Actor–Network Theory
Actor–network theory (ANT) is best understood as a relational, material-semiotic approach that emerged within science and technology studies to explain how heterogeneous associations among humans, artifacts, texts, institutions, organisms, technical systems, and infrastructures become stabilized into what later appears as “society,” “technology,” “nature,” or “fact.” Its canonical formulation is associated with Michel Callon, Bruno Latour, John Law, and related work by Madeleine Akrich at the Centre de Sociologie de l’Innovation in Paris. Callon’s sociology of translation, Law’s account of heterogeneous engineering, and Latour’s analyses of mediation, inscription, and black-boxing define ANT’s central commitments: agnosticism about what counts as an actor before inquiry, generalized symmetry across human and nonhuman actants, attention to translation and mediation, and a methodological preference for tracing associations rather than invoking “the social” as a separate explanatory substance.
ANT’s most distinctive contribution is its reconceptualization of agency. It does not treat agency as a private human capacity grounded only in intention. Instead, agency is distributed across associations and often delegated into artifacts, standards, texts, interfaces, devices, classifications, protocols, and infrastructures. This is why black boxes, inscriptions, scripts, technical objects, and institutional routines are central to ANT: they are the media through which relations become durable and action becomes repeatable. Yet ANT’s symmetry principle is methodological, not moral. It does not claim that humans and objects are equivalent in every respect. It claims that analysts should not decide in advance that only humans matter causally.
The major critiques of ANT remain familiar. Critics argue that it can become overly descriptive, politically thin, weak on macro-structures, and too willing to flatten differences between intention-bearing humans and nonhumans. Classic objections from Harry Collins and Steven Yearley, and later critiques from Andrea Whittle and André Spicer, argue that ANT risks sacrificing explanatory force and normative leverage. Defenders respond that ANT’s descriptive rigor is precisely what allows it to show how power, institutions, expertise, and infrastructures are assembled in practice. Later “ANT and after” work also addresses politics, multiplicity, ontology, and critique more directly.
Recent work shows that ANT remains active beyond classic laboratory studies. It is now used in healthcare systems research, sustainability studies, design, education policy, digital platforms, AI governance, and robotics or educational technology. A 2024 healthcare review argues that ANT helps analyze relationships among human and nonhuman actors in healthcare systems, including governance processes, interdependencies, and power dynamics. Recent sustainability literature also treats ANT as a useful framework for tracing socio-technical transitions, while newer AI-focused work uses ANT to examine how algorithmic systems, clinical AI tools, surveillance infrastructures, educational robots, and digital platforms redistribute competence, visibility, responsibility, and accountability.
ANT and enactivism converge in their rejection of strong mind–world dualism, their relational account of action, and their insistence that cognition or action cannot be understood apart from embodied, material, and environmental coupling. But they diverge sharply in scope and ontology. Enactivism is a theory of cognition grounded in autonomy, sense-making, embodiment, and organism–environment coupling. ANT is not primarily a theory of mind; it is a method and ontology for tracing socio-material assemblages. Put differently: enactivism asks how cognition is enacted by living, embodied agents; ANT asks how agency, order, facticity, and accountability are distributed and stabilized across heterogeneous networks.
The most promising bridge between ANT and enactivism is not a full synthesis but a family of adjacent approaches, especially material engagement theory, ecological-enactive cognition, and other 4E traditions. These approaches examine how minds, practices, artifacts, bodies, and environments are co-constituted through material activity. In that space, ANT contributes a method for tracing networks of mediation, while enactivism contributes an account of embodied sense-making and cognitive autonomy.
ANT’s most distinctive contribution is its reconceptualization of agency. It does not treat agency as a private human capacity grounded only in intention. Instead, agency is distributed across associations and often delegated into artifacts, standards, texts, interfaces, devices, classifications, protocols, and infrastructures. This is why black boxes, inscriptions, scripts, technical objects, and institutional routines are central to ANT: they are the media through which relations become durable and action becomes repeatable. Yet ANT’s symmetry principle is methodological, not moral. It does not claim that humans and objects are equivalent in every respect. It claims that analysts should not decide in advance that only humans matter causally.
The major critiques of ANT remain familiar. Critics argue that it can become overly descriptive, politically thin, weak on macro-structures, and too willing to flatten differences between intention-bearing humans and nonhumans. Classic objections from Harry Collins and Steven Yearley, and later critiques from Andrea Whittle and André Spicer, argue that ANT risks sacrificing explanatory force and normative leverage. Defenders respond that ANT’s descriptive rigor is precisely what allows it to show how power, institutions, expertise, and infrastructures are assembled in practice. Later “ANT and after” work also addresses politics, multiplicity, ontology, and critique more directly.
Recent work shows that ANT remains active beyond classic laboratory studies. It is now used in healthcare systems research, sustainability studies, design, education policy, digital platforms, AI governance, and robotics or educational technology. A 2024 healthcare review argues that ANT helps analyze relationships among human and nonhuman actors in healthcare systems, including governance processes, interdependencies, and power dynamics. Recent sustainability literature also treats ANT as a useful framework for tracing socio-technical transitions, while newer AI-focused work uses ANT to examine how algorithmic systems, clinical AI tools, surveillance infrastructures, educational robots, and digital platforms redistribute competence, visibility, responsibility, and accountability.
ANT and enactivism converge in their rejection of strong mind–world dualism, their relational account of action, and their insistence that cognition or action cannot be understood apart from embodied, material, and environmental coupling. But they diverge sharply in scope and ontology. Enactivism is a theory of cognition grounded in autonomy, sense-making, embodiment, and organism–environment coupling. ANT is not primarily a theory of mind; it is a method and ontology for tracing socio-material assemblages. Put differently: enactivism asks how cognition is enacted by living, embodied agents; ANT asks how agency, order, facticity, and accountability are distributed and stabilized across heterogeneous networks.
The most promising bridge between ANT and enactivism is not a full synthesis but a family of adjacent approaches, especially material engagement theory, ecological-enactive cognition, and other 4E traditions. These approaches examine how minds, practices, artifacts, bodies, and environments are co-constituted through material activity. In that space, ANT contributes a method for tracing networks of mediation, while enactivism contributes an account of embodied sense-making and cognitive autonomy.
ANT Core Concepts and Canonical Texts
What ANT IsActor–network theory (ANT) is a material-semiotic and methodological approach that examines how heterogeneous associations among humans, technologies, texts, organisms, institutions, artifacts, and infrastructures become stabilized into relatively durable arrangements such as organizations, facts, markets, devices, and institutions. Law’s classic formulation argues that social order, power, and organization are network effects produced through materially heterogeneous relations, rather than pre-given structures. Latour later radicalized this point by arguing that “the social” should not be treated as an already-existing explanatory substance. Instead, inquiry should trace the associations that compose it.
A key caution from Latour is that “network” in ANT does not mean a simple technical grid, internet network, or social-network graph. In his later reformulations, a network is a chain of transformations, mediations, and translations. Latour warned that the popular internet-era meaning of “network” as transport without deformation is almost the opposite of what ANT intended.
Key Concepts and Definitions
Actors or actants. In ANT, an actor or actant is any entity that makes a difference in the course of action. The analytic point is not that all entities are identical, but that human and nonhuman contributions to action should be traced rather than excluded in advance. Law describes agents, texts, devices, and architectures as participants in the heterogeneous networks that generate social effects. Later ANT scholarship clarifies that agency here means the capacity to affect and be affected, not necessarily consciousness, intention, or autonomy.
Networks. Networks are dynamic configurations of heterogeneous associations whose apparent stability is continuously reproduced through coordination, maintenance, and translation. They are performative and precarious rather than fixed containers. Law emphasizes that institutions, organizations, and machines are effects of ongoing ordering work, while Latour stresses that analysis should trace the transformations that hold them together.
Translation. Callon’s “sociology of translation” remains foundational to ANT. In the scallops paper, translation is the process through which actors define problems, assign roles, enroll others, and mobilize spokespeople. Callon famously describes four moments of translation: problematization, interessement, enrolment, and mobilization. Translation is therefore not simple communication; it is a process of negotiation, displacement, and alignment through which actors redefine one another’s identities, interests, and capacities for action.
Inscription and script. In the Akrich–Latour line of work, inscription refers to the way programs of action, anticipated uses, competences, and constraints are written into artifacts, standards, diagrams, documents, or interfaces. Akrich’s “de-scription” of technical objects and Latour’s discussion of delegated action in mundane artifacts both show how artifacts can embody, stabilize, and displace social intentions into material form.
Black boxes. A black box is a stabilized association whose internal complexity disappears from ordinary view once it works reliably. Science in Action proposes studying science and technology “before the box closes,” while Law explains how heterogeneous networks come to appear as single-point actors. Breakdown, controversy, or failure can reopen the black box and make its internal complexity visible again.
Generalized symmetry. Callon’s principle of generalized symmetry requires analysts to explain competing claims in the same terms and to avoid a priori divisions between the social and the natural. This principle is one of ANT’s most controversial moves because it suspends the assumption that humans should receive exclusive explanatory privilege.
Obligatory passage point. During translation, actors often attempt to position themselves as indispensable intermediaries through which other actors must pass. By establishing an obligatory passage point, a network gains coordination and stability while also redistributing power.
Mediators and intermediaries. Latour distinguishes between intermediaries and mediators. Intermediaries transport meaning or force without significant transformation, whereas mediators transform, translate, distort, or amplify what passes through them. ANT generally treats actors, objects, texts, and technologies as mediators rather than passive conduits.
Canonical Texts
The core ANT literature is usually associated with Michel Callon, Bruno Latour, John Law, and Madeleine Akrich. Callon’s 1986 scallops paper provides the canonical account of translation. Latour’s Science in Action develops the analysis of black boxes, facts, laboratories, and technical networks. Law’s 1992 account of heterogeneous engineering explains how social order and power are generated through materially diverse relations. Akrich’s work on technical objects develops the concepts of script and inscription. Latour’s later Reassembling the Social reformulates ANT as a method for tracing associations rather than a theory that explains society from above.
What ANT IsActor–network theory (ANT) is a material-semiotic and methodological approach that examines how heterogeneous associations among humans, technologies, texts, organisms, institutions, artifacts, and infrastructures become stabilized into relatively durable arrangements such as organizations, facts, markets, devices, and institutions. Law’s classic formulation argues that social order, power, and organization are network effects produced through materially heterogeneous relations, rather than pre-given structures. Latour later radicalized this point by arguing that “the social” should not be treated as an already-existing explanatory substance. Instead, inquiry should trace the associations that compose it.
A key caution from Latour is that “network” in ANT does not mean a simple technical grid, internet network, or social-network graph. In his later reformulations, a network is a chain of transformations, mediations, and translations. Latour warned that the popular internet-era meaning of “network” as transport without deformation is almost the opposite of what ANT intended.
Key Concepts and Definitions
Actors or actants. In ANT, an actor or actant is any entity that makes a difference in the course of action. The analytic point is not that all entities are identical, but that human and nonhuman contributions to action should be traced rather than excluded in advance. Law describes agents, texts, devices, and architectures as participants in the heterogeneous networks that generate social effects. Later ANT scholarship clarifies that agency here means the capacity to affect and be affected, not necessarily consciousness, intention, or autonomy.
Networks. Networks are dynamic configurations of heterogeneous associations whose apparent stability is continuously reproduced through coordination, maintenance, and translation. They are performative and precarious rather than fixed containers. Law emphasizes that institutions, organizations, and machines are effects of ongoing ordering work, while Latour stresses that analysis should trace the transformations that hold them together.
Translation. Callon’s “sociology of translation” remains foundational to ANT. In the scallops paper, translation is the process through which actors define problems, assign roles, enroll others, and mobilize spokespeople. Callon famously describes four moments of translation: problematization, interessement, enrolment, and mobilization. Translation is therefore not simple communication; it is a process of negotiation, displacement, and alignment through which actors redefine one another’s identities, interests, and capacities for action.
Inscription and script. In the Akrich–Latour line of work, inscription refers to the way programs of action, anticipated uses, competences, and constraints are written into artifacts, standards, diagrams, documents, or interfaces. Akrich’s “de-scription” of technical objects and Latour’s discussion of delegated action in mundane artifacts both show how artifacts can embody, stabilize, and displace social intentions into material form.
Black boxes. A black box is a stabilized association whose internal complexity disappears from ordinary view once it works reliably. Science in Action proposes studying science and technology “before the box closes,” while Law explains how heterogeneous networks come to appear as single-point actors. Breakdown, controversy, or failure can reopen the black box and make its internal complexity visible again.
Generalized symmetry. Callon’s principle of generalized symmetry requires analysts to explain competing claims in the same terms and to avoid a priori divisions between the social and the natural. This principle is one of ANT’s most controversial moves because it suspends the assumption that humans should receive exclusive explanatory privilege.
Obligatory passage point. During translation, actors often attempt to position themselves as indispensable intermediaries through which other actors must pass. By establishing an obligatory passage point, a network gains coordination and stability while also redistributing power.
Mediators and intermediaries. Latour distinguishes between intermediaries and mediators. Intermediaries transport meaning or force without significant transformation, whereas mediators transform, translate, distort, or amplify what passes through them. ANT generally treats actors, objects, texts, and technologies as mediators rather than passive conduits.
Canonical Texts
The core ANT literature is usually associated with Michel Callon, Bruno Latour, John Law, and Madeleine Akrich. Callon’s 1986 scallops paper provides the canonical account of translation. Latour’s Science in Action develops the analysis of black boxes, facts, laboratories, and technical networks. Law’s 1992 account of heterogeneous engineering explains how social order and power are generated through materially diverse relations. Akrich’s work on technical objects develops the concepts of script and inscription. Latour’s later Reassembling the Social reformulates ANT as a method for tracing associations rather than a theory that explains society from above.
Concise takeaway
Actor–Network Theory continues to face three enduring critiques—over‑descriptiveness, political thinness, and confusion around nonhuman agency—but recent scholarship argues that ANT is evolving beyond Latour’s original formulations, extending into political ecology, feminist STS, digital governance, and environmental crises. New empirical work in healthcare, sustainability, design, education, AI, and robotics shows ANT’s ongoing methodological utility and its gradual shift toward more reflexive, critical, and normative engagements.
Critiques, Defenses, and Recent Developments in ANT
Major critiques
1. ANT is “too descriptive”
This critique remains central. ANT’s methodological commitment to tracing associations rather than positing macro‑structures is often seen as an explanatory limitation. Critics argue that ANT underplays structural determinants, historical asymmetries, and systemic inequalities—an issue highlighted in recent scholarship examining ANT’s difficulty addressing domination and hierarchy without reintroducing the very macro‑entities it brackets.
This echoes long‑standing criticisms from the science wars, Collins & Yearley, and organizational theory: ANT maps networks well but struggles to say why certain networks matter more than others.
2. ANT is politically and normatively thin
Critics such as Whittle & Spicer argue that ANT’s descriptive stance can naturalize existing arrangements and obscure domination. Recent work reinforces this concern: ANT’s refusal to engage structural determinants can limit its ability to address political or ecological crises, making it appear insufficiently critical or normatively engaged.
Some critics go further. Stamenkovic (2025) claims ANT’s hybridism—mixing humans and nonhumans, facts and values—renders it conceptually incoherent and politically dangerous, accusing Latour of relativism and “soft obscurantism.”
3. Nonhuman agency is misunderstood or misapplied
A persistent critique is that ANT anthropomorphizes objects or collapses distinctions between intentional and non‑intentional action. Stamenkovic’s 2025 critique exemplifies this misreading, arguing that ANT absurdly attributes interests or intentions to bacteria or rivers.
ANT scholars counter that agency is not intention but distributed capacity to make a difference within a network. Still, this remains a point where ANT is frequently misinterpreted.
Main defenses
1. Description is a methodological stance, not a weakness
Latour and Callon frame ANT as a method for tracing how associations are assembled, not a theory imposing macro‑explanations. What critics call “mere description” is a refusal to smuggle in structural causes before showing how they are enacted. This defense remains central to ANT’s self‑understanding.
2. Power is not absent—it's enacted materially
Law’s work on “the mechanics of power” shows how power is distributed across documents, infrastructures, standards, and interfaces. Recent critiques acknowledge that ANT can analyze power asymmetries, but argue it must do so more explicitly. Mladenović (2025) notes that ANT’s future depends on reconciling empirical openness with critical engagement.
3. ANT has evolved “after Latour”
Blok & Jensen (2025) argue that ANT is undergoing metamorphosis: extending into feminist STS, political ecology, and more‑than‑human anthropology. They emphasize ANT’s descriptive capacities being pushed toward new ontological and ecological questions—especially under planetary distress.
Recent developments and empirical applications
Healthcare and health ITA 2024 systematic review finds ANT valuable for analyzing governance, interdependencies, and power dynamics in care delivery—especially when human and nonhuman actants (software, workflows, policy, clinicians) must be integrated. This continues ANT’s long tradition in EHR and infrastructure studies.
Sustainability and socioecological research
A 2025 review of 197 articles shows rapid growth in ANT‑based sustainability research, with thematic clusters around:
This marks ANT’s movement from micro‑case studies toward broader infrastructural and socioecological questions.
Design and digital tools
Recent design research uses ANT to show how digital tools are reconfigured within hierarchical and co‑acting design networks. ANT helps explain how design outcomes emerge from heterogeneous alignments—not just human intention.
Education and institutional policy
A 2024 study of “Maths Mastery” in England uses ANT to show how policy becomes stabilized through assemblages of documents, pedagogical devices, accountability systems, and actors. This demonstrates ANT’s utility for institutional analysis and governance.
AI governance and public safety
A 2024 drowning‑prevention case study uses ANT to analyze the heterogeneous network required to move from intention to operation in AI surveillance systems. A 2026 medical article proposes an ANT‑informed normative framework for AI governance using:
This is notable because it pushes ANT toward explicit normative reflection—addressing the critique of political thinness.
Robotics and educational technology
A 2026 study of children’s engineering learning shows how robots, classroom objects, teachers, journals, and prototypes co‑constitute learning roles. This is especially relevant for embodied cognition: material artifacts participate in pedagogical action rather than merely supporting it.
Synthesis: Where ANT stands now
ANT is at a crossroads. Recent scholarship argues:
In short: ANT is not exhausted. It is being re‑scaled, re‑politicized, and re‑ontologized.
Actor–Network Theory continues to face three enduring critiques—over‑descriptiveness, political thinness, and confusion around nonhuman agency—but recent scholarship argues that ANT is evolving beyond Latour’s original formulations, extending into political ecology, feminist STS, digital governance, and environmental crises. New empirical work in healthcare, sustainability, design, education, AI, and robotics shows ANT’s ongoing methodological utility and its gradual shift toward more reflexive, critical, and normative engagements.
Critiques, Defenses, and Recent Developments in ANT
Major critiques
1. ANT is “too descriptive”
This critique remains central. ANT’s methodological commitment to tracing associations rather than positing macro‑structures is often seen as an explanatory limitation. Critics argue that ANT underplays structural determinants, historical asymmetries, and systemic inequalities—an issue highlighted in recent scholarship examining ANT’s difficulty addressing domination and hierarchy without reintroducing the very macro‑entities it brackets.
This echoes long‑standing criticisms from the science wars, Collins & Yearley, and organizational theory: ANT maps networks well but struggles to say why certain networks matter more than others.
2. ANT is politically and normatively thin
Critics such as Whittle & Spicer argue that ANT’s descriptive stance can naturalize existing arrangements and obscure domination. Recent work reinforces this concern: ANT’s refusal to engage structural determinants can limit its ability to address political or ecological crises, making it appear insufficiently critical or normatively engaged.
Some critics go further. Stamenkovic (2025) claims ANT’s hybridism—mixing humans and nonhumans, facts and values—renders it conceptually incoherent and politically dangerous, accusing Latour of relativism and “soft obscurantism.”
3. Nonhuman agency is misunderstood or misapplied
A persistent critique is that ANT anthropomorphizes objects or collapses distinctions between intentional and non‑intentional action. Stamenkovic’s 2025 critique exemplifies this misreading, arguing that ANT absurdly attributes interests or intentions to bacteria or rivers.
ANT scholars counter that agency is not intention but distributed capacity to make a difference within a network. Still, this remains a point where ANT is frequently misinterpreted.
Main defenses
1. Description is a methodological stance, not a weakness
Latour and Callon frame ANT as a method for tracing how associations are assembled, not a theory imposing macro‑explanations. What critics call “mere description” is a refusal to smuggle in structural causes before showing how they are enacted. This defense remains central to ANT’s self‑understanding.
2. Power is not absent—it's enacted materially
Law’s work on “the mechanics of power” shows how power is distributed across documents, infrastructures, standards, and interfaces. Recent critiques acknowledge that ANT can analyze power asymmetries, but argue it must do so more explicitly. Mladenović (2025) notes that ANT’s future depends on reconciling empirical openness with critical engagement.
3. ANT has evolved “after Latour”
Blok & Jensen (2025) argue that ANT is undergoing metamorphosis: extending into feminist STS, political ecology, and more‑than‑human anthropology. They emphasize ANT’s descriptive capacities being pushed toward new ontological and ecological questions—especially under planetary distress.
Recent developments and empirical applications
Healthcare and health ITA 2024 systematic review finds ANT valuable for analyzing governance, interdependencies, and power dynamics in care delivery—especially when human and nonhuman actants (software, workflows, policy, clinicians) must be integrated. This continues ANT’s long tradition in EHR and infrastructure studies.
Sustainability and socioecological research
A 2025 review of 197 articles shows rapid growth in ANT‑based sustainability research, with thematic clusters around:
- urbanization
- practices and tools
- transitions
- industry
This marks ANT’s movement from micro‑case studies toward broader infrastructural and socioecological questions.
Design and digital tools
Recent design research uses ANT to show how digital tools are reconfigured within hierarchical and co‑acting design networks. ANT helps explain how design outcomes emerge from heterogeneous alignments—not just human intention.
Education and institutional policy
A 2024 study of “Maths Mastery” in England uses ANT to show how policy becomes stabilized through assemblages of documents, pedagogical devices, accountability systems, and actors. This demonstrates ANT’s utility for institutional analysis and governance.
AI governance and public safety
A 2024 drowning‑prevention case study uses ANT to analyze the heterogeneous network required to move from intention to operation in AI surveillance systems. A 2026 medical article proposes an ANT‑informed normative framework for AI governance using:
- translation
- composition
- reversible black‑boxing
- delegation
This is notable because it pushes ANT toward explicit normative reflection—addressing the critique of political thinness.
Robotics and educational technology
A 2026 study of children’s engineering learning shows how robots, classroom objects, teachers, journals, and prototypes co‑constitute learning roles. This is especially relevant for embodied cognition: material artifacts participate in pedagogical action rather than merely supporting it.
Synthesis: Where ANT stands now
ANT is at a crossroads. Recent scholarship argues:
- ANT must evolve beyond pure descriptivism to address structural inequality, ecological crises, and digital governance.
- ANT is already metamorphosing through feminist STS, political ecology, and more‑than‑human anthropology.
- ANT’s empirical vitality remains strong, especially in healthcare, sustainability, design, education, AI, and robotics.
- The main critiques persist, but ANT’s defenders increasingly acknowledge the need for reflexivity and normative engagement.
In short: ANT is not exhausted. It is being re‑scaled, re‑politicized, and re‑ontologized.
ANT, Enactivism, and 4E Cognition
What enactivism brings to the comparisonEnactivism is a theory of cognition, not a sociology of associations. In canonical texts, enactivism argues that cognition is not best understood as internal symbol manipulation but as embodied, situated, and world-involving activity. The Embodied Mind helped establish the enactive orientation within embodied cognition; Action in Perception formulated the sensorimotor idea that perception is something organisms do; Radicalizing Enactivism rejected contentful representation for basic minds; and participatory sense-making extended enactivism into the social domain by arguing that interaction can take on a form of autonomy.
Within the wider 4E family, embodied cognition is a broad anti-computationalist research program, and enactivism is one of its strongest strands. The SEP entry on embodied cognition explicitly notes distinct enactivist strands, while externalist/extended-mind discussions supply the “extended” component in 4E cognition.
Points of convergenceANT and enactivism converge most clearly on anti-dualism. Both reject neat separations between organism/subject and world/object, although for different reasons. ANT dissolves the nature/society and human/object split by tracing hybrid assemblages; enactivism dissolves the mind/world split by treating cognition as enacted through embodied coupling.
They also converge on relationality. ANT insists that actors are effects of relations within networks; enactivism insists that cognitive significance arises through active organism–environment coupling. In social enactivism, interaction can acquire autonomy; in ANT, networks can stabilize and act through aligned mediations. Neither perspective is satisfied by an isolated, internalist subject.
A third convergence is material participation. ANT foregrounds artifacts, inscriptions, devices, and infrastructures as mediators of action. Enactivism, especially when widened toward 4E and material engagement theory, also treats tools and material practices as constitutive rather than merely causal supports. Recent bridge work in material engagement theory explicitly states that it draws on both 4E accounts of cognition and actor-network theory to argue that mind is made up in part by material engagement.
Points of divergence
The deepest divergence concerns what is being explained. Enactivism explains cognition, perception, agency, and meaning-making. ANT explains how socio-material arrangements become durable and consequential. ANT is therefore not a competing cognitive architecture, and it should not be read as a theory of perception or consciousness. Latour’s mature work frames ANT as a method for tracing associations, while enactivism is explicitly a cognitive-scientific and philosophical account of mind.
A second divergence concerns agency. For enactivism, agency is rooted in autonomy, adaptivity, and living organization; for social enactivism, interaction may acquire a derivative autonomy. For ANT, agency is distributed across heterogeneous associations and may be delegated into nonliving artifacts, standards, and protocols. ANT therefore offers a broader social ontology of action, whereas enactivism offers a more constrained account centered on living systems and sense-making.
A third divergence concerns representation. Enactivism is largely motivated by dissatisfaction with representationalist cognitive science, especially in radical enactivism. ANT, by contrast, is not primarily anti-representational in that cognitive sense; instead, it studies how inscriptions, diagrams, datasets, and texts circulate and stabilize as representational media within networks. Thus both are suspicious of internalist mentalism, but they use the term “representation” very differently.
A fourth divergence concerns embodiment. Embodiment is constitutive in enactivism: bodily organization and sensorimotor skills are central. ANT recognizes bodies, but embodiment is not its privileged explanatory anchor; bodies are one among many materially heterogeneous actants. In practice, this makes ANT stronger on institutions and infrastructures, and enactivism stronger on perception, autonomy, and lived experience.
Comparative table
The table below synthesizes canonical ANT texts by Callon, Law, and Latour with standard enactivist texts by Varela, Thompson, Rosch, Noë, Hutto, Myin, and De Jaegher and Di Paolo.
Bottom-line relation
The strongest analytical conclusion is this: ANT and enactivism are not rivals, but differently scaled relational frameworks. ANT works best when the question is how humans, artifacts, institutions, and texts co-produce orderly action. Enactivism works best when the question is how embodied living systems enact meaningful worlds. Bridge work is most plausible where artifacts and material culture are not merely external supports but participate in cognition, as in material engagement theory and related 4E scholarship.
The strongest analytical conclusion is this: ANT and enactivism are not rivals, but differently scaled relational frameworks. ANT works best when the question is how humans, artifacts, institutions, and texts co-produce orderly action. Enactivism works best when the question is how embodied living systems enact meaningful worlds. Bridge work is most plausible where artifacts and material culture are not merely external supports but participate in cognition, as in material engagement theory and related 4E scholarship.
Recommended Sources and Open Questions
Recommended primary sources
Recommended secondary scholarship
The following secondary works are especially useful for a concise analytical report and poster bibliography.
Recommended enactivism and 4E sources for the comparison
Open questions and limitations
The strongest limitation in the current literature is that direct ANT–enactivism comparison is still relatively sparse. Most of the robust bridge work proceeds indirectly through material engagement theory, sociomateriality, archaeology, or ecological-enactive approaches rather than through direct head-to-head comparison between Latour/Callon/Law and enactivist cognitive science. That means the comparison in this report is mostly a high-confidence synthesis of primary commitments, supplemented by explicit bridge scholarship where available.
A second limitation is access. Some canonical ANT texts now circulate more reliably through author archives, repositories, or publisher metadata than through open full text. Where official publisher pages were unavailable or restricted, I prioritized author-hosted PDFs, institutional repositories, or formal DOI records.
Recommended primary sources
- Callon, Michel. “Some Elements of a Sociology of Translation: Domestication of the Scallops and the Fishermen of St Brieuc Bay.” The Sociological Review 32(S1), 1984. Best starting point for translation. Official page above.
- Law, John. “Notes on the Theory of the Actor-Network: Ordering, Strategy, and Heterogeneity.” Systems Practice 5, 1992. Best concise conceptual overview.
- Latour, Bruno. Science in Action. Harvard University Press, 1987. Best source for black boxes and following actors.
- Akrich, Madeleine. “The De-scription of Technical Objects.” In Shaping Technology/Building Society, 1992. Best source for script and inscription.
- Latour, Bruno. “Where Are the Missing Masses?” In Shaping Technology/Building Society, 1992. Best source for delegated action in artifacts.
- Latour, Bruno. We Have Never Been Modern. Harvard University Press, 1993. Best source for symmetry and hybridity.
- Latour, Bruno. “On actor-network theory: A few clarifications plus more than a few complications.” Soziale Welt 47, 1996. Best source for what ANT is not.
- Latour, Bruno. “On Recalling ANT.” The Sociological Review 47(S1), 1999. Best transitional reflection.
- Latour, Bruno. Reassembling the Social. Oxford University Press, 2005. Best mature statement of ANT.
- Law, John. “Actor-network theory and material semiotics.” In The New Blackwell Companion to Social Theory, 2008. Best later conceptual consolidation.
Recommended secondary scholarship
The following secondary works are especially useful for a concise analytical report and poster bibliography.
- Akrich, Madeleine. “Actor Network Theory, Bruno Latour, and the CSI.” Social Studies of Science 53(2), 2023. Best short historical reflection on ANT’s origins.
- Baiocchi, Gianpaolo, Diana Graizbord, and Michael Rodríguez-Muñiz. “Actor-Network Theory and the Ethnographic Imagination.” Qualitative Sociology 36, 2013. Excellent clarification of agency as affect/being-affected rather than intention.
- Sayes, Edwin. “Actor–Network Theory and Methodology: Just What Does It Mean to Say That Nonhumans Have Agency?” Social Studies of Science 44, 2014. Best focused defense and clarification of nonhuman agency.
- Whittle, Andrea, and André Spicer. “Is Actor Network Theory Critique?” Organization Studies 29, 2008. Essential critique on normativity and politics.
- Alcadipani, Rafael, and John Hassard. “Actor-Network Theory, Organizations and Critique.” Organization 17, 2010. Useful defense-and-extension of critique within ANT.
- Cresswell, Kathrin M., et al. “Actor-Network Theory and Its Role in Understanding the Implementation of Information Technology Developments in Healthcare.” BMC Medical Informatics and Decision Making 10, 2010. Strong empirical methods piece for health IT.
- Ryan, Tadhg, et al. “The Integration of Human and Non-human Actors to Advance Healthcare Delivery: Unpacking the Role of Actor-network Theory.” BMC Health Services Research 24, 2024. Good recent systematic review.
- Aka, Kadia Georges. “Actor-network theory-based applications in sustainability: A systematic literature review.” Cleaner Production Letters 8, 2025. Best recent synthesis of application growth.
- Noesgaard, Maria S., et al. “Making Society a Safer Place: Revisiting Actor-Network Theory in the Age of AI.” ICIS 2024 proceedings. Good AI/public-safety application.
- Hahn, Matthias. “An actor-network-theory informed normative framework for the governance of AI-systems in medicine.” Humanities and Social Sciences Communications, 2026. Best example of a normative turn in ANT.
- Jung, Sung Eun. “Young Children’s Engineering Learning with Educational Robots: An Actor-Network Theory Approach.” Journal of Science Education and Technology 35(3), 2026. Strong robotics/education case.
- Kee, H. “Between social cognition and material engagement: the cooperative body hypothesis.” Phenomenology and the Cognitive Sciences, 2025. Useful bridge text across ANT and 4E/material engagement.
Recommended enactivism and 4E sources for the comparison
- Varela, Thompson, and Rosch. The Embodied Mind. MIT Press.
- Noë. Action in Perception. MIT Press.
- Hutto and Myin. Radicalizing Enactivism. MIT Press.
- De Jaegher and Di Paolo. “Participatory sense-making.” Phenomenology and the Cognitive Sciences 6, 2007.
- Shapiro. “Embodied Cognition.” Stanford Encyclopedia of Philosophy.
- Rowlands. “Externalism About the Mind.” Stanford Encyclopedia of Philosophy.
Open questions and limitations
The strongest limitation in the current literature is that direct ANT–enactivism comparison is still relatively sparse. Most of the robust bridge work proceeds indirectly through material engagement theory, sociomateriality, archaeology, or ecological-enactive approaches rather than through direct head-to-head comparison between Latour/Callon/Law and enactivist cognitive science. That means the comparison in this report is mostly a high-confidence synthesis of primary commitments, supplemented by explicit bridge scholarship where available.
A second limitation is access. Some canonical ANT texts now circulate more reliably through author archives, repositories, or publisher metadata than through open full text. Where official publisher pages were unavailable or restricted, I prioritized author-hosted PDFs, institutional repositories, or formal DOI records.