Enacting memories through and with things: Remembering as material engagement

For mainstream theories, memory is a skull-bound activity consisting of encoding, storing and retrieving representations. Conversely, unorthodox perspectives proposed that memory is an extended process that includes material resources. This article explains why neither representationalist nor classical extended stances do justice to the active and constitutive role of material culture for cognition. From Material Engagement Theory, we propose an alternative enactive, ecological, extended and semiotic viewpoint for which remembering is a way of materially engaging with and through things. Specifically, we suggest that one remembers when one updates their interactions with the world, a form of engagement previously acquired through sociomaterial practices. Moreover, we argue that things are full-fledged memories, since they accumulate and bring forth how we have materially engaged with them over different timescales. Last, we highlight the need for studies considering the cognitive ecologies where remembering takes place in its full complexity.

theories of extended cognition. This material turn agrees with current proposals that consider it crucial to seriously examine the cognitive relevance of things, an aspect long neglected in cognitive studies. This article aims to elucidate the relationships between memory, remembering and material culture.
In the first section, we analyse the mainstream and traditional accounts of memory, outlining the problems that derive from considering it an internal mental activity, consisting of the storage and recovery of mental states. In the second section, we examine theoretical proposals about memory that recognise material culture as having some role in mnemonic processes. We argue that these proposals, despite constituting unorthodox views, do not do justice to the role things have for cognition since they still consider material culture as subordinate to internal mental processing. Accordingly, we uphold the need for a model that considers materiality in all its seriousness and clarifies the process by which people and things co-constitute each other.
In the third section, we show why the Material Engagement Theory (MET) is a suitable theoretical framework for achieving such an enterprise and situate it within the landscape of ecologicalenactive perspectives on cognition. MET is a contemporary theoretical framework emerged in the field of cognitive archaeology. It combines elements of the embodied, embedded, extended, enacted, ecological and semiotic approaches to cognition. According to this framework, mind and matter co-constitute and intertwine to such an extent that they are ontologically inseparable. For MET, then, the study of cognition is the study of how humans engage with and through materiality over time.
Drawing from MET, in the fourth section we present our original view on memory. We claim that remembering is not an internal phenomenon but a way of interacting in and with the world. More specifically, we contend that remembering is a cognitive act by which we update in our interaction with the world a form of material engagement previously acquired through sociomaterial practices. Furthermore, we propose that thanks to processes of semiotisation, things become memories themselves, since they accumulate meaningful ways of engaging with the world over different timescales of material engagement. In our view, memory is an extended and continuous process that unfolds over different timescales of material engagement with and through material signs. Finally, contrary to traditional accounts of memory, we conclude that things endure not because we form an internal copy of them but because they remain meaningful for material engagement. This is achieved thanks to the processes by which, over different timescales, the material qualities of things become intertwined with the specific sociomaterial practices in which humans take and have taken part. Altogether, our proposal seeks to contribute to the field of memory studies carried out from ecological-enactive perspectives of cognition (e.g. Caravà, 2020;Heersmink, 2020a;Hutto and Peeters, 2018;Michaelian and Sant'Anna, 2021;Peeters and Segundo-Ortin, 2019).

Mainstream accounts of memory
Theoretical models of memory come in a variety of flavours. There are general and specific ones (i.e. dealing with a particular mnemonic system, such as procedural, episodic or semantic memory) (see Ribot, 1907;Roediger et al., 2002). Some models claim to be holistic, while others are specific to disciplines such as psychology, archaeology or neuroscience (Solomon et al., 1989). There are also formal models (i.e. devoted to explaining the structure of mnemonic processes) (e.g. Oberauer and Kliegl, 2006) and others that are clearly oriented towards practical applications (e.g. that investigate the use of memory in education or decision making) (e.g. Klein, 2015). 1 Needless to say, we could not possibly produce a detailed exegesis of all models of memory. Nevertheless, we do wish to draw attention to a series of ontological assumptions that underlie most of them. Some assumptions are linked to the very definition of 'memory ' and 'remembering' (i.e. what 'memory' is and what 'remembering' means). One of the perennial legacies in cognitive science is the conception of memory as a place (Roediger, 1980). In Locke's (1881Locke's ( [1689) words, memory is the repository or store-house where we store and retain memories (i.e. the objects of our past thoughts) to retrieve them later. Remembering is 'the power to revive again in our minds those ideas which after imprinting have disappeared, or have been as if it were laid aside out of sight ' (p. 97). Thus, memory is primarily about the psychological survival (Ebbinghaus, 2013(Ebbinghaus, [1885) of external stimuli under the form of mental states, somewhere on a private inside. Many others have endorsed this view by defining memory as a treasure-house or the backstage of performance (Cicero, 1967), a room (Freud, 1962(Freud, [1924) or a store (Atkinson and Shiffrin, 1968). The adoption of such spatial metaphors, so pervasive in definitions of memory and other cognitive phenomena, almost discloses the extent to which Cartesian metaphysics have influenced (and live on within) cognitive studies. It led theories to assume the primacy of the first person and the existence of an executive centre, independent of embodied action and the material world, where high-level reasoning is carried out (Smit, 2020). For even when contemporary researchers do not speak of memory in the same terms as Cicero, they continue to hold a localisationist ontology.
Meanwhile, in computational paradigms, it remains hegemonic to speak of memory as a process of encoding, storage and retrieval of information in the mind (McBride and Cutting, 2019). Information encoding formats can be varied: bits and chunks of information (Miller, 1956), propositional mental representations (e.g. Baddeley, 1988;Tulving, 1983), mental plans (Miller et al., 1960), mental images of various types (e.g. Bergson, 2005Bergson, [1911; Cornoldi et al., 2008), semanticconceptual networks (e.g. Lindsay and Norman, 1977) and scripts (e.g. Abelson, 1981;Schank and Abelson, 1977), among others. Regardless of the format, they all refer to mental, internal entities. For others, the locus of memory is not an abstract mind but the brain. Ribot (1895Ribot ( [1881), who proposed that 'memory is, in essence, a biological fact; by accident, a psychological fact' (1), and James (1900), for whom retention takes place as a 'purely physical phenomenon (. . .) in the finest recesses of the brain's tissue ' (pp. 291-292), pioneered this perspective. Based on this, Penfield's (1952) experiments showed that stimulation of the human cortex could evoke recollection and concluded that the place of memory is the brain circuits. Added to the proposal that the brain is an organ of computation (Marr, 2010(Marr, [1982), this view motivated studies that, to this day, attempt to unravel the nature of the neurobiological mechanisms of memory and the location of memoryrelated information in anatomical structures of the brain (e.g. the hippocampus or the medial septum) (e.g. Allen et al., 2020;Hanslmayr et al., 2019; for earlier neurophysiological research, see Scoville and Milner, 1957).
Another assumption associated with localisationism is that memory is an individual psychological capacity (Tulving, 1983): as private entities in charge of memory, the mind and the brain are inaccessible to others. Interestingly, this assertion is even shared by the somewhat more recent field of collective memory. Thus, several studies attempt to explain how social contexts unidirectionally affect individual memory processes and 'how memories can be biased through contact with others' (Echterhoff and Hirst, 2009: 106). The underlying hypothesis is that 'a given social setting can dictate what sorts of recollections are most appropriate or commensurate with individual goals of communication' (Roediger et al., 2017: 17, emphasis ours). Meanwhile, other studies inspired by the contributions of social psychology regarding social representations and culturally shared knowledge (Moscovici, 1961) focus on how individuals appropriate a community's understanding of its shared past (e.g. Carretero and Van Alphen, 2018).
We want to argue that the localisationist and individualistic commitments of mainstream perspectives seriously limit our understanding of memory. First, by situating mnemonic processes exclusively in the mind or brain of the subject, they neglect the analysis of how contexts of activity and material culture define the pragmatic usefulness of memory. If there is anything for which memory is cognitively relevant, it is how it eases the achievement of our objectives in everyday contexts of activity. In such contexts, things do not only matter to the extent that they generate sensory impressions that can be internally stored, as mainstream perspectives suggest. Second, by thinking of memory as a process of inputting, storing and outputting, mainstream views adhere to a substance metaphysics that neglects the dynamic aspects (Bickhard, 2019) of memory as a process. Furthermore, the emphasis on the individual dimension of memory leaves relevant questions unexplained (e.g. how do we communicate with others about what we remember if mnemonic contents are individual and private?). We will argue here that there is more to the social dimension of memory than the collective forces that can affect individual performance: memory is, itself, a joint activity. To move towards our characterisation of memory as a situated and collective practice with, through, and about material culture, in the next section we analyse contributions that have claimed that things do play some role in mnemonic processes.

Memory and things: a troubled relationship
Pragmatic and cultural perspectives have repeatedly drawn attention to the intimate relationship between cognition and the practical settings where it takes place (e.g. James, 1900;Merleau-Ponty, 1963; Sereda, 2011). Since cognition is situated, its study must acknowledge specific activity forms and their changing functions (e.g. Leontiev and Luria, 1972;Nelson and Fivush, 2000). From such a view, memory could not be a self-sufficient psychological function. As 'a by-product of everyday life' (Nelson, 1989: 131), memory is oriented towards obtaining a pragmatic profit (Alessandroni and Rodríguez, 2019). Therefore, understanding memory involves, inevitably, acknowledging where and for what we remember.
Since material culture is a fundamental part of our cognitive ecologies (Hutchins, 2010), considering pragmatic settings necessarily implies rethinking the role that things have in cognitive processes. On one point there seems to be some agreement: things are more than just lifeless substrates that motivate sensory impressions, and their importance far exceeds that of being the aboutness of inner mental states (i.e. external referents to which our internal states would be causally linked). As to what the concrete role of things is, the agreement is not so blunt. In what follows, we review two perspectives on the matter.

Objects as proxies for remembering
'I was still a little girl . . . one night . . . I remember . . .'. 2 This is how Leïla begins to narrate one of the plot knots of Les pêcheurs de perles, Bizet's famous opera (Cormon andCarré, 1923 [1863]: 15). In the story, a fugitive called Zurga desperately knocks on the door of Leïla's cottage, asking for shelter. Assuming great risks, Leïla promises to protect the man, thus saving his life. In gratitude for not handing him over to his pursuers, Zurga decides to give the girl a pearl chain necklace. In doing so, he says, '(. . .) take this chain and keep it always, in memory of me! I will remember!' 3 (p. 15). Leïla would always carry the necklace as a symbol of her loyalty to that promise. Years later, Zurga finds himself caught up in a love dilemma. Despite trying to seduce an enigmatic woman whose identity he does not know, she is determined to love Nadir, one of Zurga's most trusted friends. The choir foretells it all: Nadir will 'unchain' 4 (p. 21) the fury of the gods. Indeed, tormented by his anger, Zurga orders the murder of Nadir and the woman, who is willing to die alongside her beloved. Before leaving for execution, the woman gives a fisherman her pearl chain necklace and asks him to hand it over to her mother. Zurga is dumbfounded: the enigmatic woman is, against all odds, the same girl who years ago saved him from death. The outcome is known: Zurga forgives Leïla and Nadir and, to fulfil his original promise, sets a fishing camp on fire so that, during the distraction, the lovers can escape. Just before the end of the opera, Zurga tells Leïla: 'Leïla, remember, you saved me once, be saved by me' 5 (p. 31) while showing the necklace and dramatically breaking its chains.
In this fine example, it is quite evident that the pearl chain necklace is somehow related to memory. Some would say it behaves as an aide-mémoire, an artificial device specifically conceived to spur remembering (D'Errico, 2001). If it had not been for the pearl chain necklace, the end of the opera would most likely have been different (i.e. Zurga would have executed the couple without knowing the woman's identity). Vygotsky (1978) considered cases like this one in positing his theory of indirect memory. For him, indirect memory is a higher psychological function, of logical nature and cultural origin. Like other higher psychological functions (e.g. creativity), this specifically human capacity is semiotically mediated, involves conscious control and allows us to reorganise and self-regulate our behaviour. What does it mean that memory is semiotically mediated? In a nutshell, that we can actively influence the way we remember by making use of signs. The paradigmatic case analysed by Vygotsky is that of the knot in the handkerchief: When a human being ties a knot in her handkerchief as a reminder, she is, in essence, constructing the process of memorising by forcing an external object to remind her of something; she transforms remembering into an external activity. This fact alone is enough to demonstrate the fundamental characteristic of the higher forms of behaviour. In the elementary form something is remembered; in the higher form humans remember something. (Vygotsky, 1978: 51, emphasis ours) The process by which we transform and use the knot as a reminder to remember something follows, from this perspective, a series of steps. First, there is a person's desire to remember some mental content in the future. Since she knows she will probably forget it, she transforms whatever object into a symbol of that content by establishing an arbitrary, idiosyncratic convention (i.e. signs are 'self-generated stimuli'; Vygotsky, 1978: 39), only valid for her (i.e. if other subjects saw that knot, they most likely would not interpret it as a reminder). In doing so, the object becomes a vehicle for the meaning to be conveyed: the externalisation of memory transforms things into a type of memoria technica (Grey, 1751). Second, thanks to semiotic interpretation processes, the object acts back on its creator by bringing back meaning. Thus, on seeing the symbol, the subject deconstructs the sign she has created, recalls that she should remember and is guided back to the appropriate mental content (e.g. a belief).
Archaeological studies on social memory have embraced this view (e.g. Borić, 2010;Olivier, 2011; Van Dyke, 2019) when analysing how meaning-loaded objects have the potential to cue memories. For instance, Williams (2003) argued that, in the context of funerary practices, the use of objects (e.g. tombstones, drinking equipment and weapons) was a strategy 'by which the living sought to remember the dead, both recalling and constructing aspects of their identity, or destroyed, subsuming, dispersing others' (p. 10). In this way, he argues that objects can become symbolic substitutes for meanings culturally attributed to the dead. Similarly, Joyce (2003) contended that Maya societies used calendars and writing systems as technologies of memory to inscribe and share social memories (see also other contributions in Van Dyke and Alcock, 2003). Conversely, Mills (2008) reflects upon the role that objects play in forgetting, through the analysis of archaeological collections from Chaco Canyon, New Mexico. She shows that Chacoans destroyed, removed and repositioned objects that had symbolic meanings as a way of actively transforming social memories.
The idea that objects become proxies for remembering has a problem, both in its Vygotskian formulation and in the one found in archaeological studies: objects do not indicate, in any way, what we should remember. They act as a wake-up call, as auxiliary means (Daniels, 2001) that trigger the subject's memory processes. In this way, memory is still the retention and recovery of information in and from the inner world. The novelty is just that external elements can mediate this activity. It is also worth noting that although in this view things are functionally important (i.e. they help us to remember), they appear to be semiotically contingent: any object can be in place of any content, regardless of its material characteristics. Accordingly, Zurga's necklace could have referred to other quite varied contents (e.g. an important date). Moreover, Zurga's story could equally have been evoked by other objects besides the necklace (e.g. a knot in a handkerchief). Things behave, therefore, as nonsemantic factors (Shanon, 1990). As here things do not add nor subtract anything to the meaning they refer to, they are mere proxies for remembering. Of course, this leaves some issues unsolved. Let us just suggest one of them: Why does Zurga break the necklace after recovering his memories? After all, if the association between the object and the memory was no longer significant or useful, it would have sufficed for Zurga to 'dissolve' it mentally, on the plane of the arbitrary symbolism that created it in the first place. Why does Zurga need to get materially engaged with the necklace? We will come back to this later.

Things as instantiations of mental states
Theories of extended mind, active externalism and vehicle externalism (Clark and Chalmers, 1998;Donald, 1991;Hutchins, 1995;Rowlands, 1999;Sutton et al., 2010) argue that, under specific circumstances, information-laden things can instantiate cognition, therefore implying that cognition need not be skull-bound (e.g. Heersmink, 2020b;Kirsh and Maglio, 1994). Donald (1991) was among the first to propose that things can instantiate memory. Particularly, he argued that the development and use of visual symbolic systems (e.g. cave paintings) resulted in a new stage of human cognitive evolution where things started to serve as external memory storages. In contrast to the engram term coined by Semon (1921), Donald referred to external symbols as exograms: while an engram is 'a single entry in the biological memory systems' (i.e. a memory trace of a past experience stored in the neural substrate, see Josselyn and Tonegawa, 2020), an exogram is an external record of an idea, an idea stored in a thing (Donald, 1991). Due to their neural nature, engrams are limited in storage, difficult to alter and short-lived. Contrarily, the external nature of exograms makes them more stable and flexible for improving thinking, as well as indefinite in storage capacity: things do not forget as humans do (pp. 314-319). As an example, Donald provides an analysis of how writing systems, such as cuneiform or Egyptian hieroglyphs, work as storages for information. He shows that, by rearranging in space little figures associated with ideas and/or sounds, people were able to offload onto the environment complex information that would have been otherwise difficult to remember and develop (e.g. literary texts or mathematical theorems) (pp. 333-334). He also argues that the continuous use of writing systems through the generations produced 'a collective memory bank' that made available a virtually indefinite storage for information (p. 311), thus highlighting the cultural and longitudinal nature of external memory. 6 More importantly, Donald believed that exograms are functionally equivalent to engrams inasmuch as they make up 'a storage and retrieval system that allows humans to accumulate experience and knowledge' (p. 309).
After Donald, other authors have argued that things can instantiate cognition resorting to the idea of functional equivalence (Clark and Chalmers, 1998;Rowlands, 1999;Sutton et al., 2010 for digital technologies see Heersmink and Carter, 2020). Clark and Chalmers'(1998) view, for instance, maintains that memory extends into the environment insofar as external resources work as internal cognitive processes. The quintessential example is the Inga-Otto thought experiment. In this experiment, whenever Otto needs to remember something, he writes and consults information on his notebook. Since the notebook plays for Otto the role usually played by biological memory (i.e. the storage and retrieval of information), the authors argue that it can be considered a part of his cognitive apparatus (p. 12). This conclusion is based on the parity principle: if as we confront some task, a part of the world functions as a process which, were it done in the head, we would have no hesitation in recognizing as part of the cognitive process, then that part of the world is part of the cognitive process. (p. 8) Relative to Vygotsky's theory of mediated memory, Donald's perspective on exograms and extended mind theories bring something new to the table: things are important for remembering not because they elicit memories (i.e. because they act as proxies), but because they can actually instantiate mental states. Given that exograms can store information, they expand and complement engrams, as does Otto's notebook. However interesting this rapprochement between things and memory, classical extended theories present some problems. First, memory remains a psychological function carried out, primarily, by an inner mind. If exograms and notebooks can act as sources of memory it is because they are considered functionally equivalent to internal remembering processes. As pointed out before, by equating external resources to inner mental processes, extended perspectives remain 'inevitably tied to the boundaries between inner and outer that they wish to undermine' (Di Paolo, 2009: 10). As a result, while defending that cognitive processes can be extended into the world (Clark, 2008: 79), these theories leave the representationalist logic of information processing untouched. Second, they advocate that external resources can instantiate memory because they carry information. For instance, it is enough that Otto's notebook contains propositional information written on its pages to be considered part of his memory. This information-centred approach overlooks important aspects of our relationship with materiality. When we engage with things, we do not simply extract information from them. We can carry, touch, use, hold and arrange things (among other possibilities), and each of these activities has cognitive relevance. As Sutton (2007) argued, external resources may hold information in different ways, or not hold information at all. This reinforces the idea that things should not be considered as resources for cognition only because of their informational properties. Last, by giving primacy to the degree to which materiality resembles the inner processing of information, classical extended theories do not account for the unique characteristics of things and how they affect or constitute memory (Sutton, 2007). Their concern seems to be only that cognition is extended but not where it is extended. Somewhat paradoxically, this implies that materiality can act as an external memory despite its materiality. 7 However, there appears to be a big difference between remembering a life event by writing it down in a notebook and remembering it through the use of a necklace. It is clear for us that differences in how materiality influences memory should be carefully addressed. This is something that cognitive theories have only recently started to consider and, therefore, necessitates further elaboration (Heersmink, 2020b;Hutto and Myin, 2013;Malafouris, 2004;Malafouris and Koukouti, 2018;Prezioso, 2021). With the aim of contributing to this ongoing debate, in the remainder of this article we develop the argument that things constitute memory because of the particular characteristics of our material engagement with them.

MET: a transactional stance on cognition
So far, we have shown that mainstream perspectives consider remembering as the manipulation of representations in the head. Other views extend the phenomenon beyond the individual, also considering things. When they do, however, they exhibit limitations. For perspectives that consider things as proxies, objects are arbitrary symbols that remind us to remember. Meanwhile, for classical extended approaches, things can work as realisations of internal mental states as long as they are functionally equivalent to the latter. But is it possible for things to be otherwise related to memory? In this article, we argue it is, from the extended, enactive, ecological, distributed and semiotic stance of MET (Malafouris, 2013(Malafouris, , 2019a(Malafouris, , 2019b. Below, we briefly explain what cognition and things are for MET. In the last section, we elaborate on why MET implies a new look at memory as a situated activity involving things.

Think-ing as thing-ing: an extended, enactive and ecological approach
If for the Cartesian tradition cognition is a property of res cogitans independent of the body and the environment, for MET it is quite the opposite: mind and matter co-constitute and intertwine to such extent that they are ontologically inseparable. Here, the study of thought processes is a situated enterprise. For one thing, this means that it is impossible to understand cognition as detached from its ecological conditions: to study cognition is to study cognitive ecologies (Hutchins, 2010), the space where brains, bodies and things conflate (Malafouris, 2004(Malafouris, , 2013. This is because thinking can only occur in structured material environments (Ingold, 2011(Ingold, [2000), under specific material and social coordinates. 8 This is a strong hypothesis, as MET goes beyond perspectives that consider the sociomaterial as a context or as a causal influence on thought: the mind is 'situated within and constituted by the material world rather than merely being about it' (Malafouris, 2018: 756). Things and sociocultural practices are the matter of which thought processes are constituted (Gallagher and Ransom, 2016;Ransom, 2019). Conversely, for MET materiality is not an instrumental dimension through which internal mental states are externalised: we do not think in the head and then express our thoughts using things as a medium. On the contrary, thought is articulated with and through things, during meaningful instances of material engagement. Hence, thinking refers to acts of consciousness (Malafouris, 2019a) that emerge during transactional and transformational exchanges with the material world (Malafouris, 2019b): thinking is thing-ing (Malafouris, 2013(Malafouris, , 2014, interacting with, through and about things within specific forms of activity. The material and interactive turn proposed by MET coincides with the aims of the ecologicalenactive stances of cognition, for which a core challenge is 'to do justice to the role of the environment in guiding and constraining cognitive processes' (Kiverstein and Rietveld, 2018: 149). First, because for ecological-enactive approaches, cognition is also a relational phenomenon resulting from the agent-environment co-determination (Varela et al., 1991). Accordingly, the minimum unit of analysis is that of an agent interacting in an environment: cognition is 'an embodied engagement in which the world is brought forth by the coherent activity of a cogniser in its environment' (Di Paolo, 2009: 12). Second, the idea of mind as minding (Malafouris, 2019a) -an extended, inbetween process emerging from the engagement with materiality -connects strongly with the ecological-enactive goal of understanding how and when, through interactions with their environment, organisms produce and sustain different forms of organisation. This point also implies the rejection of a substance metaphysics in favour of a process ontology (Bickhard, 2019), in which cognition is cognitive becoming (Gosden and Malafouris, 2015). Last, MET and ecological-enactive paradigms agree that the usually considered representation-hungry problem types (Clark and Toribio, 1994) can be addressed without postulating the existence of skull-bound mental representations, namely, 'stand-ins' for external states of affairs (Kiverstein and Rietveld, 2018). As Malafouris (2019a) nicely explains, 'things have a cognitive life not because of what they represent, or of how they can be represented, but for what they do' (p. 8).
As far as the study of the relationships between cognition and things is concerned, MET is a much richer perspective than orthodox paradigms, in at least two senses. First, unlike the empiricist tradition that sees things as incidental collections of sensations (Locke, 1881(Locke, [1689), MET maintains that things are active agents of the transactional process that thinking is. Second, MET rescues things from the place of the aboutness of cognition. Things are not fixed referents to which our mental representations point (Malafouris, 2004) but full-fledged cognitive agents.

Material engagement and enactive signification
Due to their material qualities, their participation in instances of material engagement and the sociocultural settings of which they are part, things become meaningful. In this sense, for MET things develop as semiotic entities rather than just being lifeless bits of reality. The process whereby things acquire specific meanings and become material signs in situated contexts where bodies, brains and things conflate is termed enactive signification (Malafouris, 2013). Material signs act in unique ways. While linguistic signs arbitrarily symbolise content that is independent of them and can be enunciated, material signs become a constitutive part of what they express by being enacted. Simply put, things acquire and articulate meanings in virtue of what we do with them (Malafouris, 2013(Malafouris, , 2019b. MET challenges the standard conception that meaning is an a priori mental product that is then passed on to things. Things do not 'sit in silence waiting to be embodied with socially constituted meanings' (Olsen, 2010: 10): meanings are the emergent product of material engagement. For Afro-Cuban diviners, for instance, the concept of power is so closely related to a consecrated powder they engage with while conjuring deities that, for them, powder is power (Holbraad, 2007). The process by which powder becomes power (i.e. powder comes to signify power) emerges in the action of conjuring divinities through and with powder. The diviners' power is, therefore, strictly correlated with the material qualities of powder, the gestures it gives rise to and the sociomaterial practices where it is used.
This example illustrates why things, due to their capacity to shape our interactions with the world, have ontological precedence over concepts. In other words, 'the concept is meaningless without the actual substance' (Renfrew, 2001; as quoted in Malafouris, 2013: 97). Things, as material signs, embody meaning through qualitative, spatial or temporal relations (Knappett, 2005) and bring forth concepts as concrete exemplars and substantiating instances (Malafouris, 2013: 97).

Material engagement, social interactions and normativity
Material engagement is social in nature. Like any other aspect of our being-and-becoming-in-theworld, it occurs in social situations that constrain and constitute it at the same time: socialisation and materiality coalesce within sociomaterial practices, thus making it impossible to separate them except for pedagogical reasons. 9 Although many empirical studies on cognition fail to consider this fact in their methods (see , the importance of the social is much more recognised in the theoretical sphere. For Leontiev (2009Leontiev ( [1978) and his Marxist model, for example, it was clear that individual activity does not exist outside the system of social relations in which it is part. The individual-society dualism, deeply rooted in the cognitive sciences, does not hold up because, ontologically, both the individual and society are dialectically co-determined (Castorina, 2020;Gergen, 2003). However, at the processual level of ontogenesis (i.e. the developmental history of an organism), the social aspect undertakes perhaps greater importance than at the ontological level. Newborns are incredibly dependent and non-specific organisms (Karmiloff-Smith, 1996). Human babies, unlike babies of other species, require social contact and constant attention from their caregivers to survive. It is only due to social interaction through increasingly complex modes of relatedness that babies can progress and acquire skills (Stern, 1985). The idea that biology is socially oriented from the start (Wallon, 1970(Wallon, [1942) has been an axiom for research in developmental psychology. Perhaps this is why the academic literature has been greatly concerned with how aspects such as the coordination of body movements or the musical components of vocalisations (e.g. Papoušek and Papoušek, 1981) are organised within dyadic interactions (i.e. adult-baby interactions, see Trevarthen and Hubley, 1978). In contrast, very little is known about the relationship between early interpersonal communication and material culture, probably due to the long-standing subject-object dualism which has skewed much of the psychological research (Malafouris, 2019b;Smit, 2020).
Fortunately, several authors have gradually begun to recognise the relevance of early triadic interactions, namely, adult-object-baby interactions that occur before 9 months of age (e.g. Striano and Stahl, 2005). This triadic turn in the study of cognitive development is crucial. On one hand, because objects are part of interactions from the beginning. For 'even when the child cannot blend in the same communicative act one object and another person, he is placed by others in meaningloaded material scenarios' (Alessandroni et al., : 1556. On the other hand, because, as mentioned, things are more than bundles of sensory impressions. By taking part in sociomaterial practices that are sustained over time, things acquire meanings that are not accessible through mere individual exploration (Rietveld and Kiverstein, 2014;Rodríguez, 2012;Vietri et al., 2022). In these cases, access is always semiotically facilitated by others: during interactions, adults dynamically modulate the opportunities for action available to the infant (Nonaka and Goldfield, 2018) by deploying meanings through a plethora of communicative strategies (e.g. demonstrations of uses, gestures, or language) Moreno-Núñez and Alessandroni, 2021;Vietri et al., 2021).
This fact has enormous implications: if children manage to materially engage with the world in a meaningful fashion and can communicate with others, it is thanks to other people who have acted as semiotic filters (Rodríguez and Moro, 1998), actively shaping niches of joint action. Caregivers are therefore guides who, little by little, introduce the child to the culturally relevant action formats that they embody during interaction (Kärtner, 2018). Recurrent patterns of activity also involve a nondiscursive and enactive normativity: once specific interactive formats become ritualised (Rączaszek-Leonardi et al., 2019), adults expect children to be increasingly capable of engaging with material culture skilfully. Thereby, action becomes 'subject to normative assessment as better or worse, as more or less correct given the specific demands of the situation' (Kiverstein and Rietveld, 2018: 154). Within this framework, 'rationality is all about acting comme il faut' (Alessandroni, 2021: 175), that is, complying with the proper ways of taking part in sociomaterial practices.
It is worth recalling, as cognitive archaeologists have pointed out, that sociomaterial practices and their normativity extend beyond the situated and ontogenetic timescales and unfold through long-term trajectories of human cognitive becoming (Gosden and Malafouris, 2015;Malafouris, 2013;Overmann and Wynn, 2019). This is so because, within the sociomaterial dynamics of interactions, 'form gives rise to patterns of practice in people, and, equally, patterns of practice help create form' (Gosden and Malafouris, 2015: 703). Through their interweaving, things and sociomaterial practices come together into traditions, namely, 'specific sets of procedures, gestures, tools, materials, finished products, and beliefs and attitudes towards actors and materials' (Gosselain, 2008: 152). In this sense, traditions are appropriated, redefined and passed on in specific contexts by interacting with things and knowledgeable others who act as guarantors of specific practices.

A material engagement perspective on memory and remembering
In the previous sections we have shown the limitations of traditional perspectives on memory and developed the foundations of MET. From this theoretical framework, things are not passive referents of cognition, nor do they possess meaning solely due to their informational capacities. Contrary to traditional cognitive studies, MET examines how our ways of thinking and socialising are embedded in 'elementary gestures of enactive material signification' through and with things (Malafouris, 2019a: 3). Applied to memory and remembering, the approach of MET situates things back into our cognitive ecologies. It considers the active qualities that things have and how they shape our social and cultural worlds over time. By taking into account the premises of MET, in this section we present a new way of understanding the relationship between things, memory and remembering.

Remembering as material engagement
Once the standard representationalist approach is abandoned, it cannot be argued anymore that remembering is an internal mental process. From MET, remembering would be a way of engaging with and through things, namely, an enactive, extended, semiotic and situated activity. Consequently, the one who remembers would not be a subject but a-subject-interacting-with-and-through-athing: the subject remembers as much as the thing remembers itself, and it is not possible to differentiate these two terms. 10 More precisely, remembering occurs in that in-between space that emerges when brains, bodies and things conflate according to the sociomaterial dynamics of interaction.
However, not every interaction entails remembering. We propose that we remember whenever we update in our interaction with the world a form of engagement previously acquired through sociomaterial practices. We should note that our approach is in line with Bartlett's theoretical proposal, who, despite not analysing the role of materiality for remembering, studied memory in relation to behavioural regularities. The notion of schema, so central to his theory, refers to 'an active organisation of past reactions, or of past experiences, which must always be supposed to be operating in any well-adapted organic response' (Bartlett, 1995(Bartlett, [1932: 201). In terms of Bartlett's (1995Bartlett's ( [1932) works, remembering can be defined as a behavioural attitude towards the world that emerges whenever we are re-enacting schemata (p. 208). A simple example is in order. Children appropriate the canonical functions of artefacts around 10 months of age. When they do so, objects become functionally permanent (Rodríguez, 2012), and children become capable of performing canonical uses of objects, that is, uses that are congruent with sociocultural norms (e.g. a cup should be used to drink liquids). These norms of cultural use are not inherent to artefacts themselves but learned through other subjects who (en)actively shape the ways children interact with things (e.g. demonstrating the canonical use of a cup). As the child incorporates the interactive dynamics that characterise the canonical uses of objects, she becomes more and more skilled at bringing forth canonical ways of materially engaging with the world. When the child re-enacts these particular kinds of interactions, she updates a normative way of engaging with material forms derived from a longer tradition of sociomaterial interactions. We would argue that she is remembering how engagement should unfold in her community of practice and, by doing so, also reshaping the characteristics of current and future sociomaterial interactions.

Things are memories
In addition to arguing that remembering is a situated activity that takes place with and through things, we also contend that things are more than symbols or material instantiations of internal mental states. Things accumulate ways of engaging with the world over different timescales of material engagement: from that of online action where memories are enacted (i.e. a child using a cup or a potter making a pot), through the ontogenetic one related to the developmental processes of individuals, up to broader transgenerational trajectories. Through enactive signification processes occurring on each of these timescales, things become material signs of the history of meaningful exchanges that took place with and through them. Thanks to this multi-temporal process of semiotisation, material signs bring forth the sociomaterial practices which gave rise to them. In doing so, they do not replicate past forms of action in the present. On the contrary, they enable new ways of interacting in the particular sociomaterial settings where we remember, structuring the characteristics of the material engagement. By virtue of this dual role (i.e. things actualise previous ways of material engagement and shape current ones at the same time), things are memories. 11 A significant archaeological example is the Minoan footed goblet, also called egg-cup (Macdonald, 2010). This goblet shape, decorated with different styles (e.g. polychrome, light-ondark), lasted from the Early Minoan IIA (2700 BC) to the Middle Minoan IB (1800 BC). As Macdonald (2010) writes, the goblet 'had been accessible to large sections of the population, and had been used in countless generations and understood by them as a key element in Knossian traditions of daily consumptions and formal celebrations' (p. 205). By continually engaging with and through goblets, members of the Minoan society brought forth and fostered the persistence of specific sociocultural practices. In this sense, it is sensible to argue that each goblet served as a memory of the particular actions required to comply with the sociomaterial practices of the Minoan culture (i.e. drinking at feasts and in communal rituals), notably those of the Knossian society (from Knossos, Crete). Egg-cups, with their peculiar forms and changing decorations, also served as memories for artisans by reminding them to employ specific technical skills, follow cultural beliefs, choose certain raw materials and adopt particular attitudes towards those materials. Finally, egg-cups also acted as memories by pragmatically supporting the making of new pots which, at the same time, updated the egg-cup making tradition and enabled the following generations to learn long-standing technical skills and acquire associated beliefs. In brief, thanks to their emerging from a tradition of egg-cup using and making (i.e. the accumulation of previous episodes of material engagement with egg-cups), these goblets became material signs and memories for the Minoan society. Furthermore, in the case of pottery-making, it makes sense to conceive of the meaning of the egg-cups as an emergent result of the recurrent meeting of the potters' hands with the clay and not as a static meaning encoded and stored in their private minds. 12 This example shows why things can be regarded as memories that are re-enacted through material engagement.

Remembering is cognitive becoming
From MET, remembering is a way of interacting with and in the world, in the here-and-now. Likewise, things are memories not because they stand for encoded information but because they accumulate and bring forth particular ways of material engagement (Malafouris, 2004). Therefore, the pragmatic coordinates in which the activity of remembering takes place are not anecdotal. Conversely, we propose that they are essential, constitutive of remembrance. If remembering is meaningful, it is not because it allows us to dive into a nostalgic past but because it is functional in the present -a present that demands us to re-enact specific modes of interaction with the world. As Nelson (1989) argued, 'what is remembered should be that that enables the individual to carry out activities, to predict, and to plan' (p. 144). Nelson alludes here to the pragmatic profit (Alessandroni and Rodríguez, 2019) of interacting with the world: what makes a memory meaningful is its practical value, the set of active possibilities that such a memory enables. These possibilities rest within the set of possible actions that specific things, acting as memories in the situated context of engagement, enable to carry out according to their material properties and histories of uses. We want to adhere to the idea that memory is not primarily oriented to the past, but that 'the actual remembering process happens in the present and usually for the sake of the future' (Malafouris and Koukouti, 2018: 175). 13 Since 'you can never plan the future by the past' (Burke, 2005(Burke, [1791: 55), to understand memory is to understand the activity of memory as it unfolds online. For this reason, the modes of interaction that remembering allows to bring forth in the present could not be just copies of previous modes of interaction. Memory does not reproduce meanings; it (re)constructs them (Bartlett, 1995(Bartlett, [1932). Therefore, we argue that remembering is cognitive becoming: a continuous situated process enacted through and with things in the context of material engagement where brain, body and culture conflate (Gosden and Malafouris, 2015;Malafouris, 2013Malafouris, , 2019a. 14 Returning to the example of the previous subsection, whenever a child uses a cup to drink, she is not bringing back a past event structure. Most probably, the situation she finds herself in now is not a reproduction of past situations. Neither the physical environment, nor the degree of thirst, nor the people around them, nor the actions being carried out are probably the same as those that characterised past situations. What remains unchanged between both contexts? We propose that it is the very possibility of re-enacting a profitable interaction in the setting in which the child is, to adapt to specific needs, through a functional material engagement with the cup.

Remembering ties together different timescales
The fact that memory is an activity that occurs online is by no means the same as saying that it does not have a history. On the contrary: memory is a widespread practice, both in space and in time. This is because each instance of material engagement is embedded in histories of interaction that unfold, at the same time, across different timescales (Bietti and Sutton, 2015;Gosden and Malafouris, 2015;Malafouris, 2013;Rączaszek-Leonardi et al., 2019).
Returning to the example of the child, the possibility of canonically using a cup at the online timescale (i.e. the here-and-now) constitutes a form of memory made possible by a history of interactions with different cups, facilitated by adults, that have taken place in earlier moments of her life (i.e. the ontogenetic timescale). Throughout the ontogenetic development, children construct, thanks to the guidance of others, general ways of engaging with the material world that, as one of us proposed, should be seen as cases of conceptual thinking . However, at the same time, these interactive dynamics could not exist without the broader transgenerational trajectories of material engagement (Gosden, 2005). If caregivers introduce babies into the practice of drinking from cups, it is because they themselves were introduced into this practice by their caregivers and because cups, as meaningful artefacts, have endured to our days as memories. In fact, as the example of the Knossian egg-cups illustrates, cups have been part of sociomaterial practices for a long time in varied cultural contexts.
We therefore propose that material engagement provides coherence and cohesion to our cognitive experience: while bringing forth specific modes of interaction with and through the world, material engagement ties together different timescales.

Remembering is material
Remembering is inseparable from the material properties of things since the dynamics into which cognitive practices are enacted are also set by the forms things take (Gosden, 2005;Gosden and Malafouris, 2015). Zurga's example is once again enlightening: remembering is so material to him that in order to forget he must break the necklace. If, as we argued above, remembering occurs through and with material signs, then it cannot be accounted for without acknowledging the cognitive relevance of the material qualities of things. While, for instance, we can remember by materially engaging both with a notebook and a cup, the what and how of the remembering is different in each case. Remembering through a notebook usually entails writing down something on its pages. Conversely, remembering through a cup involves the ability to use it dexterously (i.e. to drink from it). In these cases, the qualities of material engagement differ as a function of the material properties of the things involved. This does not mean that the material properties of things predetermine our engagement with things. As mentioned, sociomaterial practices encourage specific ways of engaging with material culture (e.g. canonical uses of objects), thus creating action boundaries. However, things could never become meaningful and endure over time as memories for engagement if it were not for their material qualities. As Nelson nicely puts it, 'the form versus function controversy is largely irrelevant (. . .) form is usually determined by function, just as function depends largely on form' (Nelson, 1983: 392). Summarising, things endure not because we form an internal copy of them but because they remain meaningful for material engagement over time. This is due to the processes by which the materiality of things becomes intertwined with the specific sociomaterial practices in which subjects take part, over different timescales.

Some concluding remarks
In this article, we developed an alternative perspective to mainstream memory theories. To this end, we first analysed the difficulties of orthodox views that claim that remembering is a skullbound activity that occurs through the storage, manipulation and retrieval of representations in the head. Then, we explained why perspectives that consider memory as an extended process are reductionist and do not really do justice to the active role things have for cognition. Based on this, we argued the need for a theoretical alternative that does not subordinate material culture to internal mental processing.
Based on the enactive, ecological, extended, semiotic and situated perspective of MET, we proposed that remembering is a way of engaging with and through things. More specifically, we suggested that we remember whenever we update in our interaction with the world a form of engagement previously acquired through sociomaterial practices. It is worth noting that certain aspects of our proposal have significant consequences for studies on memory. First, we defined remembering as a transactional process that does not occur in the individual mind but in the space that emerges when brains, bodies and things conflate during material engagement. In this sense, we affirmed that the one who remembers is not the subject but a-subject-interacting-with-andthrough-a-thing. Defining memory in terms of interaction with and in the world collaborates with the understanding of cognition in concrete terms and is congruent with current challenges of ecological-enactive perspectives.
However, we provided arguments to consider things as full-fledged memories. Things accumulate, over different timescales, the ways in which we have materially engaged with them, according to the sociomaterial practices in which we have participated. Hence things, as material signs, are capable of both bringing forth past ways of material engagement and shaping current ones. Things are not the aboutness of cognition but active agents. They do not collaborate with cognition by providing input for internal processing: things are the very arena in which and through which cognition occurs. Finally, our proposal entails innovations for the design and implementation of empirical research. We have argued that the study of memory should account for the situations of material engagement that occur in specific sociomaterial coordinates. This implies a preference for studies that prioritise the analysis of the cognitive ecologies where remembering takes place over those that seek to achieve maximum control of variables in artificial environments (e.g. laboratory settings). In this way, our article is a defence for a situated study of cognition, which seriously considers its material, pragmatic and transactional nature.
integration between agents and artefacts (e.g. Heersmink, 2015), these contributions are somewhat aligned with the premises of Material Engagement Theory and the idea that cognition is a process of continuous becoming in which a coalition of human and non-human agents co-constitute each other (see Ihde and Malafouris, 2019;Malafouris, 2021 for a discussion on the concept of participatory mentality). Nevertheless, it should be noted that, as argued by Loughlin (2020), minimising or eliminating the parity principle could result in a distortion of the extended perspective altogether. 8. The idea that the sociomaterial coordinates of action are constitutive of cognition also applies to cognitive processes usually considered abstract and disembodied, such as imagination. Thinking, whatever its form, is always entangled with matter and the affordances of things (for a discussion, see Koukouti and Malafouris, 2020). 9. Our proposal for considering both sociocultural practices and materiality is also shared by archaeological views on memory (Borić, 2010;Mills and Walker, 2008;Olivier, 2011;Van Dyke and Alcock, 2003). Of particular interest is Jones (2003), who specifically argues that one cannot understand memory without considering the material dimension of things and the social practices associated with them. 10. Psychology classifies types of memory according to the 'different kinds of information they process and the principles by which they operate' (Squire and Dede, 2015: 3; for a discussion, see Michaelian, 2011). However, as we argue that memory is a transactional process occurring with and through things rather than the processing of internally stored information, our proposal does not concur with the mainstream classification of memory types (i.e. once memory is defined as a cognitive act involving brains, bodies and things, categories based on how information is mentally processed are no longer useful). In our view, a more productive undertaking would be to describe mnemonic activities and their sociomaterial coordinates (see Malafouris, 2013;Sutton, 2015). 11. Olivier (2011Olivier ( , 2013) offered a related perspective when arguing that the past is the materiality of the present. In his view, 'the present is not actually what is happening right now, but rather the accumulation of all past times that have been materially preserved' (p. 124). Accordingly, because they come from a variety of pasts living on in the present, things act as memories. However, in his perspective the past gets recorded within things themselves (Olivier, 2013: 121). Following a process and transactional metaphysics, we argue that the meaning that emerges from material engagement is not encoded within things but gives rise to a potential for future interactions. This potential is relational in that it pertains to the brainbody-things network and becomes actualised in each new instance of material engagement. 12. We agree with Malafouris and Koukouti (2018) that 'it is the actual engagement with clay, in the throwing, that offers the opportunity of remembrance' (p. 173) and enables the re-enactment of specific forms of material culture. 13. Archaeological studies on memory have long recognised the importance that memory has for present aims or future projects (e.g. Bradley and Williams, 1998;Joyce and Pollard, 2010;Meskell, 2008). 14. Psychology generally maintains that the self and memory are inextricably linked thanks to processes that create and maintain self-continuity (Beike et al., 2004;Neisser, 1988). Similarly, it is often assumed that subjects form representations of the self that can be updated over time (e.g. Greenwald, 1980;Tulving, 2005). In adhering to the non-representational, interactive and situated framework provided by MET and the ecological-enactive approaches, our proposal raises a question about the connection between memory and the emergence and diachronic continuity of the self (i.e. if memory is continually (re)constructed by interacting through and with things in the world, how does our sense of self emerge and endure?). In line with Malafouris (2008), we argue that the self is 'located neither inside nor outside the brain/body, but is instead constantly enacted in-between brains, bodies and things and thus irreducible to any of these three elements taken in isolation ' (p. 1997; see also Malafouris, 2015;March, 2019). In principle, then, the self would be an emergent result of the process of cognitive becoming.