On the importance of things: a relational approach to agency

Malafouris’ book presents a new theory of the part played by things in cognitive processes. Malafouris conceives of the interactions between persons and things as forming a “continuum” in the spirit of radical enactivism (Hutto 2005) as well as extended mind (Clark and Chalmers 1988), distributed cognition (Hutchins 1991; Hutchins 2008), and similarly approaches such as action-network theory (Latour 2005). The focus of this review essay is on the relational ontology of human beings and things. In this regard, Malafouris suggests that agency is not the property of humans or things but rather it is the emerging property of material engagement (Malafouris 2013: 18). He proposes that the notion of agency should be understood as a materially engaged cognitive process. The book is structured in three parts: “Cognition and material culture;” “Outline of a theory of material engagement;” and “Marking the mental: Where brain, body, and culture conflate.”

Central to Malafouris’ theory is a critique of “cognitivism” — which he describes as the idea that the mind is simply a computer located within the brain. Malafouris holds that artefacts are not external products of human thought but rather integral parts of the thinking process. The thesis is that an ontological unity between cognition and material culture drives cognitive evolution. There is a phenomenological osmosis with things (ibid.: 7). According to Malafouris, there is an absence in the literature of any real concern with the semiotic ontology of material culture. A “return to the things” (ibid.: 133) is necessary to resist the idea of the human mind perceived as a disembodied information-processing ghost captured in the labs of the neurosciences.

The book deploys a direct critique of the Cartesian worldview immanent in cognitivism, and representational views of mind (internalist approaches). The dualistic conception of the relationship between mind and matter is said to be a category mistake. Representation in cognitivism is “the mechanism by which we could feed our cognitive apparatus with facts and information of the external world” (ibid.: 26). Criticizing the notion of internal representation, Malafouris claims “the only representations with any substantial or real implications for human cognition are to be found outside the head” (ibid.: 31).

In contrast to this, Malafouris proposes material engagement theory as an attempt to “restate the problem of interaction between cognition and material culture in a more productive manner by placing it upon a new relational ontological foundation” (ibid.: 35) and thus to abandon the logic of boundaries and delimiting lines, opting for a continuum of mediated action between things and cognition. The compatibility with the distributed cognition approach (Hutchins 2008) is evident, in the sense of those interactions being the process of thinking itself. Interactions with artefacts not only influence cognition; they are a constitutive part of it. According to Malafouris, concepts of vital materiality (Bennett 2015) and metaplasticity (Parisi 2019) are convergent with material engagement theory. The former refers to the need to “take things seriously” — that is, material culture is not “merely the backdrop against which human cognition takes shape. Things mediate, actively shape, and constitute our ways of being in the world” (Malafouris 2013: 44). Metaplasticity, on the other hand, refers to experiential activities affecting the organization of the brain and the way it works, which account for the payoffs from learning and training. For Malafouris, metaplasticity “may be what makes change and alterability the natural state of a human intelligence that is unlike anything we see in other animals” (ibid.: 46). Evolution of cognition is a “dynamic co-evolutionary process of deep enculturation and material engagement” (ibid.: 45).

Malafouris points out two problems that embodied cognition seeks to resolve. First, what is the exact nature of the pre-conceptual experiences that define how humans make sense of the world? Second, how can abstract and higher-level cognition be explained in terms of such experiences? The first question is answered by claiming that pre-conceptual experience consists mainly of directly meaningful universal schemata that exist prior to categorization. The second problem is explained by the thesis that we import pre-conceptual structures by way of metaphorical and integrative conceptual mappings to deal with higher-level cognition. The problem with this approach, according to Malafouris, is that it is “simply an expansion of the ontological boundaries of the res cogitans rather than the dissolution of those boundaries altogether” (ibid.: 65). Embodied cognition is thus a kind of “embodied cognitivism” while material reality remains external and epiphenomenal.

In the same vein, Malafouris argues that symmetric functional isomorphism is misleading since the way that artefacts operate is different from the way the brain works; they are complementary but not coextensive, and together they make up the cognitive process as a whole. As such, a cognitive process is a relationship between elements without placing any importance on the spatial co-location of the elements (inside or outside the brain). Malafouris asserts that it is difficult to draw the boundaries between internal and external parts of the cognitive system and his argumentation tends to equate a cognitive system with the notion of agency.

Nevertheless, I would contend that a distinction should be made between a cognitive system and a cognitive process, and between a cognitive process and the notion of agency. According to Gabbay and Woods (2003: 7), a cognitive system is “a three-tuple of a cognitive agent, cognitive resources, and cognitive tasks performed dynamically in real time.” Pursuing this line of thought, I suggest that a cognitive process which is also a process of meaning-making (i.e. a semiotic process) occurs when agency is manifested in actions. A proper cognitive-semiotic system can only be understood as such when meaning-maker appears on the scene, establishing an asymmetrical relationship. Other kinds of systemic relations are symmetrical, for instance, an autonomous system in which relations depend on inputs and outputs could work without the interpretation of an agent. This is the case of automatized production lines. Nevertheless, the robot’s actions are meaningless until a proper agent enters the scene. Hence, meaning-making, and intentionality — including those of sedimented semiotic acts (Sonesson 2009; Sonesson 2019) — introduce a phenomenological asymmetry into the cognitive systems, which should not be compared with an action-reaction relationship.

As proposed by Mendoza-Collazos (2016: 87ff), the asymmetry of our relation to artefacts is highlighted by the need for the agent to incorporate artefacts to fulfill their purposes and make sense of their actions even if the agent is not aware, at the moment of its occurrence, of their own agency. In contrast, cause and effect relationships do not require agency, but an agent certainly is necessary to make sense of such relations.

For a proper understanding of our mental machinery as Malafouris claims, it is more accurate to enquire into how the knowledge for performing a task is propagated across the elements of a cognitive system (people, artefacts, time and space) than to investigate who is responsible for performing the task (Malafouris 2013: 67–68). This argument avoids the central question on agency, translating it from an entity to a process. Following this line of argument, Malafouris uses the examples of the hotel key and the speed bump (ibid.: 124ff), derived from actor-network theory (Latour 2005), in order to support the thesis of material agency being symmetrical. In the case of the room keys, the large, cumbersome weights attached to the keys are supposed by the hotel manager to prevent customers from taking away the keys. The speed bump, meanwhile, is not only a reminder to slow down; it is, according to Malafouris, a sign and, more specifically, a moral agent stating: “Pass over me at a speed that will allow your back and your car’s suspension not to suffer any damage” (ibid.: 124). In a similar vein, Manar Hammad (1990: 60ff) presents the wall as being less a physical obstacle than a dissuasive device.

These examples overlook the original intention of the creator and seem to take for granted the existence of things while the cumbersome shape of the hotel room key and the strategic location of a speed bump were designed by someone to fulfill these purposes. Design is the conception and planning of the artificial (Buchanan 1992: 14). This pursuit highlights the teleology of human actions and reveals the importance of studying the nature of the agent in order to reach a complete understanding of the phenomenon. Artefacts do not simply appear in the world ex nihilo, they are conceived based on prior intentions and needs. The process of making is usually an a posteriori stage of design. Earlier phases of this process may include sketches and external representations of the intended artefact or it may be the result of direct engagement with the material. Malafouris only attends to the latter case, suggesting that the agent shares the creative act of invention with the material, as epitomized by the example of the potter’s wheel (Malafouris 2013: ch. 9).

Further, Malafouris explains how we are engaged with the world and how this world can be used as a cognitive artefact — that is, something that enhances our cognitive skills by supporting choice, perception, and problem solving. Artefacts are not only externalized information but “also the actual process of information” (ibid.: 73). The distinction between “internalism” and “externalism” is not at issue. External conditions clearly form part of the process of cognition itself (not only the content of mental states). This we can agree on, although there are disagreements on other issues presented here.

The book presents two fundamental theses on distributed cognition: What is outside the brain is not necessarily outside the mind; and cognition is not simply a matter of internal representation. The mind is not a storehouse of internal representations. Cognition also involves transformation and propagation of “internal representational states.” Representation is not only intracranial; it can also be outside or external. With the notion of enactive sign, Malafouris tries to move beyond representations, dealing with the concept of sign and what counts for a sign. He argues that “material culture is the capacity of material things to operate as signs” (ibid.: 89). The important question is what a material sign means, not what a material sign is. Material signs do not have communicational or representational functions, and they operate by means of an enactive logic: “meaning is not the product of representation but the product of a process of conceptual integration between conceptual and material domains” (ibid.: 90).

This requires some clarification. A representational function is performed only by means of signs. This is to say, only external representations correspond to proper signs (Sonesson 2015; Niño 2015; Thompson 2007). Mental states do not have the status of signs; they are not representations at all. At the best they are presentifications in the sense of Sonesson (Sonesson 2015). On the other hand, practical actions (without representations) are a kind of enactments. This way of obtaining meaning is not related to signs and has its own logic for operating. Although Malafouris wants to convey the same hypothesis, he is trapped in the oxymoron of “enactive sign” (added to the oxymoron of “material agency”). Any semiotic item (such as a hammer or a picture) can act as a sign or have only an enactive use.

Malafouris turns to Peirce to avoid the representational character of signs. The question is how a sign emerges and acquires symbolic force. My answer to this question involves highlighting the role of agents. Agents assign the status of signs and the item can change its semiotic function according to the purpose of these agents (Niño 2015).

Malafouris deploys arguments against the exclusiveness of human agency, based on the impossibility of disentangling humans from material things. He repeats that, while agency and intentionality “may not be properties of things, they are not properties of humans either; they are the properties of material engagement” (Malafouris 2013: 18, 119). However, we think material engagement is only one condition among others (Mendoza-Collazos 2016; Niño 2015) for producing and obtaining meaning from the world. The capability of an agent to act (their agency) should not be overlooked.

Malafouris argues that there are deficiencies in the concept of agency, and that the category of material culture “remains neglected and theoretically marginalized even within philosophical frameworks” (Malafouris 2013: 120). He rejects the approach to the concept of agency within the framework of “human individuals of the modern Western type” as strongly contaminated by the Cartesian lens, but at the same time he considers “too obvious” that intentionality is a necessary conditions for agency (ibid.: 121). He asserts that to understand or try to define the fuzzy notion of agency is unnecessary (ibid.: 147). In order to put the emphasis on the relational approach, he proposes an argumentum ad ignorantiam fallacy: We should not opt to reject the study of the concept of agency because there are deficiencies in the explanations of what agency is. On the contrary, the fuzziness of the concept invites us to reach a greater understanding of how agency operates, what its nature is.

In order to understand the social and cognitive life of things, Malafouris states that “it is more sensible and productive to treat material things as agents (and be wrong) than to deny their agency (and be wrong)” (ibid.: 134). Thus, material agents should not be treated as persons but as “fetishes”. Methodological fetishism (using artefacts as fetishes) is the conceptual apparatus proposed by Malafouris to differentiate the agency of artefacts from human agency. Artefacts are not extensions of human-body parts (ibid.: 135). Moreover, artefacts are not homologous to human beings, that is, they do not interact like humans, but rather have their own way to operate. Humans endow artefacts with material agency by means of fetishization, but artefacts do not have human-like agency. In other words, artefacts act like artefacts with material agency but not like human beings with human agency. In spite of such apposite observations, Malafouris most of the times seems to reason as if artefacts had agency in the same sense as animate beings.

Malafouris is interested in the cognitive process underlying the generation of fetishes, the process of “fetishization” which takes place in four steps: First, concretization when an abstract concept is objectified into an artefact; second, animation when “interactions between persons and fetishes resemble interactions between persons rather than interactions between persons and things” (ibid.: 133); third, conflation of signifier and signified, thus, object operating as causative agents; and fourth, ambiguous conceptualization of power, that is, ambiguity of control (people controlling objects or objects controlling people).

In this context, he introduces the Peircean notion of abduction in the following terms: “the properties of fetishism are abducted and projected into the general domain of material culture and used as a comparative reference point for detecting the agency of things” (ibid.: 134). The business of abduction is to bring something unfamiliar into a familiar frame in order to make it comprehensible. I would add that Peirce’s abduction is a special kind of inference (Fann 1970), which is a key evidence of the manifestation of human agency.

Malafouris presents three key concepts derived from Searle (1979, 1983): Prior intentions, intention-in-action, and background. Ignoring the notion of prior intentions, he prefers to focus on intention-in-action, using the concept to propose rethinking the nature of agency from “categorical” to “relational”. In short, agency is to be understood as emergent product, derived from the relational ontology of material engagement.

In contrast, I suggest that a better understanding of our relational ontology with things can be obtained by positing a distinction between agency and meaning. Agency is the capability of acting, signifying, producing, and obtaining meaning, and this capability is exclusive to living organisms. Meaning, on the other hand, is the outcome that emerges based on our agentive relationship with the environment, and it depends on a complex network of conditions (Mendoza-Collazos 2016; Niño 2015), one of which is material engagement. Agency is thus an exclusive capability of proper agents, not a process or an emergent property. Note that this does not imply falling into Cartesianism or denying a relational approach between agents and the world. Agentive semiotics (as conceived by Niño 2015; cf. Mendoza-Collazos 2016), for instance, is far from being an anthropocentric or a Cartesian conception of the way agency operates. The exclusiveness of agency within living organisms could in fact enhance the force of Malafouris’ relational approach, explaining the key role of material culture without the need for resorting to methodological fetishism.

The best definition of what it means to be human, according to Malafouris, is homo faber, arguing that intelligence is the faculty of manufacturing artificial objects, especially tools for making tools. Human tool usage is the prosthetic gesture par excellence (Malafouris 2013: 193). As this point, the intriguing enigma of hand axe symmetry enters the equation. Was this symmetry planned and preconceived? Malafouris opts to explain it as a result of the process of manufacturing, compelled by the conditions of matter. This is indeed a coherent and relevant explanation of one part of a typical design process. To explore the proclivities of materials in order to find solutions via intention-in-action is a customary practice in design processes. Nevertheless, design implies also a prior need; ergo it is a teleological action. In the case of symmetrical hand axes, while involved in a typical exploration of materials, the designer also intends to accomplish his or her purpose, even if the process of design is fused with the process of manufacturing. There is evidence that this was already the case for ancient toolmakers (Mendoza-Collazos et al. 2020).

According to Malafouris, marks, engravings, and drawings found in archaeological records are not representational intents but the result of an enactive attempt to trace lines. Pictures respond to an “enactive logic” derived from an “active sensorimotor engagement” (quoting Noë 2004). Thus, these drawings are “a form of acting in the world”, rather than a way of representing the world (Malafouris 2013: 203). This idea is based on the differentiation between the experience of pictures and the awareness of pictures as representations. The process of perception as such is not ordinarily made part of our awareness, so it may not have an important effect on our immediate phenomenological perception of the world. This proposal is consistent with the perspective of agentive semiotics (Mendoza-Collazos 2016; Niño 2015), as well as that of cognitive semiotics (Sonesson 2001: 3, 2009; Zlatev 2015). It is indeed plausible that practical actions and intentions-in-action supported the emergence of human representational capacity, as we argue in our paper on evolution of design (Mendoza-Collazos et al. 2020).

While criticizing “internalist” approaches to creativity because they contend that the creative process is something that exists inside the brain, Malafouris is also critical of “externalist” approaches that concede the influence of matter, but still accept that the ultimate source of creativity resides in the brain. In contrast, he claims that the creative process is “a binding of materials — a dynamic flow of the organic into the inorganic that can be understood as a new or ‘surprising’ blend of ingredients that act or be acted upon” (Malafouris 2013: 213). His aim is to demonstrate that this activity demands neural activation but “few if any internal representations.” And again, we would argue that while a productive part of the design process is based on material engagement, the creator has to have a prior intention (not necessarily a representation, but the intention of fulfilling a need) to produce something, even if the final product is then obtained by serendipity or material processing and constraints (Mendoza-Collazos et al. 2020).

Malafouris makes a distinction between agency and the sense of agency: The agentive capacity is common to any living organism, but the sense of agency is distinctively human (Malafouris 2013: 214). Nonhuman animals as well as artefacts lack the sense of authorship. This claim, however, overlooks the difference between living organisms and material things. Arguably, this is misleading if agency is an intrinsic feature of all living organisms, to the exclusion of any other items, in spite of being manifested in different grades of intentionality, animation (movement) and attention (Niño 2015; Zlatev 2003).

The force of material engagement theory is not disminished when we accept that living organisms have the exclusive role of agents. We act upon artefacts and engage with them, but we do not interact with them in the same way as with other agents. Since meaning can be understood as the result of engagement of agents with things or with others in order to accomplish specific purposes, there is no meaning on the side of artefacts themselves, but only on the side of agents proper. The function of things only makes sense to the living beings within a mediated interaction, or contributing to fulfilling the purpose of agents. Therefore, the study of the meaning-maker, with his/her capability to act — that is, his/her agency — is still important for a thorough understanding of the role of our material engagement.

In conclusion, I would like to present cognitive semiotics version of some of the theses of material engagement theory:

1. Meaning, not agency emerges when an agent act in the world. Agents discover affordances in characteristics of the material world; those affordances do not exist per se, but as emergent product of agent-world interaction.

2. Artefacts are human creations. They do not simply appear in the world. The creative act of transforming materials implies a purpose. Thus, agency shapes the way we think under conditions established — among other conditions — by the materiality of things.

3. Agency and intentionality are exclusive properties of living beings, and not properties of things. As stated in (1) the emergent product of our material engagement is meaning, and meaning only makes sense to agents. Inert matter exists, but it does not act.

With these re-interpretations, the contribution of Material Engagement Theory would be also useful for cognitive semiotics, for its eroding of the “tyranny of dichotomies” and highlighting the importance of materials, things, and artefacts in (human) meaning making. More broadly, it can help vindicate the notion of material culture. I hope that my comments can contribute to such an agenda.