3.1 Rejection of the Claim that “Meaning is Solely Determined by Internal States”

Internalism maintains that meaning is entirely determined by an individual’s mental states (such as beliefs and intentions) within the brain, independent of the external environment. Externalism counters this with the well-known “Twin Earth” thought experiment: on Earth, the word “water” refers to H₂O, as determined by its chemical structure. Suppose that on “Twin Earth” there exists a substance whose structure is “XYZ,” which is identical to water in appearance and function. The inhabitants of both Earth and Twin Earth use the word “water,” yet the meanings differ because the referents are different – this difference is not fixed by mental states but by the environment (Putnam, 1973). In other words, even if your mental states are identical to those of your counterpart on Twin Earth, the word “water” you use refers to different substances. Thus, internal mental states cannot fully determine semantic content.

Most semantic externalists employed Putnam’s thought experiment of Twin Earth to demonstrate that the meaning of words does not solely depend on mental states. Twin Earth, which differs from Earth only in the chemical formula of water, involves a scenario where Oscar and the Twin Oscar both say, “Please give me water.” The reference of the two statements differs – one refers to H₂O, the other to XYZ. It is therefore clear that externalism emphasizes external environmental factors such as the relationship between meaning and the external world in which the speaker resides. In other words, externalism argues that meaning is determined by what the speaker refers to.

3.2 Rejection of “Private Language” and “Individualism”

Internalism maintains that the meaning of language can be independently grasped by the individual, whereas externalism questions its ability to explain the “commonality and stability of meaning.” Externalism emphasizes that language is inherently public and shareable, and its content can only be understood and shared by others insofar as it relies on the knowledge and usage of the community to which the speaker belongs.

Wittgenstein demonstrated the impossibility of a private language, arguing that meaning depends on public linguistic practices and the normative standards of external society (e.g., the correctness of “addition” is determined by communal usage). He criticized the philosophical tradition of idealism, pointing out that mental images are not static entities labeled and stored; rather, if one is not conscious of the images, they vanish. When attempting to evoke a mental image of “red” again, how do you confirm the specific image that you wish to conjure up? This requires prior knowledge of the meaning of “red”: “‘Red’ means the color that occurs to me when I hear the word ‘red’” – would be a definition (Wittgenstein, 1967, p. 88e). Yet, that mental image is precisely what explains your knowledge of the meaning of “red.”

From this perspective, the internalist model presumes that you “have the ability to recognize which internal objects are red” when explaining “how you can recognize external objects as red.” This applies to any other scenario: while you can correctly identify external objects, you must also be able to select the correct internal objects. In other words, the internalist model makes presuppositions about the phenomenon that demands clarification. Wittgenstein, employing the analogy of chess, offered a compelling rebuttal: If we wish to know if someone can play chess, we are not interested in anything happening inside him. If he perceives knowing how to play chess as an internal process, then what interests us is the internal occurrences, and we can only ask him to contemplate what criteria would demonstrate his ability and what would be the criteria for verifying the “internal state.”

If words and signs derive their meaning fundamentally from being associated with internal objects – that is, ideas – how do these internal objects acquire their meaning? If you associate the word “red” with a mental image of a red square, does this confer a new meaning upon “red”? The answer is no. The association in this case is akin to a red square printed on a card, which is interpretable in countless ways – they lack inherent meaning. The reason is that the internalist model offers a circular explanation. According to internalism, to understand meaning is to be in a particular mental state; any two individuals would have the same understanding of the meaning of words if they were in the same mental state. This model seeks to explain how public words and signs gain meaning by appealing to private, internal objects, and it assigns meaning directly to these internal objects.

In the internalist model, the action of language consists of two parts: the handling of signs, and understanding, meaning, and interpreting these signs. These latter activities seem to take place in a queer kind of medium, the mind (Wittgenstein, 1972, p. 3). Wittgenstein reminds us to resist the temptation to regard meaning and understanding as enigmatic internal activities or processes. He endeavors to redirect our focus from internal occurrences to our publicly observable capacities. Meaning isn’t concealed but rather contingent upon the usage accorded to words and other signs. To understand a word’s meaning is not to associate the word with some mysterious internal object but to understand how it is used; as Wittgenstein argued, “Let the use teach you the meaning” (Wittgenstein, 1967, p. 212e). Wittgenstein did not propose an alternative theory of meaning; in his view, such a theory is superfluous because the meaning of language arises from its use, and it is the use that shapes the present and future meaning of language.

Compared with Wittgenstein, Burge places greater emphasis on the decisive role of linguistic division of labor and communal norms in shaping individual thought. He prefers the term “anti-individualism” to define his position with greater precision. Specifically, the external environment (including the physical world and the social community) is not merely a cause that triggers or influences mental content; it constitutively determines the nature of an individual’s mental states (what makes a thought be about water rather than about XYZ) (Burge, 2012, pp. 263–270). Without these external relations, the corresponding mental state would not be the state it is (or would not exist at all). For example, differences in how a concept is used across communities lead individuals to possess different mental contents. The relevant external factors need not be internally accessible or self-known to the individual: even if one cannot fully access or grasp certain details of the environment, those details can still determine the content of the thoughts one has.

Burge argues that linguistic reference does not always depend on whether an individual can exhaustively describe what is referred to. Even when a person’s beliefs and descriptions concerning a proper name or a natural-kind term are mistaken, he can still successfully refer to the bearer of that term. Ordinary speakers may be unable to describe Aristotle’s life accurately, yet they can still, via the causal chain sustained by the linguistic community, correctly refer to the historical Aristotle. This indicates that the constitutive determinants of reference for proper names (the causal chain) are independent of an individual’s internal descriptive capacities. The difference between linguistic reference and physical properties shows that reference is fixed by environmental and nonpsychological factors; meaning depends on the practices of the external linguistic community rather than on individual understanding.

In Burge’s view, the essence of this referential externalism is ultimately rooted in the “externality” of perceptual states themselves. Even if an observer cannot discriminate an object from its duplicate, what is perceived is still the particular object before the observer rather than some other item that merely resembles it in appearance, because perceptual reference depends on the causal impact of the actual object on the senses rather than on the observer’s prior perceptual classifications. Perception is the mind’s most immediate point of contact with the world, and it is itself an anti-individualist process: the nature of a mental state (here, a perceptual state) is constitutively determined by its relations to the external environment. The “externality” of mental content is an extension and complication of the “externality” of perception.

3.3 Opposing the View that Meaning Is an Abstract Entity or Mental Image Separate from Language

Compared to the physical externalism and anti-individualism mentioned earlier, Quine offered a harsher criticism of idealism. The ideational theory of meaning posits meaning as a psychic entity that exists in the human mind independently of language. Quine strongly opposed this elusive notion of “meaning” that lacks explanatory power. First, the notion of “idea” itself is vague and ambiguous. For example, what precisely constitutes the idea of a horse? What are the ideas of words such as “and,” “or,” or “whether”? Does the same word produce the same idea corresponding to the same meaning in the minds of different individuals? While we learned to apply “red” to blood, tomatoes, ripe apples, and cooked lobsters, how do we associate “red” with the same idea? Second, the term “idea” is pernicious, as its use breeds an illusion. This is akin to explaining a hypnotic agent as “hypnotic.” It seems to explain something when, in fact, it explains nothing. Therefore, we cannot employ “idea” to explain anything. In his view, “The associated idea, the associated sensation, is as may be. Language bypasses the idea and homes on the object. Than the idea there is little less useful to the study of language” (Quine, 1990, p. 35).

For Quine, the approach of positing “psychic entities” or “mental entities” solely for the purpose of facilitating explanations of physical phenomena is acceptable, yet these phenomena ultimately demand physiological or physical explanations. If, however, mental entities are posited to explain psychological events or states, like why molecules and atoms are posited in physics, then a theory that can be explained through mental states or events can also be explained through physical or physiological states related to these mental behaviors. Since the existence of physical or physiological states is unequivocal, there is no need to posit other elusive mental states (Quine, 1960, p. 264).

According to Quine, language is a social activity, and its meaning should be the behavioral responses of people, devoid of anything transcending what should be evident from observable circumstances and conspicuous actions. There, the study of linguistic meaning must commence from the observable behaviors of individuals, starting from affirmative or negative responses to observation sentences. He proposed to explore the meaning of words within a “stimulus-response” framework. Quine pointed out that individuals acquire language and comprehend meaning through the linguistic game of inquiry-agreement-disagreement.

First, Quine, starting from a behaviorist standpoint, introduced the key concept of “stimulus meaning,” which sums up the speaker’s behavioral disposition to assent to or dissent from a sentence once the conditions for stimulation are established, while the stimulation serves as the trigger of this behavioral disposition. In other words, Quine took the view that what triggers human behavioral responses is the combination of external stimulation and the stimulus meaning of language. Moreover, the aggregated stimulation that gives rise to the stimulus meaning of a sentence is publicly observable (Quine, 1960, pp. 32–38).

Quine then put sentences into two categories based on their reliance on sensory stimulation: occasion sentences and standing sentences. In particular, observation sentences, which are special occasion sentences, hold a distinctive status in his theory of meaning. Quine defined observation sentences as follows: “An observation sentence is an occasion sentence that the speaker will consistently assent to when his sensory receptors are stimulated in certain ways, and consistently dissent from when they are stimulated in certain other ways” (Quine, 1981, p. 25).

The chief characteristic of observation sentences lies in their observability, that is, the meaning of an observation sentence can be examined based on publicly observable behaviors among subjects. How do we derive rich theories about the external world from meager sensory stimulations? The answer is that observation sentences are directly associated with our sensory stimulations. Subjects stimulated can clearly assent to or dissent from such sentences, and the truth value of observation sentences depends on the current stimulation. Furthermore, observation sentences also exhibit intersubjectivity. They describe physical objects rather than reports of sensory materials, and physical objects are external, public entities capable of forming objective standards.

Consequently, meaning becomes an object of empirical investigation. While stimulation is a private affair of subjects, the meaning of class as stimulation is obtained among subjects. The meaning of linguistic expressions can only be grasped through observation based on the context of stimulation. Meaning is the behavioral deposition of individuals to assent to or dissent from a particular observation sentence in response to the existing conditions of stimulation.

3.4 Questioning Descriptive Properties

Internalism holds that the meaning of a name depends on the descriptive content possessed in the speaker’s mind. Externalism challenges this, especially with respect to the reference of names. Physical externalism maintains that the meaning of natural-kind terms is determined by the material essence of the external world. For example, the meaning of “gold” depends on its atomic number – element 79, Au – rather than on a subjective impression of a “yellow metal.” Subjective impressions not only struggle to account for the necessity of reference, but they also fail to secure rigid designation. Relatedly, the Causal Theory of Reference emphasizes the direct referentiality of names, rather than mediation by descriptions. Descriptions themselves cannot stably fix a referent or capture its real identity. For instance, “Aristotle” refers to a specific historical individual by virtue of a historical-causal chain tracing back to an original naming event; even if someone falsely believes that “Plato was Aristotle’s student,” their use of “Aristotle” still picks out the actual historical person.

It follows that successful reference does not depend on whether a speaker commands sufficient or accurate descriptive content; on the contrary, such content is often constrained by ignorance or error. On this basis, Kripke proposes that a name is first linked to a particular individual through a “baptism,” and is then transmitted and sustained through communicative chains within a social-linguistic community (Kripke, 1980, pp. 97–106). In other words, the meaning of a name is rooted in an external historical-social network rather than in the descriptive content within an individual mind, thereby further refuting the internalist view that semantics depends on subjective representations.

In summary, semantic externalism adopts a fundamentally different stance from semantic internalism on the issue of meaning. Its central thesis – “meanings just ain’t in the head” – shifts the focus of inquiry regarding the “source of meaning” from the traditional internalist framework, which concentrates exclusively on individuals’ internal cognitive states, toward an examination of external factors. This has profoundly influenced contemporary philosophy of language, compelling theorists to regard the physical environment, social conventions, and the causal-historical chains of word usage as indispensable components in the constitution of meaning, thereby reshaping the direction of inquiry into reference, content, and the very nature of meaning.

4.1 Distinguishing between “Meaning” and “Reference”

The Twin Earth thought experiment is regarded as one of the most powerful criticisms of semantic externalism. However, this case in fact conflates meaning and reference. Meaning is constituted by internal mental states, such as conceptual networks and intentional states, whereas reference concerns the external objects to which words point. The Twin Earth thought experiment merely demonstrates that external reference changes (H₂O vs. XYZ), but this difference does not amount to a refutation of the core claim of internalism. The psychological meaning or conceptual representation of “water” on Earth and on Twin Earth can remain entirely consistent (a transparent, drinkable, tasteless liquid). When the subject does not consciously distinguish between the two, the difference lies only in the externally assigned reference, not in any substantive difference at the level of the subject’s pragmatic use. There is no contradiction between the internality of meaning and the externality of reference – semantic internalism can fully accept that reference is determined by external environments without abandoning the internal nature of meaning itself.

Here, we must mention Frege. Frege was the first to expel idealism from the realm of semantic theories. He offered the following criticism:

Thus everything leads into idealism and with perfect logical consistency into solipsism. If everyone designated something different by the name “Moon”, namely, one of his ideas, then admittedly the psychological way of looking at things, including language, would be justified; but a dispute about the properties of the Moon would be pointless: one person could quite well assert of his Moon the opposite of what another person, with equal right, said of his. If we could grasp nothing but what is in ourselves, then a genuine conflict of opinions, a reciprocity of understanding, would be impossible, since there would be no common ground, and no idea in the psychological sense can be such a ground. (Frege, 1997, p. 206)

Frege did not mean that we cannot view things from a psychological perspective. Rather, he argued that we cannot rely on the speaker to determine the truth value of a sentence, which risks losing the public foundation necessary for understanding.

While mentalists attempt to explore meaning and judgment through the study of inner processes, Frege argued that the meaning of a linguistic expression lies not in it being a mental image generated by the speaker or listener but in its objective role in determining the truth value of an entire sentence. He pointed out that logicians study linguistic expressions, which signify objectively real entities. According to Frege, meaning stems from the examination of expressions, not conjectures about mental processes. For instance, concepts like whales and mammals are objective entities, not subjective constructs. A specific whale belongs to the concept of whales, which is subsumed under the concept of mammals. These conceptual connections are objective. Linguistic expressions possess qualities available for public scrutiny, and the study of meaning depends on examining these qualities rather than speculating about mental processes.

It should be noted in particular that although Frege refused to reduce meaning to psychological states, he did not deny the internal connection between meaning and the rational subject. On the contrary, his theory of meaning emphasizes that meaning is closely tied to an individual’s internal understanding: to understand the meaning of an expression, one must be able to “grasp” it in thought. Frege begins his analysis with the problem of sameness, posing the question, “Is sameness a relation? A relation between objects? Or a relation between names or signs?” (Frege, 1996, p. 186). If A=B signifies merely a sameness between signs, then those signs are associated and possess a relation of sameness because they not only name or refer to something but also refer to the same referent. If that is true, the relation of sameness does not solely pertain to signs. If A=B represents solely a sameness between the objects signified by signs, then, in the case where A=B is true, the cognitive significance of A=B (two signs referring to the same object) and A=A (a sameness between the object and itself) would be substantially the same. The difference lies only in the presentation of the signs. In this case, the relation of sameness pertains not only to objects but also to the relation between signs and objects.

Frege, therefore, posited, “Associated with a sign (name, phrase, expression) are not only the objects it denotes, namely the reference, or nominatum, of the sign, but also its sense, connotation, or meaning; the sign’s sense encompasses the manner and context in which it presents itself” (Frege, 1996, p. 187). His theory of meaning is thus divided into two parts: one concerning reference and the other concerning sense. In a nutshell, the “reference” of a word (name) denotes concrete objects, while “sense” denotes the meaning represented by the object. For example, “morning star” and “evening star” both refer to the same celestial body, Venus, but why does “morning star is evening star” offer more knowledge than “morning star is morning star”? Why is the former an important astronomical discovery while the latter is a logical truth devoid of empirical content? Because “morning star” and “evening star” have different senses.

Frege provided clear distinctions between signs, the sense of signs, and the reference of signs, stating, “What corresponds to a sign is a specific sense, and what corresponds to the specific sense is a specific reference. What corresponds to a reference (object), however, is not just one sign. In different languages, and even in the same language, the same sense can be expressed by different expressions” (Frege, 1996, p. 187). According to Frege’s definition, meaning is both an objective, public domain shared by all and also a domain comprehensible to all. When I say “traffic light,” I am actually expressing the sense of the word “traffic light,” and you are able to understand me because you understand the word’s sense and the physical object.

Building on this basis, Frege further pointed out that the sense of a sign cannot be conflated with the “image” with which it is associated. If the reference of a sign is a sensory object, then the image of this object in my mind is an internal mental image stemming from sensory impressions and memories generated by activities. Even for the same individual, the same sense does not always accompany the same image. Images are subjective; a person’s image is not the same as that of another. Consequently, differences arise between images associated with the same sense. The term “Bucephalus,” for instance, may evoke vastly different images in the minds of a painter, a rider, and a zoologist. Images, therefore, are fundamentally different from the connotation of signs. The connotation of a sign may encompass common features shared by many images and is, therefore, not a part of individual minds or individual mental patterns. This is the case because we cannot deny the shared repository of ideas inherited by humanity across generations (Frege, 1996, p. 188).

Frege’s notion of “image” is the psychological “image,” which pertains to individuals’ “understanding” of the sense or reference of a sign, or the “subjective sense” formed by individuals based on their understanding of the sign. However, signs also possess an “objective sense” that reflects the objective world and remains unchanged regardless of the cognitive subject. The objectivity of a sign’s sense allows different cognitive subjects to grasp and share the same “sense.” It is precisely the fact that the “objective sense” of signs is shareable and universal that allows for the dissemination of human thought. In other words, while the process of understanding a sense is an internal mental activity of individuals, the understood sense is neither private nor inside the mind, as different minds can apprehend the same sense, and people from different periods of history can also apprehend the same sense.

To illustrate the distinction between image, sense, and reference, Frege offered a brilliant explanation using the example of the “moon.” In his analogy, the “moon” is akin to the true image displayed on the objective lens of a telescope when observing the moon. This image is objective and can be used by many different observers. However, the moon as seen by an observer, that is, the image on the observer’s retina, is a subjective image. There may be as many images of the moon as there are eyes to behold it. Nevertheless, whether it is the objective image on the lens or the subjective image seen by an individual, they both share a common reference – the celestial body, the moon (Frege, 1996, p. 188).

While the distinction between sense and reference provided by Frege did not directly lead to or negate any particular theory of meaning, it offered a well-conceived framework for subsequent studies on the problem of meaning. Along this path of exploring meaning pioneered by Frege, meaning seems to be a mode of presentation: on one hand, meaning is related to the psychological life of the thinker, a representation in the mind; on the other hand, meaning is objective and communicable. This, to a certain extent, liberates meaning from the constraints of reality. Meaning is thus regarded as how objects present reason: objects present themselves in different ways, and each representation corresponds to a particular meaning of a word.

4.2 Clarifying “Mental State” or “Internal State”

Internalists do not deny that environment and society may influence word use, but they maintain that the decisive factor that determines meaning still lies in the subject’s internal mental states. Such internal meaning possesses cognitive stability, thereby ensuring that the cognizing subject can continue to employ concepts in a steady and sustained manner in thought. It should be noted that what internalism calls a “mental state” is not a purely internal event cut off from the external world but a cognitive state that includes intentionality – namely, mental representations such as beliefs and attitudes that are always directed toward the external world. Within the internalist framework, the meaning of a word is determined by its inferential function within an individual’s overall cognitive system (Greenberg & Harman, 2006, pp. 295–322), and different conceptual networks give rise to differences in mental states.

In the Twin Earth case, Oscar and Twin Oscar differ in their mental states when they use the word “water.” When Earthlings mention “water,” they tacitly invoke knowledge related to H₂O (e.g., “water comes from rivers and rain,” etc.) and form an inferential network with concepts such as “transparency” and “quenching thirst,” associations that ultimately point to H₂O. Twin Earthlings, by contrast, tacitly rely on a body of knowledge about XYZ; their conceptual network is connected with twin “rain,” twin “rivers,” and the like, ultimately pointing to XYZ. The two mental representations are not essentially the same, and the difference in mental representations directly constitutes a division in mental states. A distinction grounded solely in internal mental states is sufficient to explain the difference in meaning, without introducing external factors. Thus, the refutation offered by the “Twin Earth” thought experiment fails.

The core tenet of Locke’s theory of meaning is that words denote ideas and indirectly refer to objects through these ideas. However, since Locke believed that words themselves have no fixed meaning and the relationship between words and what they denote or connote is arbitrary (he acknowledged the existence of private languages), theories of meaning and knowledge based on idealism inevitably carry the stamp of mentalism. Locke’s theory of meaning contains elements that cannot be reconciled with each other so easily: his linguistic thesis according to which words as arbitrary signs are imposed by the mind on its ideas has a mentalistic and even solipsistic ring to it, yet communication is a social activity, governed by rules, customs, and conventions (Nauta, 2021, pp. 215–244). Therefore, the question left for us is: How does the social world shape our private minds?

For Locke, one’s mental state determines the meaning of his sentence. Since mental states determine the meaning of sentences, why does a person refer to different objects when the content of his mental state is the same? This is precisely the problem with internalism. It was pointed out by Searle that traditional internalism fails to address this problem because, in its framework, each intentional content is isolated from others, independent of any nonrepresentational capacities, and intentional states are not related to causality (Searle, 1983, p. 65).

But, in fact, mental states are interrelated; any one mental state must be accompanied by the presupposition of mental states in order for it to be reasonably explained. For example, my belief that “Jenny will return this book tomorrow” must be accompanied by the presupposition that I have (whether I am aware of it or not) mental states such as “Jenny exists,” “the book exists,” the memory of “Jenny borrowing the book,” and the belief that “Jenny will keep her promise.” Otherwise, we cannot explain my original belief. On the other hand, mental states must also be accompanied by the presupposition of nonrepresentational capacities in order for them to be reasonably explained. For instance, for the desire “I want to go to the botanical garden to see the flower exhibition” to be reasonably explained, I must be presupposed as having capacities that include knowing how to go to the botanical garden, knowing how to purchase the ticket, knowing how to find directions, and so on; otherwise, my desire cannot be fulfilled. These related mental states or capacities are theoretically inexhaustible because when we specify some related mental states or capacities, they may further presuppose other mental states or capacities. Furthermore, intentional states are causally self-referential. For example, the conditions of satisfaction for “I see that apple” are as follows: I see that apple, and my perceptual experience of “seeing that apple” is caused by the apple I see.

Since mental states are interrelated, the analysis of the intentional structure of meaning constitutes the core component of Searle’s theory of intentional meaning. Searle suggested that in performing a speech act, there are two levels of intentionality: one is the intentional state expressed during the speech act, and the other is the intention to perform the speech act itself. For instance, when I make the statement “It is raining outside,” I am not only expressing the belief “It’s raining outside” but also performing the intentional act of making that statement.

Hence, when a speaker says something and means something, he is performing an intentional act, and his production of the sounds is part of the conditions of satisfaction of his intention in making the utterance. But when he makes a meaningful utterance, he imposes conditions of satisfaction on those sounds and marks. In making a meaningful utterance, he thus imposes conditions of satisfaction on conditions of satisfaction. (Searle, 1998, p. 141)

It can be said that the primary, internal intentionality of the speaker’s intentional state is transferred to the words, sentences, or signs, which acquire intentionality derived from the speaker’s intentional state when being uttered meaningfully. The key to this derivation lies in consciously imposing conditions of satisfaction on conditions of satisfaction.

For instance, when I make the statement “The sun is setting,” it clearly expresses or reflects my belief about an external state of affairs. As previously stated, intentional meaning has two levels: one is the intentional state expressed during the speech act, which in this example is the state of my belief about the sun setting, and the other is the intention to perform the speech act itself, which in this example is me uttering the sentence “The sun is setting.” In other words, uttering this sentence constitutes part of my representational intention. Now, the uttered sentence itself also possesses conditions of satisfaction, that is, what is demanded: the sun is indeed setting.

It is noteworthy that the literal meaning of utterances in everyday life is often inaccurate. If we were to speak strictly according to the literal meaning, the conditions of satisfaction of our linguistic actions would not be clear. For example, when we say, “That cat is on the table,” is our assertion true if there is actually a three-millimeter gap between the cat and the table? Is it true if the statement is made in zero gravity? The satisfaction conditions of an utterance are relative to a certain context, such as the related physical environment, the physiological makeup of the subject, cultural norms, and so on. All these elements are nonrepresentational; they are not semantic in themselves, but they play crucial roles in discerning semantic content. Therefore, the same statement in different contexts may determine different conditions of satisfaction (Searle, 1979, p. 128, p. 133); if a certain context is not available, an utterance cannot determine a specific set of truth conditions.

The same applies to mental states. To possess a belief or desire, one must have an entire network of other beliefs and desires, as any conceptual content can be interpreted in various ways, and their meanings depend not on their own experiential content but on the presupposed background – an external criterion for “inner processes” that determines how an experience is to be interpreted; yet, this background is not part of the conceptual network (Searle, 1992, p. 176). If a person only knows his mental state, it is impossible for him to know his mental state’s conditions of satisfaction. Without knowing the conditions of satisfaction, he cannot know what mental state he is in. Only when given different backgrounds can one distinguish one’s mental state. The background is the set of stances, habits, skills, and practices that enable intentional contents to work in the various ways that they do, and it is in that sense that the background functions by providing a set of enabling conditions for the operation of intentional states. This means that the conditions of satisfaction of speech acts are collectively determined by the holistic mental state (Searle, 1983, pp. 153–154).

As mentioned earlier, language conveys meaning through intentionality, but how can “what I say” express “what I mean”? Searle divided a sentence into propositional and intentional contents: the intentional content refers to the meaning the speaker intends to convey through a sentence, while the propositional content refers to the content directly expressed in the sentence. The connection between the two arises from conventional concepts. Specifically, the speaker uses a sentence to convey his intention, and the listener uses the heard sentence to understand what the speaker intends to express, which may be either the literal meaning of the sentence or something else the speaker refers to using the sentence. Since the expression and interpretation of the sentence by the speaker and the listener are based on their respective knowledge, cognitive structures, and language usage, the speaker’s intention may differ from the intention inferred by the listener. Convention, however, bridges the two. For instance, in a poorly ventilated classroom, when a teacher says, “It is hot today,” his intention is for the student to open the windows, while the student may understand the implied intention and open the windows or simply perceive it as a statement about the weather. Regardless of how the sentence is perceived, it is clear that the realization of the teacher’s intention depends on their shared cognitive background, which is the conventional experience. Convention acts as a bridge between speakers and listeners.

Nevertheless, the background consists of mental phenomena. While the background derives from the entire network of relations inherent in every being that is both biological and social, without biological makeup and social relations, I could not have the background that I have. But all these relations are only relevant to the production of the background because of the effects it has on my mind (brain). Here, Searle implied, without explicitly stating, biological, social, and physical factors constrain my mind – hence the emergence of the entire background – ultimately through their internal influence on my mental state and intentional state. In this context, nothing external to the mind can act independently. In Searle’s view, the background is a mental phenomenon, rather than social or physical relations.

Consequently, the meaning of language is not only a matter of intentionality but also a matter of convention; it is jointly determined by intentionality and convention. However, conventions or social rules are realized through the intentional network and background, and biological, social, and physical relations can be associated with the background only because they can influence my mind (brain).

4.3 Emphasizing the Cognitive Accessibility and Epistemic Transparency of Semantic Content

A theory of meaning must not only explain how language refers to the world; it must also be able to give a clear account of individual behavior and psychological causal relations. Yet the environment-dependent factors posited by externalism (e.g., objective causal histories, communities of experts) are often inaccessible or opaque to the subject, which poses a fundamental difficulty for epistemology. If meaning depends on factors in the environment of which the individual is unaware or which are not accessible, then the subject, in effect, cannot ascertain what the meanings of one’s own words are; psychological explanation will face severe difficulties, thereby falling into the philosophical predicament of epistemic skepticism or semantic nihilism and contravening the basic function of language as a tool for expression and thought.

Against this background, only by understanding meaning as wholly internal to the subject’s cognitive structure can we coherently explain the language user’s behavior, reasoning, and psychological interaction. Semantic internalism does not deny the role of the external world; rather, it stresses that the language user’s cognitive activity must be grounded in psychologically accessible states. In other words, while external factors may influence the formation of meaning, the substantive constitution of meaning must ultimately be anchored at the level of the individual subject’s cognition, because the explanation of meaning maintains a direct and intimate connection with the subject’s cognitive capacities and mechanisms of mental representation. From the perspective of contemporary cognitive science, semantic content is grounded in a functional state – that is, a specific computational state realized by the nervous system. This state not only has a clear neurobiological basis but also implements information processing through intracerebral mechanisms and thereby supports higher cognitive functions such as human reasoning, understanding, and linguistic behavior. For example, studies indicate that the semantic activation of concrete concepts depends on modality-specific brain regions, whereas abstract concepts require the integration of social and affective dimensions (Hoffman & Bair, 2025). Taken together, these empirical findings reveal that semantics, in essence, reflects the specific representational forms that mediate between the brain’s internal structure and its information-processing operations.

One line of externalism’s critique of internalism holds that meaning depends on external factors such as expert definitions and the social community. Yet this stance faces a core difficulty, namely a crisis of general usability: if linguistic expressions must rely on expert knowledge (e.g., H₂O) in order to have meaning, then everyday communication (such as using “water” to refer to an experientially available liquid) is stripped of semantic validity. Such semantic elitism in effect nullifies the expressive authority of nonexpert language users.

The externalist view also generates a gap in causal explanation. What a social community supplies is only a normative framework rather than a semantic ontology; these external points of reference do not directly constitute meaning itself. For example, how is the social consensus “water is H₂O” concretely transformed into an individual’s motivation to drink water? Externalism finds it hard to provide a clear causal mechanism, because social norms operate at the group level and cannot directly explain individual behavior. By contrast, such phenomena as misunderstanding, divergence, and creative usage (e.g., poetic metaphor) among language users indicate that an individual’s internal semantic system possesses a relative independence. Linguistic communication depends on the cross-situational stability of meaning; if meaning cannot be sustained by internal structure, then cross-context understanding, translation, and inference all become unreliable.

While affirming the important role of linguistic communities and social norms, internalism places greater emphasis on the fact that these external factors exert genuine influence only insofar as they are cognitively accepted and psychologically reconstructed by the individual. Its core claim is that the ultimate effect of any external influence is rooted in the subject’s internal cognitive states. In my view, the interaction between individual psychological construction and social norms can be coordinated through the following three mechanisms:

• Internalization of norms: Social norms take effect only after being understood, adopted, and mastered by the individual (e.g., learning the medical definition “arthritis = inflammation of the joints”); this process depends on updating mental representations and reorganizing the network of beliefs.

• Dispositional states: Even if one accepts “meaning is use,” its essence still lies in the dispositional states an individual exhibits in particular contexts (Anjum & Mumford, 2011) – for instance, the tendency to say “Please give me water” when thirsty.

• The cognitive basis of shared intentionality: Even expert-dependent terms (such as “gold”) derive their efficacy from the linguistic community’s acquisition mechanisms, which convert social reference into an individual’s default cognitive state (Wolf & Tomasello, 2025).

Evidently, internalism does not deny social factors; rather, it delineates the boundaries of their efficacy and establishes the priority of causal explanation: the proximate cause of behavioral explanation must point to internalized mental states, not to external facts. The stability of meaning rests on a dual safeguard: on the one hand, cross-context semantic consistency requires the modulation provided by social norms; on the other – and more crucially – it requires the autonomy of the individual cognitive system, thereby ensuring that meaning remains durably stable at the psychological level.

4.4 Upholding Nonessentialism about Meaning in Theories of Reference

Externalism typically presupposes an essentialist stance about meaning – namely, that meaning is determined by the objective environment or by the intrinsic nature of entities. Internalism, by contrast, upholds a nonessentialism about meaning. Nonessentialism holds that meaning has no single, fixed, and immutable essence. That is, meaning is not defined by some unique, external, objective property but is constituted by a set of diverse and complex conditions and by actual linguistic use. The meaning of a word is knitted together by many intersecting and overlapping uses, rather than unified by some essential feature.

Nonessentialism denies that there is a necessary, essential linkage between words and external objects (as in Kripke’s “rigid designation”) and thereby weakens the monopoly of public language over meaning. In other words, meaning is not determined by some external, objective fact (such as chemical composition or physical property). For example, “water” does not possess a fixed, unchanging essence of meaning that is necessarily bound to H₂O. If so, then the essential properties of the external world cannot determine meaning, and this opens space for the view that “meaning originates within the individual.” In response to Kripke’s critique of traditional descriptivism (e.g., that “Einstein” does not depend on any single description), we can appeal to the cluster theory: reference is fixed by a set of descriptions (rather than a single description), and these descriptions arise from an individual’s internal cognitive network. Even if some descriptions are mistaken (e.g., “Einstein invented the telephone”), so long as sufficiently many descriptions converge on the same object, reference can still be fixed (Costa, 2011).

According to nonessentialism, the meanings of natural-kind terms cannot be determined solely by a material base; rather, they are manifested in the processes of linguistic use and analysis. The relevant cognitive mechanisms include lexical retrieval, syntactic parsing, and broader language processing, all of which occur within the language user. In the case of terms such as “water” and “lion,” their meanings can neither be fully reduced to physical attributes (e.g., H₂O, a felid of the genus Panthera) nor be exhausted by a fixed reference; on the contrary, they are always shaped by sociocultural and other contextual cognitive factors. The phenomenon of copredication, supported by inherent polysemy, also provides strong evidence (Murph, 2023). For example, in the statements “Lead is a heavy metal” and “Lead is a poisonous metal,” the term “lead” functions with different aspects of meaning: in the former sentence it is characterized by its physical properties, whereas in the latter it is characterized by its health-related properties. These two descriptions (“heavy” and “poisonous”) appear independent, and each relies on different attributes, yet both jointly target the same concept – lead. That is, the meaning of “lead” here does not depend on some concrete external environment or physical property as such but becomes polysemous in light of how we use the terms and how we cognize these properties.

Internalism acknowledges that natural kinds have essential properties, but it sharply distinguishes the essence itself from the mental representation of the essence, emphasizing that the latter is decisive for determining meaning. Its key move is to resist externalism’s reliance on “essence”: only when an essence is represented by an individual as a theoretical belief (e.g., “water has a certain microstructure”) does it play a role in the constitution of meaning; the essence itself does not directly determine the meaning of a term. Within this framework, the meanings of natural-kind terms can be constituted by people’s mental representations of prototypes or exemplars, without depending on objective essences in the external world. For instance, the meaning of “tiger” derives from our cognitive schema of striped fur, ferocity, and feline characteristics, rather than from its DNA. Even if it turns out that the “tiger” before us is in fact a robot, so long as it fits the foregoing cognitive schema, the term’s meaning would remain in place.

From a nonessentialist perspective, meaning is not fixed and unchanging but realized through multiple uses and rules across different contexts. Hence, to understand a word is precisely to understand how people use it in psychological and linguistic practice. This diversity and flexibility can be adequately explained only by reference to the concepts and rules in the speaker’s mind. Accordingly, internalism construes semantic determination as a process of interaction between the cognitive subject and the language system, rather than as a static reflection of the essence of things. In contrast to externalism’s rigid essentialist stance, internalism highlights the constructivism, context-sensitivity, and openness of meaning, thereby affording a deeper explanatory power for meaning.

In sum, when confronted with the criticisms of semantic externalism, semantic internalism does not adopt a rigidly oppositional stance; rather, it answers externalist challenges in a stepwise manner by reappealing to the classical distinction between sense and reference, underscoring the fundamental status of psychological content as the source of meaning, reaffirming the principle that individual cognition has access to the contents of one’s own thoughts, assessing the scientific operationalizability of explanations of meaning, and delineating the applicable scope of social factors. By virtue of this defensive strategy, semantic internalism forcefully safeguards its theoretical foundation – namely, the core principle that meaning is determined by the subject’s internal psychological states and cognitive structures.

5.1 Distinguishing Types of Semantic Content and Developing a Theory of “Narrow Content”

Traditional semantics has often treated “meaning” as a single-dimension object of inquiry (e.g., truth-conditional or compositional semantics). With the deepening of research, however, an increasing number of contemporary scholars have come to recognize that meaning comprises both literal and implicit semantic structures and encompasses socially shared cultural connotations as well as individualized affective projections; it is therefore multiple, open-ended, and variable. Hence the notion of a “multilayered semantic structure” has gradually emerged. It holds that linguistic meaning is not monolithic but a complex structure jointly constituted by several layers that differ in nature and function. Among these layers, two are especially crucial and far-reaching in impact: the cognitive semantic layer, which is tightly connected to the human conceptual system and the world of experience and is used for reasoning, memory, and the expression of intentions; and the communicative semantic layer, which focuses on how meaning is realized within socially shared practices, interpersonal interaction, and contextual exchange. The cognitive semantic layer provides the “content” or “material” basis of meaning: it concerns how we organize and understand the world through language. The communicative semantic layer, by contrast, concerns the meanings language carries in actual interaction so as to realize specific communicative aims.

A paradigmatic example is Kasia M. Jaszczolt’s treatment of the construction of utterance meaning as a process of modular integration, whose sources include four principal layers: (1) literal meaning, generated by grammatical analysis and lexical composition; (2) world knowledge, covering common sense, cultural background, and social conventions; (3) contextual clues, including the conversational setting, pragmatic implicatures, prosodic modulation, facial expressions, and other multimodal cues; and (4) speaker intentions, namely what the speaker aims to convey as their real import. Jaszczolt argues that, in the absence of specific contextual interference, hearers preferentially opt for the most typical and economical construals – default interpretations. On this basis, she proposes Default Semantics and further distinguishes two kinds of defaults: cognitive defaults, grounded in universally human cognitive tendencies (such as relevance and economy), and social-cultural defaults, which depend on the shared knowledge of particular cultures or groups (Jaszczolt, 2016). By drawing this distinction, Jaszczolt underscores the joint operation of cognitive mechanisms and sociocultural factors in discourse comprehension.

Building on this line of thought, we can develop a theory of “narrow content” and use two-dimensional semantics (Two-Dimensional Semantics) to refine internalism.

Narrow content and wide content are two opposing notions that describe the scope of the content of mental states (such as beliefs and desires). Narrow content refers to the meaning of an expression in a given context that depends only on the speaker’s internal psychological states and linguistic knowledge, and not on specific facts about the external world. It seeks to explain the “cognitive significance” of linguistic expressions – namely, why certain expressions can retain the same cognitive import across different possible worlds – and is primarily employed to account for action, thought, and reasoning (Kriegel, 2008). Suppose someone has the concept “water” in mind: regardless of whether, in the world they inhabit, water is H₂O or XYZ, so long as their psychological states are the same, that narrow content does not change. Wide content, by contrast, includes the actual reference of words and depends on the individual’s relations to their environment (e.g., causal history, social-linguistic practices); it is suited to social communication. Continuing with “water”: if in one world “water” refers to H₂O while in another it refers to XYZ, then their wide contents differ, even if the relevant psychological states are similar.

Wide content emphasizes the public character of language and thought, but it struggles to account for the independence and causal efficacy of the individual’s internal psychological processes. For example, when a person drinks water because they are “thirsty,” their motive for action depends solely on their internal state, whether or not real water is present in the environment (as in a hallucination). It follows that defining semantic content solely by appeal to external factors (such as history or environment) cannot capture the autonomy and regularities of the individual’s mental activity, nor can it ground stable psychological laws. By contrast, the independence of internal states is more clearly articulated within a theory of narrow content. Narrow content supplies robust psychological objects and thus supports effective research in the psychological sciences and causal explanation. Even hallucinatory experience has content that is constructed by the subject’s internal representational system. As Descartes’ evil-demon hypothesis reveals, although external perception may be wholly deceptive, the subject’s internal cognitive structure still provides the basis on which experience acquires meaning.

Corresponding to narrow and wide content are the Two-Dimensional Semantics posits for linguistic expressions: primary intension and secondary intension. Primary intension – also called epistemic intension – reflects the cognitive possibilities associated with a term and is tied to the speaker’s a priori knowledge and conceptual capacities; it is equivalent to narrow content. For instance, the epistemic intension of “water” is “a colorless, transparent, thirst-quenching liquid” (a descriptive profile): even if “water” picks out different substances (such as XYZ) in different possible worlds, its cognitive role remains unchanged. Secondary intension – also called metaphysical intension – is the expression’s actual referent: once we fix what the term designates in the actual world, the term designates that very thing in other possible worlds, whether or not the thing there retains the same properties; this corresponds to a particular kind of wide content. Thus, if in the actual world water is H₂O, the secondary intension of “water” rigidly designates H₂O in all possible worlds (Chalmers, 2004).

The theory of narrow content furnishes a more fine-grained framework for defending semantic internalism: we acknowledge that metaphysical intension is influenced by the environment, but the subject’s understanding and use of words rely first and foremost on epistemic intension. External factors affect only how reference is realized, whereas primary intension is the core of meaning. What truly determines the meaning language has for the individual is the structure of psychological content. Two-dimensional semantics accommodates both “narrow content” (internal) and “wide content” (external) while maintaining that narrow content is a necessary condition for meaning. By clearly distinguishing epistemic from metaphysical intension, we can uphold the thesis that “meaning originates in the mind” and still explain the externality of science and reference. In other words, what the subject relies on in communication and reasoning is the internal cognitive meaning – precisely the locus of internalist concern.

5.2 Grounding Intentionality in Conscious Experience

In twenty-first-century philosophy of mind, an important research program is Phenomenal Intentionality Theory (PIT). Its core claim is that the intentionality of mental states is ultimately rooted in, and even wholly determined by, their phenomenal character – the “what-it-is-like” aspect of experience. Put simply, a thought can be directed at or about an object (e.g., thinking “that tree”) because having that thought is accompanied by a distinctive subjective experience. Proponents of PIT – such as Terence Horgan, Uriah Kriegel, and Angela Mendelovici – have articulated a radically internalist picture of the mind (Horgan, 2024, pp. 461–482).

At its theoretical core, research on phenomenal intentionality offers a highly attractive philosophical basis for semantic internalism. Its justificatory force is chiefly evident in two respects: (1) internal determination, and (2) determinacy and subjectivity of content.

Internal determination. The most fundamental source of intentionality lies in the phenomenal features of consciousness. Phenomenal features (such as the “burning feel” of pain or the visual qualia of color) themselves carry aboutness, without relying on external causal history. For example, a “red experience” in hallucination still exhibits directedness (it is about red), even when there is no corresponding physical object. Phenomenal character – the qualitative feel of consciousness – is widely regarded as an individual’s most private and internal attribute. If intentionality is grounded in phenomenal experience, then intentionality itself must likewise be internal. This stance directly challenges the Naturalistic-Externalist Research Program (NERP) that dominated late twentieth-century theorizing, which sought to explain intentionality through causal, informational, or evolutionary relations between minded subjects and the external world.

Determinacy and subjectivity of content. Advocates of PIT hold that phenomenal experience can furnish thoughts with determinate content. Such determinacy derives from the immediacy and inwardness of subjective experience rather than from unstable external causal chains. If the content of the thought “water” is fixed by the inner experience of “what it is like to think of water,” then the environmental differences posited in the Twin Earth thought experiment (H₂O vs. XYZ) do not affect the thought’s core content. The Earthling and his twin can fully share the same thought content determined by their internal phenomenology. This constitutes a direct rebuttal to semantic externalism.

Accordingly, semantic internalism secures an ontological foundation rooted in the study of consciousness – precisely what traditional defenses of semantic internalism have often lacked. PIT distinguishes two levels of intentionality: original intentionality – conferred directly by phenomenal experience – and derived intentionality – arising from external factors such as social pragmatics and environmental semantics. The theory emphasizes that the intentionality possessed by phenomenal states is original, whereas the intentionality of pragmatic or other external representations is derived. In other words, Original over Derived (Bourget, 2010). This implies that our understanding of semantic content should first be grounded in subjective qualia, rather than in environmental conditions or social practices. Even when external factors influence semantics, their role is limited to constraining or revising phenomenal intentionality; they are not the fundamental source that determines meaning.

The phenomenal character of consciousness also receives support from neuroscience and experimental psychology. Studies on multimodal representation and perception–action simulation show that semantic comprehension is accompanied by the activation of specific sensorimotor patterns; even in the absence of external correlates, subjects can retain content (e.g., color concepts in blind patients; object recognition in hallucination), with conscious experience associated with particular neural patterns (Barsalou, 2008; Dijkstra, Bosch & van Gerven, 2019). For instance, research on patients with brain damage (such as visual agnosia) and on virtual-reality paradigms indicates that grasping meaning depends on modality-specific activations of internal representations (e.g., simulations in the visual cortex for “apple”) rather than on external referents (Bottini et al., 2020). This moves the “internal accessibility” of meaning beyond mere philosophical intuition and brings it underwritten by empirical findings.

Phenomenal intentionality thus indicates that, although we perceive parts of the external world, our understanding of it is always mediated by inner experiential frameworks. An individual’s inner conscious experience possesses the capacity to construct semantic content independently. Our direct access to the contents of our own thoughts is achieved through reflection on present experience. This means our understanding of language and symbols is deeply dependent on the subject’s inner experience and perception. The content of a given mental state is manifested in the distinctive qualitative feel (what-it-is-like) that accompanies it; and this feel arises from the subject’s psychological state – it is internal and immediately generated. This aligns closely with the “first-person access” emphasized by semantic internalism and provides it with potentially powerful support. If an individual’s perception is internal and shaped by the qualitative features of experience, then the intentional content of that perception should, naturally, be determined by those internal features.

5.3 Appealing to the Listener’s Interpretation

We know that the primary aim of a theory of meaning is to elucidate the meaning of an expression uttered by a speaker in successful communication. Davidson, however, shifted to a different approach when answering this question. Rather than appealing to the speaker’s understanding and use of language like other philosophers, he attended to the understanding of the listener. According to Davidson, meaning is not an independently existing object; when we say that language is meaningful, we are merely stating that it can be understood; if the elements necessary for understanding language are sufficiently explained, then the meaning corresponding to language is also explained. In short, a theory of meaning is a theory of interpretation. Rejecting the objectification of meaning marks a clear departure from the philosophical tradition spanning from Locke’s empiricism to Frege’s theory of meaning.

For Davidson, words or sentences have meaning because people can understand them and use them to communicate with others; conversely, if one or a string of signs, or sounds, cannot be understood by people, then it is not meaningful. Davidson proposed to use Tarski’s notion of truth as the foundation for the study of semantic theories. He borrowed and modified Tarski’s Convention T to meet the formal requirements of a reasonable theory of meaning (Davidson, 2001a, pp. 23–24). With this approach, “being true” is something that can be understood by others. With regards to the sentence “Snow is white,” for example, Tarski first assumed that the meaning of the sentence is clear, and the question is why this sentence is true. In contrast, Davidson focused on what “Snow is white” means and asked how to interpret this sentence. The reason why the truth theory can be used to explain a specific sentence is that we not only have all the evidence about other T-sentences but also know that the sentence’s T-sentence is interpretable. In short, meaning must be closely bound with the ability to “understand” and “interpret” language because the latter meets the requirement of an ability shared by the speaker and the interpreter.

The focus here is not on how the speaker understands the sentence but on how the listener understands it. A more extreme case is how two people speaking different languages understand or interpret each other. One might answer that this could be achieved using dictionaries about these two languages, or what Quine calls a “translation manual.” Davidson, on the other hand, was concerned with how the two people interpret each other in the absence of a translation manual. This involves what he called “radical interpretation.” Radical interpretation refers to interpreting a speaker’s utterance without presupposing any knowledge about the meaning of the speaker’s utterance or his attitude. It relies on evidence that takes neither the form of knowledge about meanings nor the form of detailed knowledge about beliefs (Davidson, 2001a, p. 135).

People may use the same language to understand each other, but whether they express the same meaning by using the same language can only be answered through questions and answers in a dialogue. The meaning of a speaker’s utterance is the result of interpretation obtained through linguistic communication. As Davidson put it:

For two people to know of each other that they are so related, that their thoughts are so related, requires that they be in communication. Each of them must speak to the other and be understood by the other. They don’t, as I said, have to mean the same thing by the same words, but they must each be an interpreter of the other. (Davidson, 2001b, p. 121)

The meaning of the speaker’s utterance is not a priori but generated in linguistic communication. As Davidson argued, “meaning is what we can abstract from accomplished verbal exchanges” (Davidson, 2005, p. 258); it is the product of specific linguistic communication. The linguistic communication between the interpreter and the speaker forms the basis of the interpretative activity.

That is why Davidson situated the act of interpretation within a triangular interaction among the speaker, the interpreter, and the world. The picture of interpretation demonstrated by the triangulation mode shows that understanding language is an activity of intersubjective communication, and meaning gradually emerges as a result of dialogue and negotiation among communicative subjects. The success of linguistic communication lies not in speaking a common language but in achieving consensus through interaction between subjects. The triangulation mode fundamentally rejects the notion of meaning as a matter of convention. Viewing language as a theory of convention or rules merely describes linguistic behaviors without interpreting or justifying their legitimacy. With his triangulation mode, Davidson attempted to explain how the interpretation and understanding of language are possible and provided evidence for this possibility. He opposed the idea of treating the meaning of language as something that existed before linguistic communication, and emphasized the priority of interaction between subjects in the determination of meaning. It is through the intersubjective relationship constituted by the dialogue between the speaker and the interpreter that the objects or events in the external world referred to by the discourse are confirmed. This, in turn, allows the meaning of language to be interpreted.

It is worth noting that in interpretation, truth, belief, and meaning are intertwined. Only when the listener understands the speaker’s intentions and assigns truth values to the speaker’s utterance can the listener correctly interpret the utterance. The interpreter’s assignment of truth values and beliefs to the utterance of the speaker is a necessary aspect of language interpretation. In other words, a speaker holds an utterance true because of the meaning of the utterance he made and his own belief. For example, when someone says, “Snow is white,” he must have an understanding of what “snow” is and believe that what he sees is indeed snow. In interpretation, if all we have to go on is the fact of honest utterance, we cannot infer the belief without knowing the meaning, and we have no chance of inferring the meaning without the belief (Davidson, 2001a, p. 142). As Davidson argued, “[W]e should think of meanings and beliefs as interrelated constructs of a single theory just as we already view subjective values and probabilities as interrelated constructs of decision theory” (Davidson, 2001a, p. 146).

From this perspective, the status of belief in interpretation ensures consistency and mutual understanding between the interpreter and the interpreted, as it requires that in every possible world where each cognizant individual encounters the same sensory evidence, the interpreter and the interpreted should be assigned the same belief. In other words, in every possible world, if a cognizant individual is assigned a belief based on certain evidence, the same belief should be assigned to any other cognizant individual in that world who is familiar with the same evidence. Meaning belongs to a network of human beliefs and is also a way of weaving this network.

However, this viewpoint faces clear challenges. First, it fails to account for perception: unlike beliefs, desires, and other mental states, sensations such as itchiness, pain, cold, and heat do not seem to point to an object. Externalism, therefore, excludes perception from mental content. Second, and more crucially, its refutation of internalism is untenable: externalism attacks the nature of the conditions under which the speaker’s views or opinions are satisfied. Here, the subject (the speaker) is singular and continuous, instead of two or more speakers. The Twin Earth argument has a crucial oversight: whether two things are the same depends on who is considering them. In this case, the word “water” means the same thing for residents of Twin Earth and Earth; however, for an observer aware of the differences between them, their meanings are different. While the thought experiment presupposes that Earthlings and Twin Earthlings share the same mental state, their actual mental states differ from each other, as the former think of H₂O, while the latter think of XYZ. Given the different mental states, Putnam’s argument fails because what he sought to demonstrate is the different reference of “water” on Earth and “water” on Twin Earth under the assumption of identical mental states. In fact, whether on Earth or Twin Earth, a person cannot explain whether they mean H₂O or XYZ when unaware of the difference, yet they still express different meanings.

According to externalism, regardless of whether a language user is aware, the meaning conveyed by his vocabulary will change depending on changes in the external world. This is perplexing. Let us consider the following scenario: Tom sees a snake in the distance, and indeed, there is a snake. His experience is real, and the word “snake” expresses the meaning of a snake. Now, consider another situation: Tom is mistaken; he thinks he sees a snake, but it is actually a model someone made from grass rope. Logically, we would say Tom is mistaken; his experience is incorrect because what he sees is not factual. However, if we accept semantic externalism, we must believe that although Tom sees a snake in front of him, even though his mental state is the same as that in the previous, genuine experience, the external world has changed, and the meaning of “snake” has also changed accordingly. In other words, what he refers to as a snake actually conveys the meaning of the grass rope. Therefore, when Tom says he sees a snake in front of him, he is correct. Clearly, externalism puts us in a dilemma where we no longer have the illusion that our mental content is reality.

The attack from externalism misses the core thesis of internalism, which argues how subjects use language signs to connect with the external world, thereby satisfying the conditions set by the mind. It is evident that whether it is water on Earth or Twin Earth, whether it is a snake or a grass rope, the two things are essentially different, but this fact does not affect the internal things that communicators can share. In everyday life, we often encounter similar situations. For example, person A says, “The Bordeaux wine we had last night was excellent,” and person B responds, “Yes, the German wine we had last night was really good.” One day, A and B meet again, both wanting the same wine as last time, and they both point to the bottle in the oak barrel. At this moment, the shopkeeper of the wine store replies, “This is Royal Australian Red …” The three individuals refer to different things that exist in the real world, but the internal things they refer to are the same, and they engaged in successful communication through verbal expression.

Conversely, even the same word may still lead to ambiguity if thoughts in the mind are not involved. For example, when a person in Hong Kong calls a woman “Miss,” (Xiaojie in Mandarin Chinese), the other person responds politely; however, the same term may provoke an angry reaction in China’s mainland, as the term can also be used to mean a prostitute in China’s mainland. According to semantic externalism, the meaning of “Miss” changes due to different social environments. In fact, such an explanation is overly simplistic and general. Specifically, the meaning of “Miss” in the minds of people in China’s mainland differs from that in the minds of people in Hong Kong. While the word “Miss,” a combination of sound and signifier, is just a combination of lines and soundwaves, humans use it for the communication of thoughts. However, due to habits and culture, people in different historical periods and different people in the same historical period have different thoughts about this particular sign, leading to unavoidable misunderstandings. Therefore, reasonable mentalism should be a refined mentalism, meaning that the meaning of language should refer to the thoughts in the speaker’s mind matching those in the listener’s mind.

In this sense, I agree with Davidson’s proposition because it is the mental state in the minds of both the speaker and the listener that enables them to understand the reference of language. Belief and truth play significant roles in meaning. We have some basic beliefs about the world that constitute the source of the meaning of the linguistic expressions we use. When a speaker utters a sentence in a communicative context, what he intends to refer to is clearly one of his beliefs. Therefore, when a listener claims to understand the meaning of a speaker’s utterance, what he understands is one of the beliefs held by the speaker. Then, how does the listener represent the speaker’s beliefs? According to Davidson, when the listener interprets the utterance made by the speaker, he is not translating the utterance into an utterance in the language he uses. In fact, interpretation is fundamentally an ascription of beliefs, that is, the listener ascribes a certain belief to the speaker. This belief can be represented by several other beliefs that the speaker possesses, and these other beliefs are used by the listener as the conditions of truth value in relation to the speaker’s belief. This belief would still be public even if it is subjective, as it is constrained by both an utterance regarded as true by private individuals and an utterance regarded as true by the public.