and the environment, with its code constituting the unit of difference. In the case of the mass media subsystem, the code distinguishes between information and noninformation. Producing information is an internal operation for every system, and the reality of the mass media is always a duplication of reality. The mass media subsystem performs two essential functions (Luhmann, 2012): (1) semantic securitization: constantly saying what is what, at the level of the society-world (re-entry of the difference between forgetting and remembering); and (2) critical awareness: constantly sensitizing the social system (to criticism, surprises, news, novelty, deviance, conflict, provocation, etc.), keeping it constantly on the alert, and exposing it to challenge (like an immune system).

The mass media provide an often dramatic staging of events: surprise, local reference, quantities, and conflictuality (transgression of norms) all constitute information selectors. In other words, what constitutes the programming of expectations within the mass media system, what can be expected as being information, and what must remain without value. The mass media system extracts events from the world, defines the reference frame, fabricates the social reality, and internally produces information. Other social subsystems may or may not be sensitized by these media events and react to them, each in its own way, according to a specific timeframe: the economic system by increasing prices, the political system by taking collectively binding decisions, the legal system by stabilizing normative expectations, the scientific system by developing new research programs, and so on.

To sum up, at the risk of repeating ourselves: social memory is not collective memory. Social memory emerges from communication operations specific to social systems, whereas collective memory emerges from the operations of individual systems of consciousness when people participate in communication operations in social life (as members of groups or in interaction situations). However, we have yet to address the important issue of their articulation, in other words, the contagious co-evolution of collective memories and the social structure (Lee et al., 2010). In our opinion, Cicourel (2015)’s fusion model is not entirely convincing, as it collapses quite distinct levels of analysis (Barnier and Sutton, 2008; Sutton, 2008).