Content Internalism and Externalism

The Problem of Content Determination

The philosophy of mental content asks what a person thinks when she believes that a glass contains water, hopes that a friend will arrive, or remembers a particular landscape. Such states exhibit intentionality: they are directed toward objects, circumstances, properties, or possibilities. Characterizing them requires more than listing sensations or physical events. A belief that a substance is potable can be true or false; a desire that the same substance be available has conditions of satisfaction; a fear that a doctor will announce a diagnosis has a determinate intentional object. The question of content determination asks which facts make those descriptions correct.

Content internalism maintains, in a specified domain, that an individual’s relevant intrinsic characteristics suffice to determine the content of the individual’s thoughts. Content externalism maintains that corresponding content can depend constitutively on facts beyond those intrinsic characteristics. The disagreement concerns individuation, the conditions under which two intentional states count as having the same content. It is therefore possible for two agents to be identical in their presently described physiology, dispositions, and subjective appearances yet differ in what their thoughts represent if the environmental or historical relations that determine reference are different. The strength of the inference depends on whether the content at issue is coarse-grained, object-involving, cognitively individuated, phenomenal, or contextually interpreted (Putnam, 1975; Burge, 1979; Lau & Deutsch, 2014).

This dispute reaches across philosophy of mind and philosophy of language because words often express thoughts and because speakers acquire concepts through linguistic practices. It also bears on epistemology: if a person’s present thought content depends on circumstances outside her reflective access, how can she know what she thinks? It bears on explanation: if two inner duplicates act identically under matching proximal stimulation, can a psychology that cites different wide contents explain their shared movements? It bears on metaphysics: what is the relation between a state’s intrinsic realization and its representational properties?

The research object here is the determination and individuation of intentional mental content. Epistemic internalism and externalism, by contrast, concern the grounds of justification or knowledge. There can be a conceptual connection between content and justification, but the disputes have different dependent variables. An agent may satisfy the same internal justification condition across two worlds while entertaining thoughts with distinct wide contents. Conversely, two speakers may share a referent while having importantly different reasons for their beliefs. Conflating the disputes turns a question about what a thought represents into a question about how responsibly a subject holds it.

The central method of this study is comparative counterfactual analysis supported by historical interpretation of original philosophical arguments. The analytic results will turn on four questions: which intentional property is compared; what internal duplication preserves; which external variable is allowed to change; and what sort of explanation the resulting content distinction is intended to provide.

Concepts, Supervenience, and the Individuation of Thought

Content is the representational contribution in virtue of which an intentional attitude concerns something or represents circumstances as obtaining. In belief attribution, a that-clause ordinarily supplies a content description: a person believes that the lake has frozen. The mental attitude and its content must be distinguished. Believing, doubting, desiring, and hoping can take related contents while retaining different psychological roles. The same person may believe that a river is polluted and hope that it becomes clean; her attitude types differ, and her contents can overlap without becoming identical.

A supervenience thesis states that one kind of property cannot differ without a difference in some specified base properties. Let I(x) denote a subject’s relevant intrinsic state and C(x) the content of a specified intentional state. A strong internalist thesis says that, for any possible x and y, equality in I guarantees equality in C. The externalist denies that this guarantee holds for at least some contents. As a counterexample, she seeks two internally duplicate subjects with different contents. The formulation is schematic because its force depends on the base: a narrow neural type, a full bodily state, an intrinsic functional organization, a subject’s experiences, or the entire historically characterized organism cannot simply be substituted without argument.

“Internal” also has several meanings. A spatial reading treats what lies inside the skin as the relevant basis. A metaphysical reading selects intrinsic properties whether their descriptions happen to involve a spatial boundary. A methodological reading seeks a psychology whose generalizations abstract from environmental variation. A phenomenological reading restricts the basis to what experience makes present. These are different constraints. The claim that a representation is neurally realized is consistent with its representational type depending upon environmental relations. The claim that a thinker experiences a stable appearance is consistent with the object represented by that thought changing with the environment.

Wide or broad content is content individuated in part by external relations; narrow content is content individuated without such relations, relative to an explicitly specified internal base. Narrow content is sometimes proposed as a genuine truth-conditional content, sometimes as a function from contexts to wide contents, sometimes as a cognitive or inferential role, and sometimes as an abstract determinant that resists direct expression in ordinary language. One should therefore resist treating narrow content as a single settled theoretical entity (Fodor, 1987; Segal, 2000; Brown, 2022).

Three relations frequently confused with content determination must be separated. Causal acquisition concerns the processes by which an individual comes to possess a representation. Metaphysical constitution concerns which facts make the representation possess a given content now. Epistemic access concerns how the individual can know that content. A history of social learning can be causally necessary for concept acquisition without every aspect of that history being constitutive of present content. Equally, a constitutive dependence need not imply that the agent can inspect the dependence.

Historical Formation: Sense, Reference, and Public Language

The modern dispute inherits a problem that predates the expressions “content internalism” and “content externalism.” Gottlob Frege’s distinction between Sinn and Bedeutung, conventionally rendered sense and reference, gives a foundational explanation of why expressions can relate to the same object yet differ in cognitive significance. An identity statement containing two distinct modes of presentation can be informative even when an identity employing one name twice is trivial to a competent thinker. The distinction creates a lasting question: what in an individual’s cognitive life explains the rational difference between co-referential expressions? Frege’s original concern was the logic and semantics of language, and present theories of narrow psychological content should not be retroactively attributed to him (Frege, 1892).

Twentieth-century approaches to meaning placed different weight on linguistic behavior, inferential role, causal history, and public practice. Quine’s Word and Object (1960) examined the empirical and social setting of linguistic interpretation, including the underdetermination of translation. A historical relation of influence should be described carefully: Quine’s project unsettles the assumption of privately fixed meanings, while his skeptical and behaviorally oriented commitments differ from later realist claims that a natural-kind term succeeds by referring to an independently constituted substance (Quine, 1960).

Kripke’s Naming and Necessity, based on lectures delivered in 1970 and published in book form in 1980 after an earlier article version, challenged theories that identify reference with descriptive material associated with a name. The causal-historical account of naming offered in the lectures made the history and transmission of a designation philosophically significant. The argument that names can be rigid designators and that some identity statements are necessary a posteriori reshaped discussions of natural-kind expressions. These semantic developments did not by themselves demonstrate that mental content is environmentally individuated, but they supplied a framework in which reference need not be exhausted by the beliefs present to an individual speaker (Kripke, 1980).

David Kaplan’s distinction between character and content in the theory of indexicals adds an important structural model. A word such as “I” has a character that yields a speaker-relative content when a context is supplied. This is a semantics of context-sensitive language, not automatically an account of every thought’s narrow content. Its philosophical importance lies in demonstrating how a stable rule and a variable contribution of context can jointly determine a proposition. Later theories of narrow content borrow the functional architecture while arguing over whether a thought possesses something sufficiently analogous to character (Kaplan, 1989).

The movement from reference in language to content in thought is the pivotal step. A speaker’s word can have a publicly determined extension without every thought associated with it having the same individuating conditions. The transition must be argued through cases in which belief attributions themselves differ between internally duplicate thinkers. That is why Putnam’s and Burge’s distinct contributions require separate examination.

Putnam: Twin Earth and Natural-Kind Reference

Hilary Putnam’s “The Meaning of ‘Meaning’” (1975) presents the celebrated Twin Earth thought experiment. Imagine a world substantially similar to Earth on which an ordinary-looking and ordinarily behaving liquid has a different molecular composition from terrestrial water. The counterpart liquid is customarily designated XYZ. An Earth speaker and a suitably matched Twin Earth speaker can occupy internally matching cognitive states while using the same sound or written word in relation to different substances. On Earth, “water” refers to H₂O; on Twin Earth, the counterpart use refers to XYZ. The agents need not know chemistry, since reference is secured partly through relations to a kind in their respective environments (Putnam, 1975, pp. 215–271).

The reasoning has three relevant components. First, an agent’s internally represented descriptions may underdetermine the chemical kind to which her word refers. Second, ordinary natural-kind reference can be anchored in a speaker’s material environment and in the community’s linguistic practices. Third, a theory making meaning wholly dependent upon current inner descriptions cannot explain the stipulated referential difference. Putnam uses this to challenge a traditional picture of meaning as a psychological state entirely within an individual’s head.

One qualification is essential. The inference from an externalist account of the meaning of a public word to externalism about every mental content requires additional assumptions. A person may deploy “water” in a belief whose wide content is determined by environmental reference, while also having a narrower cognitive characterization shared by her Twin Earth counterpart. Putnam’s argument therefore motivates content externalism, and it becomes an argument specifically about beliefs when a further account connects the reference of the used concept to the individuation of the corresponding mental attitude.

The thought experiment should not be described as an empirical discovery of a real second planet. Its purpose is modal: to compare possible environments while holding the stipulated internal variables fixed. Objections may challenge the coherence of the duplication stipulation, the move from linguistic reference to mental content, or the assumption that different extensions force the relevant mental states to count as different. A fully internalist reply may individuate the agent’s thought by her cognitive presentation, treating worldly reference as a later relational classification. A two-factor reply preserves both an inner shared determinant and an externally individuated truth-conditional content. The dispute is then relocated to the question of which content has explanatory and philosophical priority.

Natural-kind externalism also resists easy generalization. One’s thought of a demonstratively perceived individual may depend on an object-specific relation; one’s mathematical thought might involve different individuating considerations; one’s imagined unicorn may lack a real-world natural kind serving as a referential anchor. The differences are central to the research program, because successful externalism about one content family does not establish a uniform externalist semantics for all possible thoughts.

Burge: Social Anti-Individualism and the Arthritis Case

Tyler Burge’s “Individualism and the Mental” (1979) shifts the focus directly to intentional mental-state attribution. Consider a patient with several correct beliefs concerning arthritis who also mistakenly believes that arthritis has developed in his thigh. In the actual speech community, arthritis is an inflammation of joints. The patient’s error is an error within his grasp of a community-governed concept, and a physician can correct him. Burge then asks us to imagine a counterpart community in which the same word is conventionally applied more broadly, including appropriate thigh ailments. The patient is internally and proximally duplicated to the stipulated degree, while the communal conditions governing his word differ (Burge, 1979, pp. 77–79).

Burge’s central point is that incompletely understanding a term need not prevent an individual from thinking with the public concept it expresses. In ordinary attribution, the actual patient can hold the false belief that arthritis occurs in his thigh. The counterpart patient, whose community applies the word differently, is attributable a different thought. The social environment thus helps determine the type of propositional attitude, even when the subject’s internal dispositions, sensations, and relevant word use are matched in the thought experiment. The argument reaches the content of belief rather than stopping at an account of public-language extension.

The case exhibits semantic deference: an ordinary speaker may depend on socially distributed linguistic expertise while retaining genuine first-order competence in a concept. Deference is often reflected in the willingness to correct one’s application in light of authoritative linguistic or scientific practice, though it should not be reduced to explicit obedience to an expert. The philosophical thesis concerns what fixes the concept under which the belief is attributed. Its social dimension is constitutive on Burge’s reconstruction, not merely a story about how the patient originally learned a word.

An individualist has several lines of response. She may argue that the counterpart patients share one psychologically relevant content and differ only in public reports, that the attribution standard is context-sensitive, or that Burge’s example fails to rule out subtle internal cognitive differences. The strongest challenge demands a principled account of why public classifications should count as the identity conditions of a thinker’s own states. The externalist replies that the individual’s participation in a meaningful communal practice is already part of what enables the state to be correctly described as a belief about arthritis.

Burge’s argument illuminates the normative dimension of concept use. A person can be wrong about the extension of a concept she nonetheless possesses. This means that mastery need not entail complete explicit knowledge of correct application. The distinction has implications for education, specialized science, law, and medicine, where people think and speak with concepts whose authoritative criteria are shared across a community. Yet communal authority itself can be contested or transformed. Social externalism must say how meaning persists or changes in communities with linguistic disagreement and competing expert practices.

Physical Externalism, Social Externalism, and Other Forms

The label “externalism” covers several dependence claims. Physical-environment externalism highlights relations to substances, objects, organisms, or events. Social externalism highlights linguistic practices, shared institutions, and deference. Historical externalism emphasizes causal chains or histories of selection that connect present representations to their targets. Singular-content externalism identifies thoughts partly through a particular object or individual; broad kind-content externalism concerns kinds or properties whose identities are environmentally fixed. These families can overlap without collapsing into one philosophical thesis.

The diversity matters because counterexamples have different structures. An internally duplicated agent might lack an actual object of perceptual attention, belong to a community with a distinct usage, or have been formed through a different history of learning and selection. Each case raises a separate question about which relation is content constituting. Physical similarity alone never tells us whether we have successfully duplicated the relevant relational property. The selection of an external factor must be justified as part of a theory of representation rather than by stipulation alone.

A further distinction separates metaphysical externalism from semantic contextualism in the ascription of beliefs. Contextual variation in how interpreters choose words to describe states can produce different reports without a corresponding metaphysical change in the subject’s content. The externalist must therefore show why the environmental difference changes what is thought, not simply how an observer chooses to describe it. Crane (1991) presses the significance of wide content for understanding the actual explanatory relations involved in such cases. Burge’s emphasis on oblique occurrences in content clauses is designed to press precisely this point (Burge, 1979, pp. 74–77).

Externalism also allows differences between content families. The presence of an object may be constitutive for a singular thought about that very object, while other general representations can remain stable across its absence. A natural-kind concept may depend on a communal or chemical anchor even though the cognitive procedures for reasoning with it are internally implemented. The content of an indexical can be context-sensitive in a way that differs from the long-term social determination of “arthritis.” A comprehensive theory must accommodate these sources of dependence in an appropriately differentiated taxonomy.

The strongest metaphysical lesson of externalism concerns the failure of a narrowly individualistic supervenience base for certain representational properties. It does not entail that a mind extends spatially to every object represented or that the material realization of thinking is external to the organism. The relation between broad content, extended cognition, and the physical organization of thought is a further and independent research question.

Narrow Content and the Individualist Response

Narrow content is introduced to capture the cognitive organization that a subject retains across counterfactual changes in environment. For an individualist, the possibility of systematic inner duplication suggests that some representationally significant structure should remain fixed. A subject and her Twin Earth counterpart could reason similarly, make matching inferences, and exhibit parallel patterns of hesitation, expectation, and action when proximally matched. The argument is strongest when the explanatory project concerns the capacities exercised by an individual cognitive system rather than the identity of the substances or persons its thoughts concern (Fodor, 1987; Segal, 2000).

Jerry Fodor’s Psychosemantics (1987), particularly its discussion of individualism and supervenience, develops the need for content usable in cognitive explanation. One proposal construes narrow content as a function taking contexts to wide contents, so that two internally alike thinkers may share the function while yielding different wide content in different environments. This strategy preserves systematic psychological similarity and grants some force to externalist truth-conditional differences. Its outstanding problem is to specify the function independently enough to supply explanatory substance rather than merely restating the range of wide contents (Fodor, 1987, ch. 2).

Gabriel Segal’s A Slim Book about Narrow Content (2000) provides a more radical internalist response. He argues that the cognitive content relevant to psychological theory supervenes on intrinsic microstructure and challenges the purported necessity of irreducibly wide contents. His position is significant because it questions whether the two-factor compromise should be accepted. An internalist theory may assign the full burden of cognitively relevant representation to narrow content and regard wide reference as a relational property of the subject rather than a further content of the thought (Segal, 2000).

The philosophical burden on narrow content is substantial. A narrow state should distinguish thoughts that differ in their rational import, not simply label internal physical states. It should support an account of mistaken belief, conceptual conflict, learning, and compositional reasoning. If characterized as an inferential role, it faces the difficulty of determining which inferences are constitutive of a concept and which are beliefs a thinker happens to hold. If characterized phenomenologically, it must explain the intentional determinacy of unconscious or unexperienced thought. If characterized functionally, it must be sufficiently determinate to answer questions about truth conditions or explain why those questions are inappropriate to the narrow level.

The dispute is not resolved by pointing out that identical neural implementations can produce matching motor outputs. A common physical mechanism is a datum for explanation, but the issue is whether the most useful explanatory kinds must themselves be narrow intentional kinds. The individualist therefore needs an account of representational significance; the externalist needs an account of how wide properties contribute to explanation without erasing the similarities internal duplicates display.

Two-Factor Accounts and Context-to-Content Functions

Two-factor views distinguish an internally determined element of cognition from a wider element dependent on context, reference, or environment. The general idea is that a subject can preserve an internal mode of thought while the external relation supplying its object changes. Such accounts respond to the fact that Putnam-style twins appear to share many reasoning dispositions and yet, by externalist criteria, think about distinct substances. A successful theory identifies what remains stable and what differs, rather than simply attaching two labels to the same unexamined state.

A functional treatment models narrow content as a rule taking a context or circumstance of evaluation to a wide content. Kaplan’s treatment of character and content makes such an architecture intelligible for indexical language (Kaplan, 1989). The analogy has limits: it is relatively clear what linguistic rule governs “I,” while the precise rule determining the environmental content of an ordinary concept may be difficult to state without circularity. An account must specify how a thinker possesses the rule and what facts make it the relevant rule.

David Chalmers’s account in “The Components of Content” distinguishes epistemic content and subjunctive content, thereby offering a structured account of the cognitive significance of thought while retaining truth-conditional resources. The epistemic component is designed to capture what is rationally accessible in a way that can remain stable across certain environmental variations; the subjunctive component tracks how a thought applies across ways the world might be. This is an articulated philosophical proposal, not a generic restatement of internalism (Chalmers, 2002).

Two-factor theories face a priority problem. If wide content alone specifies what is literally believed, how does narrow content explain rationality without becoming a second proposition? If narrow content is the genuinely psychological content, why regard the wide component as constitutive of mental content rather than a property of its environmental relation? The relative answers depend on whether the explanatory goal is rational assessment, prediction of behavior, account of truth conditions, or public belief reporting.

A useful result is that “the same content” can express different equivalence relations. Two subjects may have the same inferential organization, the same context-to-content function, the same singular referent, or the same truth conditions, and these equalities need not coincide. A philosophical assessment must declare its intended identity conditions before deciding whether a theory is successful.

Cognitive Significance, Frege Cases, and Modes of Presentation

Content theories must explain why a rational person can accept one identity claim and doubt another that concerns the same object. The classic Fregean problem arises when two co-referential names provide different cognitive routes. A person may believe that the morning star appears before sunrise without knowing that the evening star is the same planet. Replacing one name with the other in a belief report can change the truth of the report even while preserving reference. This problem establishes that referential sameness alone is insufficient to characterize the rational organization of a thinker’s beliefs.

For externalists, the explanatory options include fine-grained modes of presentation, belief files, structured propositions, or different levels of content. A theory can be externalist about the object referred to and still distinguish ways of thinking about it. The question is which elements of cognitive significance are themselves internally determined. One subject could possess two distinct files referring to the same individual, while another connects the files through an identity judgment. An account that identifies content entirely with external objects risks losing these distinctions unless it supplements reference with some cognitively articulated structure.

For internalists, Frege cases furnish a principal reason for taking cognitive significance as a content-determining factor. However, a theory based wholly on private descriptive presentations faces its own difficulties: speakers often use names successfully despite possessing incomplete or mistaken descriptions. A theory must allow cognitive differences among thinkers while explaining their ability to coordinate reference and communicate about the same objects. This is why social externalism and modes-of-presentation approaches are often more complementary than their simple opposition suggests.

The problem also helps locate the limits of introspection. A person can know that she is entertaining a thought under one presentation while remaining ignorant that its object is identical to the object of another thought. Knowing one’s manner of thinking does not entail knowing all identity or reference facts about what is thought. This distinction becomes crucial to the compatibility of externalism and first-person knowledge.

Mental Causation and Psychological Explanation

A longstanding objection to wide content holds that psychological explanation should track causally relevant internal mechanisms. Suppose two internally identical systems, placed under matching proximal stimulation, execute the same behavior. If their wide contents differ because one is embedded in an H₂O environment and the other in an XYZ environment, the explanatorily decisive properties may appear to be the shared internal mechanisms. Fodor’s discussion of individualism makes this pressure central to the case for narrow content (Fodor, 1987).

The objection requires careful reconstruction. First, psychological sciences seek generalizations across cognitively similar systems. Second, internal duplicates can share immediate causal powers. Third, if wide contents differ across these duplicates, wide content may fail to capture the relevant generalization. The conclusion that every genuine psychological explanation must therefore be narrow is stronger than the premises establish. A causal account can explain common mechanisms narrowly and also explain successful, world-directed behavior by invoking stable organism–environment relations. Different explanatory questions can legitimately select different variables.

Fred Dretske’s Explaining Behavior (1988) illustrates the ambition to relate reasons, learned representational states, and causal explanations of action. A contentful state’s explanatory relevance may depend on a history through which it acquired a function and on the worldly conditions with which it is coordinated. Ruth Millikan’s teleosemantic project similarly locates representational function in historical relations of biological selection and use rather than in a present intrinsic snapshot (Millikan, 1984; Dretske, 1988). Neither theory can be reduced to the claim that adding any external fact makes a causal difference.

Causal relevance must also be distinguished from causal efficacy. A state’s wide property can figure in a true explanation of why an agent successfully identifies water even if the immediate motor sequence is implemented through local neural processes. Whether the wide property is indispensable to a particular explanatory model is a further empirical and theoretical issue. The same phenomenon can support mechanistic explanation of movement and relational explanation of competent engagement with a specific kind.

Externalist accounts of content therefore need not imply that distant objects instantaneously modify a thinker’s neural condition. A historical or relational property can be essential to a representational description while being realized through ordinary causal interactions. Conversely, an internalist account of mechanism must explain how its purely internal kinds support the interpretation of world-directed correctness and error. The most fruitful research design separates mechanism, content, and ecological success and asks which explanatory function each performs.

Self-Knowledge and the Apparent Incompatibility

Externalism seems to generate a paradox about knowing one’s thoughts. A person ordinarily knows immediately that she is thinking about water. Yet if that content depends on whether her environment contains H₂O or XYZ, and if she cannot determine the chemical nature of her environment from reflection alone, how can her self-knowledge be authoritative? The tension emerges when privileged access to present thoughts is combined with an environmental determination thesis and the possibility of internally indistinguishable worlds (Boghossian, 1989; McKinsey, 1991).

Paul Boghossian’s “Content and Self-Knowledge” (1989) develops the challenge that a thoroughly relational view of content may threaten self-knowledge across environmental changes. Slow-switching scenarios are important here. Suppose a thinker is gradually moved between environments in which the reference of a familiar term differs, without noticing the shift. If the externally determined content changes while her inward phenomenology remains fixed, what enables her to discriminate the relevant thoughts? The problem concerns the epistemic basis of content discrimination over time, not merely the fact that the referent is external.

Michael McKinsey’s “Anti-individualism and Privileged Access” (1991) presses a related concern by exploring what a priori self-knowledge would entail when combined with externalist dependence. The force of such arguments depends on how one understands privileged access. An externalist can accept that one knows the content of a present thought in ordinary circumstances without accepting that introspection also reveals its entire environmental preconditions. A person may know that she thinks that water is refreshing while lacking chemical knowledge of water or detailed knowledge of how the reference of her term is fixed.

Burge’s “Individualism and Self-Knowledge” (1988) argues for the compatibility of anti-individualism with a distinctive first-person entitlement. The content identified in a second-order thought can inherit the same environmental determination as the content of the first-order thought. Such self-knowledge does not require independently deriving worldly reference from an inner description. The thinker’s self-ascriptive act may be epistemically distinctive even though it employs socially and environmentally individuated concepts (Burge, 1988).

The resulting distinctions are precise. Knowledge that one is thinking a particular content, knowledge of the referent’s properties, knowledge of the history fixing reference, and ability to distinguish slow switches are separate achievements. An argument against externalism needs to demonstrate that the commitments of an externalist account actually contradict an independently defensible form of first-person authority. Merely pointing to ignorance of chemistry does not establish that contradiction. Contemporary discussions elaborate contextual, entitlement-based, and discriminatory-ability responses while continuing to debate the required strength of self-knowledge (Parent, 2025).

Skepticism, Brains in Vats, and Environment-Dependent Thought

Content externalism interacts with philosophical skepticism because thought about an external world may require suitable relations to that world. Some anti-skeptical arguments propose that if a subject can meaningfully think about certain external objects, total isolation from all such objects may be incompatible with the possession of the very concepts used to formulate skeptical doubt. In these arguments a substantive thesis about reference combines with assumptions about what the subject can know concerning her own utterances or thoughts.

The reasoning must not be abbreviated into the slogan that externalism defeats skepticism. A skeptic can ask whether the thought one entertains has the content the anti-skeptical argument presumes, whether the purported a priori knowledge of its content is available, and whether a self-referential argument proves the existence of the ordinary surrounding world rather than merely some relation capable of fixing meaning. Content externalism alone does not guarantee that every ordinary perceptual judgment is accurate or that the thinker possesses evidence of its accuracy.

Brains-in-vats cases also differ depending on history. A newly envatted subject retaining a history of ordinary interactions may plausibly preserve established reference for a time; a lifelong isolated system might possess differently fixed representations. These are distinct cases, and externalist theories differ over the importance of historical anchoring, learning conditions, and present environment. The ordinary meaning of a word can be sustained by causal history and community even where the agent’s immediate stimulation is misleading.

A central philosophical lesson is that the skeptical scenario cannot be assessed independently of a theory of content. What a brain can think depends on what makes its representations refer. But this dependence cuts both ways: the skeptic can test the theory’s conditions of reference, and the externalist can reject a scenario whose assumed thoughts are inconsistent with its causal and social history. The live issue is an account of the counterfactual possibilities under which intentional content exists, not a verbal proof that deceptive appearances are impossible (McKinsey, 1991; Ebbs, 2025).

Empty Names, Fiction, and Uninstantiated Kinds

The externalist emphasis on real-world relations meets a distinctive problem when thoughts concern what does not exist. A person can fear a fictional monster, hope to find a legendary island, or mistakenly believe that a substance has a distinctive hidden nature when no substance satisfies the relevant description. Such representations have inferential, emotional, and behavioral significance despite the absence of an external object apparently available to determine their reference.

An internalist can use empty representations as evidence that significant cognitive content can be constituted without a successful relation to an existing referent. Segal’s treatment of empty natural-kind concepts makes this point especially sharp. A thinker can possess an organized concept that resembles a natural-kind concept in structure even when the natural kind is missing. If such content is fully intelligible, a theory giving constitutive priority to actual external kinds appears incomplete (Segal, 2000).

Externalist responses distinguish kinds of content and conditions of reference. A general or descriptive content may remain when singular reference fails; a fictional representation may depend on a socially maintained narrative rather than a real object; an unsuccessful natural-kind term may possess a use-governed or deferential role without securing a determinate worldly kind. Some externalists allow gappy contents in object-dependent propositions, while others reconstruct the content under a broader interpretive or descriptive treatment.

There is no simple inference from the meaningfulness of fictional thinking to the falsity of all content externalism. Nor does a successful causal theory of ordinary natural-kind terms automatically handle an empty term. The proper test asks whether one explanatory system can accommodate successful reference, systematic misreference, and failed reference while preserving their distinctions. The existence of empty and fictional thought is a major constraint on the scope of externalist theses.

Phenomenal Intentionality and the Reach of Experience

The phenomenal intentionality research program argues that some intentional content is grounded in the phenomenal character of conscious experience. If what it is like to undergo an experience constitutively determines a kind of intentional directedness, and if identical phenomenal character can be sustained across environmental differences, this content may be narrow. The proposal is important because it offers an internalist source of representational significance that does not depend solely on abstract computational roles (Horgan & Tienson, 2002; Bourget & Mendelovici, 2026).

The argument requires two premises: a phenomenal basis is sufficient for the relevant intentional property, and that phenomenal basis is itself internal under the chosen duplication relation. Both are contested. Some representationalist accounts treat phenomenal character as partly dependent on externally individuated perceptual content. Even if two subjects are phenomenally indistinguishable, it remains a further question whether what each perceives or thinks is the same. A theory may identify an appearance-level content shared across environments while allowing object-involving perceptual content to differ.

The strongest form of phenomenal internalism is also under pressure from unconscious cognition. Unconscious beliefs, recognitions, and inferential states may be intentionally interpretable without current phenomenal presentation. A phenomenal intentionality theorist must either connect them to actual or potential phenomenology or explain their intentionality as derived from a narrower phenomenal base. The externalist can challenge this priority while acknowledging that phenomenology is indispensable to an account of conscious cognition.

This research program illustrates a recurring structure of the broader debate. The internalist needs a content that captures how the world appears to a subject. The externalist needs a content that captures what worldly objects or properties the subject's state concerns. Their interaction becomes philosophically productive when each content claim specifies its truth conditions, phenomenological realization, and role in rational explanation. Phenomenal intentionality remains a competing theory of content, not a conclusive refutation of every environmentally grounded account.

Teleosemantics, Causal Histories, and Information

Naturalistic theories seek to explain content by reference to the causal, informational, or functional relations of representational states. A state may carry information about the environment because it systematically covaries with a condition. Yet covariation alone faces the disjunction problem: a signal produced both by its normal target and by abnormal causes may seem to represent the disjunction of every cause capable of producing it. An account of misrepresentation must explain why a token can be incorrect about its target even while causally triggered by some actual condition.

Dretske’s informational approach connects cognitive significance with learning and the role through which a signal becomes causally relevant to action. Millikan’s teleosemantics invokes proper functions and histories of selection to distinguish what a representation is for from every condition that can produce it (Dretske, 1988; Millikan, 1984). These proposals are important instances of externalist determination because the present state can inherit representational correctness conditions from relations extending beyond its current intrinsic realization.

A historical-function approach distinguishes normal success from accidental correlation. The challenge is to specify the relevant history and function without assigning content by a retrospective choice that merely reproduces our interpretive intuitions. Biological selection provides one sort of grounding. Learning histories, social practices, artifact functions, and institutional conventions might provide others, but the extension from biological selection to these settings requires independent arguments.

For internalist theories the corresponding difficulty is to explain normative correctness on an intrinsic basis. If any currently available causal disposition counts equally, what constitutes a representation as representing one condition correctly and another incorrectly? A computational or phenomenal organization can supply structural constraints, but the relation between such constraints and determinate reference remains contested. Both families therefore confront the question of how representations possess standards of success and failure.

Embodied, Extended, and Distributed Cognition

The debate over mental content intersects with externalist approaches to cognition, but the claims should be carefully separated. Content externalism concerns the individuation of representational properties. Extended-mind hypotheses concern whether material processes outside the organism can partly constitute a cognitive process. Distributed cognition examines how thinking can be organized across agents, instruments, records, and environments. An externalist about thought content can hold that the machinery of thought remains wholly neural. A theorist of extended cognition can maintain that some representational operations are defined by functional coupling without adopting strong externalism about reference.

This separation is especially important in scientific practice. A researcher may use a measuring apparatus and a communal vocabulary that jointly determine what a laboratory inscription is about, while the immediate inference is implemented in a local brain or computational system. The public calibration of an instrument can be relevant to the truth conditions of a report. The internal computation that manipulates the report can remain identical across two environments where the calibration differs. A good explanation of the research activity may require both an internal computational account and an external account of measurement, reference, and institutional standards.

Content externalism offers a caution against interpreting representational tokens in isolation. Identical signs can serve different purposes in distinct communities and historical circumstances. Yet distributed cognition adds a separate hypothesis: external artifacts might function as parts of a wider cognitive routine. Whether such inclusion is justified depends on the organization of memory, access, reliability, and cognitive coupling, not simply on the fact that symbols refer externally.

These interdisciplinary connections therefore supply test cases rather than automatic confirmation of a single metaphysical thesis. They sharpen the need to distinguish constitutive content, mechanism, social practice, and environmental success.

Computation, Artificial Systems, and Semantic Attribution

Artificial systems furnish a demanding test for the individuation of content. A computational state can be described by its internal configuration, its role in transforming signals, its historical training process, or the external conventions connecting its outputs to objects and situations. These descriptions can coexist while answering different questions. The issue is whether a token’s representational identity derives wholly from its internal functional organization or also from a history of use, learned regularities, referential grounding, and public interpretive practices.

Consider two systems with an identical frozen model and matching internal states that operate within different institutional environments. Their outputs may be mechanically identical while an authorized database field, local naming convention, or referential assignment gives the same string different public significance. This example supports the possibility of semantic externalism at the level of use and attribution. Whether it establishes externalism about genuinely possessed machine mental content depends on a prior account of the criteria for intentional states. That philosophical issue cannot be settled merely by showing that identical strings have different conventional meanings.

A stronger artificial-cognition research program asks which content levels support generalization and error correction. Internal computation is relevant to the system’s inferential transitions and stable capacities. External reference matters when assertions concern particular documents, persons, resources, or events. Historical provenance may matter when the functional identity of a representation is linked to its learning or deployment trajectory. These dimensions require explicit articulation if a theory is to avoid attributing a complete semantic property simply from superficial verbal resemblance.

The horizon of the Artificial Era makes this question philosophically consequential: philosophical theories of reference originally elaborated around human speakers must be assessed for their applicability to engineered nonbiological systems. That extension is an analytic research question. It neither follows automatically from Putnam’s Twin Earth case nor licenses the inference that every language-producing device is a bearer of mental attitudes. A defensible extension must determine representational vehicles, individuation criteria, environmental relations, and standards of correctness in the system under investigation.

The contribution of the internalism–externalism debate is methodological. Machine representation should be analyzed at the levels of mechanism, functional role, public signification, and object-directed correctness. A theory that specifies these relations can investigate artificial cognition without confusing token manipulation with an established account of intentionality.

Comparative Evaluation of the Major Positions

Strong content internalism offers a unified basis for comparing internally duplicate thinkers and for explaining similarities of reasoning and immediate behavior. Its central difficulty is referential determinacy. It must account for natural-kind terms, incomplete competence, singular reference, and the social coordination of meaning without treating the public reference of thought as merely incidental. Segal’s defense shows that radical internalism remains a serious philosophical option, but one must evaluate whether its preferred cognitive contents do all the explanatory work traditionally assigned to belief content (Segal, 2000).

Strong content externalism offers a compelling explanation of why social and physical relations affect ordinary content attributions. Putnam and Burge supply different arguments for this explanatory dependence. Its central difficulty concerns common cognitive structure across environmentally different thinkers and the explanatory status of broad content in mechanistic psychology. Strong externalism also must address empty terms and the possession of content when ordinary anchoring relations fail. The strength of externalism is therefore clearest when a theory makes explicit the content family to which its arguments apply.

Dual-content and two-dimensional views provide a systematic reconciliation by distinguishing cognitive organization from world-involving truth conditions. Their distinctive burden is to make the internal component precise, psychologically significant, and more than an informal restatement of similarity. They must also specify whether the two components are genuine contents, aspects of one content, or different forms of representation employed for different explanatory purposes (Chalmers, 2002).

Teleosemantic and informational accounts place representational determination in a causal-historical structure that promises a naturalistic explanation of correctness and error. Their success is constrained by the difficulty of individuating functions, historical types, and information-bearing relations. Phenomenal intentionality accounts, in contrast, give conscious presentation a constitutive role; their difficulties concern unconscious intentionality and the external determination of the content of perceptual awareness. A comparative evaluation cannot assign these accounts to a simple sequence of philosophical progress, because their objectives differ in important respects.

Stalnaker’s examination of what is “in the head” clarifies why externalist content attribution can remain useful even when a finer internal psychology is needed (Stalnaker, 1989). The most illuminating comparison therefore uses four questions. Does the theory explain which object or property a thought represents? Does it characterize differences in cognitive significance and rational inferential role? Does it support counterfactual generalizations used in psychological science? Does it distinguish success, error, and reference failure? The answers may favor different theories for different tasks. That plural evaluation has explanatory merit only if the relations among the levels are specified and empirically or conceptually defensible.

Open Questions and Contemporary Research Directions

Several issues remain structurally open. First, which forms of content are fundamentally truth conditional? Broad propositions provide ordinary truth conditions, but proposed narrow contents range from functions to modes of presentation to phenomenal structures. One must decide whether a narrow content can be evaluated directly for truth or only relative to a context.

Second, how much social change can a concept accommodate while retaining its identity? Burge’s arthritis example presupposes intelligible differences in communal usage. Historical conceptual transformations can be gradual, contested, or institutionally fragmented. A theory should explain when such variation changes the concept itself and when it changes prevailing beliefs concerning a stable concept.

Third, how should a representational theory individuate content under radically novel conditions? Counterfactual agents without familiar earthly environments, biological organisms with unusual modalities, and engineered computational systems test the generality of familiar natural-kind and social-deference arguments. The point is not to stipulate that the traditional cases transfer unchanged, but to identify which referential dependence relations survive the change of subject matter.

Fourth, what is the relation between a subject’s conscious perspective and the environment-dependent proposition expressed in thought? A satisfying account should explain cognitive significance, first-person authority, perceptual variation, and the difference between being wrong about an object and merely lacking an object.

Fifth, what should the successful empirical integration of content theory look like? Psychology often models internal processing with functional or neural variables while explaining ecological success through agent–environment relations. The philosophical dispute can clarify how these explanatory levels connect and what evidence would count against a particular claim of content determination. A theory that cannot distinguish constitutive dependence from causal history or interpretive convenience leaves its empirical predictions underspecified.

The contemporary literature on narrow content, phenomenal intentionality, externalism and self-knowledge, and content externalism and skepticism demonstrates that the debate remains active across several subfields (Brown, 2022; Parent, 2025; Ebbs, 2025; Bourget & Mendelovici, 2026). Continuing research depends on sharper individuation of the target properties rather than repeated assertion that content is simply “in” or “out” of the head.

Philosophical Synthesis: A Layered Account of Content

The arguments surveyed support a systematic distinction among the intrinsic realization of a cognitive state, its cognitive significance for a thinker, and its public or world-involving representational conditions. These distinctions are analytical results of the present research; they do not constitute a newly canonized philosophical theory. They describe explanatory commitments that any adequate account of content must confront.

Intrinsic realization concerns the organization through which a subject performs cognitive operations. Matching realization can support matching dispositions or computations without by itself settling every question of reference. Cognitive significance concerns how a state is integrated into reasoning, discrimination, planning, and awareness. Externally determined content concerns the specific objects, kinds, or circumstances that a thought represents under an account of reference. The three can overlap, and different theories propose different dependence relations among them.

The central result is that externalist thought experiments successfully challenge the claim that all ordinarily ascribed representational content supervenes on a narrowly specified intrinsic state. Putnam shows why shared inner descriptions can be insufficient to fix a natural-kind term’s reference. Burge extends the dependence claim to propositional attitude attribution within socially organized conceptual practices. Those arguments do not automatically eliminate every viable notion of narrow content, because sameness in cognitive organization can remain explanatory even when wide reference differs.

The corresponding result for internalism is equally important. Psychological similarity is a real explanatory phenomenon, and a theory of mental content should account for it. Yet the existence of such similarity does not prove that the full truth-conditional object of thought is intrinsically determined. If narrow content is to be a genuine philosophical explanation, its structure and conditions of possession must be specified. The possibility of a narrow determinant is stronger than the assumption that all ordinary contents are narrow.

Self-knowledge demonstrates that content constitution and knowledge of content belong to distinct philosophical levels. Environmental dependence can coexist with privileged first-person entitlement, provided that knowing what one thinks is not confused with knowing all the external facts that make the thought possible. Mental causation demonstrates a related separation between causal mechanism and environmentally situated successful action. Empty names and fictional objects show why content types must be differentiated according to their referential commitments.

The resulting philosophical synthesis preserves the distinct questions that motivated both sides of the debate. Content determination describes what makes a representation have its particular intentional identity. Cognitive explanation describes how that representation participates in thought and action. Epistemic authority describes how thinkers know and assess their own intentional states. A mature philosophy of mind must relate these questions while allowing their answers to differ. The unit of analysis is always the intentional property specified together with its individuation conditions and explanatory role.

Conclusion

Content internalism and content externalism investigate the metaphysical and explanatory conditions that determine the objects and circumstances represented in thought. Their defining disagreement is a supervenience dispute: whether the intrinsic condition of a subject suffices for identity of intentional content, or whether some contents constitutively depend on environmental, communal, or historical relations.

The historical trajectory from sense and reference to causal theories of naming, Twin Earth, social anti-individualism, and the debate over narrow content reveals an expanding analysis of representation. Putnam’s physical-environment argument and Burge’s social argument provide powerful grounds for externalism about major classes of ordinary referential thought. Individualist theories remain philosophically significant because internal cognitive organization explains stable reasoning patterns and shared causal capacities across otherwise different environments. Two-factor models, phenomenally grounded theories, and teleosemantic approaches each attempt to reconcile or prioritize dimensions of this problem.

Three results govern the final evaluation. The identity of a referent, the cognitive significance of a presentation, and the mechanisms implementing a state can vary independently in important thought experiments. Philosophical disputes arise when a theory treats one of these explanatory dimensions as automatically exhausting the others. A successful account must therefore specify its content type, the relation of intrinsic duplication used to compare thinkers, and the task for which the content is theoretically introduced.

The questions remain central to self-knowledge, skepticism, cognitive science, and the interpretation of artificial representation. The durable philosophical achievement of the internalism–externalism debate is the demonstration that the determination of content demands an explicit account of the relation among thinkers, their cognitive organization, their communities, and the world they represent.