Direct Reference and Rigid Designation

The Problem of Reference Across Possibilities

The central question of a theory of reference is how a linguistic expression places thought and assertion in contact with a particular object. The question acquires a distinctive structure when we consider alternatives to actual circumstances. A speaker calls an individual by a proper name and asks whether that very individual might have had different parents, chosen another occupation, or never acquired the properties by which the speaker currently recognizes them. The name appears to preserve the subject of evaluation while the imagined circumstances vary. This phenomenon is the intuitive setting of rigid designation.

Direct reference raises a related question about what the name contributes to the assertion. When a speaker says that Ada was a mathematician, a semantic theory may treat Ada herself as a constituent of the proposition expressed. An alternative theory may treat an identifying condition, or some mode of presentation of Ada, as the semantic contribution. Both theories can permit the speaker to discuss the actual individual. They disagree about the structure of what is said, the relation between thought and object, and the explanation of cognitive significance.

The two issues become difficult when reference is learned under conditions of ignorance. A person may name a planet without knowing that it has already been named under another description. The same object can enter separate patterns of belief, discovery, and communication. An expression may be semantically stable across counterfactual circumstances while its users entertain contradictory judgments about its bearer. Any comprehensive theory must account for both stability and epistemic difference.

The historical debate therefore reaches beyond names. It concerns identity, scientific discovery, necessity, de re thought, the social transmission of linguistic conventions, and the boundary between semantics and pragmatics. The modern developments associated with Kripke and Kaplan made these dimensions systematically visible, while the problems motivating them can be traced through nineteenth- and early twentieth-century investigations into meaning and logical form (Frege, 1892/1948; Russell, 1905; Kripke, 1980; Kaplan, 1989).

Four Explanatory Questions

Reference is the relation by which an expression, on an occasion or under an interpretation, is associated with an object, kind, or other semantic value. Four explanatory questions are frequently compressed into a single theory. First, what establishes the referent of an expression in the actual linguistic practice? Second, what does the expression contribute to the truth conditions or structured proposition expressed? Third, how does the expression behave when the truth of that proposition is evaluated in counterfactual circumstances? Fourth, how does a thinker recognize, retain, and communicate information about the referent?

These questions have different logical types. Reference determination is a metasemantic question about why an expression bears its semantic relation. Semantic contribution concerns content: whether a name contributes an object, a descriptive condition, or a complex representational structure. Rigidity concerns modal behavior under an assignment of reference. Cognitive access concerns a subject's capacity to think, infer, identify, and coordinate with others. Their answers constrain one another, but no single answer follows simply from the use of a word such as reference.

In a standard regimented account, a name is directly referential if the object itself, rather than a descriptive condition identifying it, is its contribution to a singular proposition. A designator is rigid when it designates the same individual relative to all relevant possible-world evaluations, assuming an interpretation held fixed. A historical theory may explain how a name was introduced and transmitted without specifying a unique theory of propositions. A theory of mental files may explain how a speaker stores information about an object without claiming that all information stored is semantically constitutive of the name (Recanati, 1993; Dickie, 2015).

This fourfold separation supplies the organizing principle of the investigation. It avoids the tempting inference that because reference is causally grounded, propositional content must be direct, or that because an expression is rigid, it lacks all cognitively significant meaning beyond its bearer. The crucial research task is to establish exactly which claims are independently supported by which arguments.

Historical Precursors: Mill and the Logic of Names

John Stuart Mill distinguished names that denote a subject while also connoting attributes from proper names that ordinarily function as labels of individuals. In Book I of A System of Logic (1843), a concrete general name such as human can signify a class through its associated attributes, whereas a proper name such as John serves principally to distinguish its bearer. The distinction has proved influential because it presents a conception of naming in which descriptive information can be associated with an individual without forming the semantic content of the individual name.

Mill's connotation and denotation belong to a nineteenth-century account of terms, predication, and induction. Equating his distinction immediately with contemporary structured-proposition semantics would obscure important developments. Mill did not furnish Kaplan's framework of characters, contexts, and propositional constituents, and the modal notion of rigidity was subsequently articulated with the resources of modern modal logic. Nevertheless, the Millian idea that a name's linguistic function can be exhausted, in an important respect, by its relation to its bearer provides a recognizable ancestor of modern direct-reference positions (Mill, 1843, Book I, ch. II; Recanati, 1993).

The historical significance of Mill also lies in his account of the separation between naming and the information people possess. A community may learn substantive facts about an individual while retaining the same proper name. Ordinary naming practices allow initial ignorance, correction, misinformation, and disagreement. This flexibility would be mysterious if successful name use required each speaker to possess one shared, uniquely identifying description.

Yet a purely label-based model leaves questions unresolved. If two names refer to the same person, why can learning the identity of their bearers constitute discovery? Why does an utterance employing one name prompt a different cognitive response from an utterance employing the other? The subsequent Fregean challenge exposes the limits of treating reference as the sole dimension of meaning. Thus the historical trajectory from Mill to Frege is driven by a genuine explanatory demand rather than a simple change in preferred terminology.

Frege: Sense, Reference, and Cognitive Value

Gottlob Frege's Über Sinn und Bedeutung appeared in 1892 in volume 100 of Zeitschrift für Philosophie und philosophische Kritik, pages 25–50. The German terms Sinn and Bedeutung are commonly rendered sense and reference, though Bedeutung also has a broader literal range and has been translated as meaning or denotation. Frege begins with the difference in cognitive value between self-identity and an informative identity assertion. If the names a and b have the same bearer, why is discovering a = b often an advance in knowledge whereas a = a is immediately recognizable? (Frege, 1892, pp. 25–27; Black translation, 1948, pp. 209–230.)

Frege's proposal distinguishes the object designated from the way in which it is presented. Two expressions can have the same Bedeutung while differing in Sinn. The morning star and evening star expressions give different routes to the same planet. The difference in cognitive value is explained by the difference in modes of presentation, not by a difference in the object itself. In the original argument, the point is embedded in Frege's broader treatment of signs, complete thoughts, and the reference of sentences as truth values.

This strategy has extensive philosophical consequences. It assigns to sense explanatory work in competent understanding and in the intelligibility of discoveries. It also makes possible an account of expressions whose sense is grasped while their reference is absent or unsettled. When attitude contexts are considered, an expression may be associated with an indirect reference, commonly the sense it ordinarily expresses. Such distinctions were meant to explain how belief reports can behave differently from extensional contexts.

Direct-reference theories confront Frege's challenge at their most vulnerable point. If two coreferential names contribute only the same object to a proposition, then, on a simple object-plus-predicate account, substituting one for the other leaves the singular proposition unchanged. But a rational subject can accept one assertion and reject the other through ignorance of identity. Frege's argument establishes a demand for an account of cognitive difference. Whether it conclusively requires sense as a semantic constituent is the contested step. Pragmatic, relational, and mental-file explanations attempt to satisfy that demand while preserving direct reference.

Russell: Descriptions, Acquaintance, and Logical Form

Bertrand Russell's On Denoting, published in Mind in 1905, transformed the treatment of apparent reference by using quantificational analysis. A definite description of the form the F, on Russell's analysis, does not automatically contribute a single object as a constituent of the proposition. A sentence asserting that the F is G can instead be analyzed as asserting that some x is F, that exactly one thing is F, and that this x is G. Its truth conditions therefore involve existence, uniqueness, and predication (Russell, 1905, pp. 479–493).

The analysis explains why a sentence containing an apparently referring description can be meaningful even when the description fails to denote. The traditional example of the present king of France illustrates how grammatical surface can mislead about logical form. Negative existentials and scope-sensitive contexts become tractable once the sentence is given quantificational structure. The analysis also produces a powerful contrast between ordinary descriptions and singular referring expressions.

Russell's wider epistemology of acquaintance complicates a simple historical classification. His theory of logically proper names granted direct access to objects of acquaintance, while many ordinary names were treated as abbreviated descriptions for purposes of analysis. Thus Russell is simultaneously a source for the modern idea of object-involving propositions and a central figure in the descriptive analysis of everyday naming (Russell, 1905; 1912).

The philosophical legacy is methodological. An apparent referential expression need not possess a referent as part of its truth-conditional contribution. We must examine semantic composition rather than inferring semantic structure from grammatical shape. Later proponents of direct reference generally reject the reduction of ordinary names to descriptions while retaining Russell's crucial insight that semantic theories must state precise truth conditions and distinguish referential terms from quantificational constructions. This methodological inheritance explains why debate with Russell remains central even where his account of ordinary names is rejected.

The Descriptivist Inheritance

A descriptivist theory associates a proper name with identifying conditions that determine its bearer. There are several importantly different versions. In simple synonymy theories, a name is equivalent in meaning to a particular definite description. Cluster theories associate a name with a weighted body of commonly accepted descriptions. Reference-fixing theories allow a description to identify the actual referent at an introduction stage without making that description synonymous with the name. Epistemic theories treat descriptive knowledge as essential to successful identification while leaving the semantic theory itself relatively open.

John Searle's 1958 paper Proper Names challenged the idea that a name is conventionally associated with one definite description. Searle instead considered a cluster of descriptions associated with the use of the name, explaining how speakers can continue to use a name after some beliefs about its bearer prove mistaken. This is a more flexible position than the thesis that Aristotle means the teacher of Alexander, and it requires a correspondingly more careful critique (Searle, 1958, pp. 166–173).

A descriptive account is attractive because it can answer a practical question: why does a speaker mean this object rather than another? Associated information can guide recognition, differentiate names, and explain an individual's inferential patterns. Descriptions may also be essential to the context in which reference is first established. An astronomer can intend to name whichever body causes an observed perturbation, even if the body's nature is still unknown.

Its principal difficulty arises when the identifying conditions are mistaken or contingent. If the bearer might have lacked the properties specified, the name seems still to concern that bearer in counterfactual evaluation. Furthermore, a whole community may be largely ignorant about an entity while still successfully talking about it. A successful critique must therefore target the thesis that descriptions supply the modal semantic content of names, rather than denying the pervasive role of descriptive information in language use.

Donnellan and the Uses of Descriptions

Keith Donnellan's Reference and Definite Descriptions, published in The Philosophical Review in 1966, distinguished attributive and referential uses of definite descriptions. In an attributive use, a speaker says something about whoever uniquely satisfies the description. In a referential use, the speaker employs the description to help an audience identify a particular individual about whom a claim is made. A speaker might use a mistaken description at a gathering to indicate a visible person, and nevertheless succeed in communicating a thought about that person (Donnellan, 1966, pp. 281–304).

This distinction reveals a dimension of ordinary speech that a purely quantificational analysis may underdescribe. The object intended by the speaker can be recoverable even when the descriptive condition is incorrect. Yet the phenomenon admits competing explanations. The description may retain Russellian truth-conditional semantics while pragmatic intentions determine the individual to whom the speaker means to refer. Alternatively, some referential uses may supply genuinely singular semantic content. Donnellan's observations alone do not determine which formal account is correct.

The significance for direct reference is substantial. They show that a speaker's communicative success and the conventional semantics of an expression can come apart. Since any semantic theory must accommodate ordinary communication, the theory cannot treat grammatical reference and speaker intention as unrelated subjects. At the same time, the distinction is a warning against inferring the semantics of an expression type from one favorable conversational example.

Donnellan also investigated proper names and identifying descriptions in 1970. Together, these works helped shift attention from idealized relations between words and qualities toward situated speech acts and causal-historical chains. Kripke's later distinction between speaker's reference and semantic reference makes explicit a critical response to interpreting all successful reference as a semantic property of the words employed.

Kripke's Reconstruction of Naming

Saul Kripke's Princeton lectures of 1970 were first published as Naming and Necessity in 1972 and later appeared as the book Naming and Necessity in 1980, with a new preface. These different dates should be preserved: the lectures preceded the article version, and the standard book edition followed them (Kripke, 1980; Burgess, 2010). The work's enduring innovation was to unite a reconsideration of proper names with a new account of the relation among modality, essence, and empirical knowledge.

Kripke argues that ordinary proper names are rigid designators. When we ask whether Aristotle might never have taught Alexander, we are considering that very man in alternative circumstances. We are not searching those circumstances for whoever happens to satisfy the description teacher of Alexander. This difference defeats the claim that the name is synonymous with that description, even where the description actually identifies Aristotle.

Kripke advances several connected but separable lines of criticism. His modal argument examines how names and descriptions behave across possible circumstances. Epistemic arguments show that speakers commonly know the identity expressed by an ordinary name without knowing a description that uniquely identifies its bearer. Semantic arguments point to cases where widespread beliefs about the bearer are mistaken but the name continues to designate the historically connected individual. The Gödel–Schmidt case is designed to make descriptive error especially vivid.

The work does not present a finished general theory of all metasemantic mechanisms, and its positive causal-historical sketch includes acknowledged difficulties. The central philosophical result is narrower and stronger: ordinary name use and modal evaluation cannot generally be explained by treating an identifying description as synonymous with a proper name. This conclusion left subsequent philosophy with a research program concerning what names contribute to propositions, how reference is historically grounded, and how their users can remain ignorant or mistaken about their subjects.

The Formal Structure of Rigidity

A semantic reconstruction can represent a proper name n as rigid with respect to a fixed interpretation I when, for any admissible worlds w and v at which the relevant denotation is defined, the referent of n is the same individual: den(n,w,I) = den(n,v,I) = a. The crucial qualification concerns worlds at which a fails to exist. One formulation holds that n designates a in every world, whether or not a exists there; another treats the designator as undefined in worlds lacking a. Both formulations preserve the principle that n never switches to a different object (LaPorte, 2026, §1.2).

A nonrigid definite description the F can have variable extensions. At one world it identifies a; at another it identifies b. When we evaluate what could have happened to a, rigid designation allows a to remain the subject while the property F changes. By contrast, a description interpreted with its ordinary world-sensitive intension picks out whoever meets its descriptive condition in the evaluated world.

This distinction can be expressed in modal logic without reducing possibility to knowledge. Let a and b be two names interpreted rigidly and actually denoting one object. Under classical identity, if a = b is true, it is true at every admissible world in which those names co-designate. Whether the resulting statement is simply necessary or qualified by existence assumptions depends on the semantic treatment of empty terms and the chosen notion of necessity.

Rigidity also concerns a term's semantic evaluation rather than a mysterious causal power of words. One can stipulate a rigid designator in a formal language even before explaining natural-language learning. Conversely, ordinary speakers can employ a name rigidly without possessing explicit modal theories. The formal semantics gives a condition that an account of practice must satisfy. It does not by itself explain the social and historical processes that establish the denotation in the first place.

De Jure and De Facto Rigidity

The philosophical distinction between rigidity de jure and de facto is essential to understanding the relationship between rigidity and direct reference. An ordinary name is typically treated as rigid de jure: its interpretation instructs us to keep tracking the very same object across counterfactual circumstances. The designator remains fixed because of its semantic manner of designating. A descriptive expression may be rigid de facto because its descriptive condition happens to identify the same entity in every relevant world. The successor of 2, on standard arithmetic assumptions, designates 3 wherever the arithmetic interpretation is maintained (Kripke, 1980; LaPorte, 2026, §2.2).

Therefore the inference from rigid to directly referential is invalid. A descriptive condition can determine a constant object across worlds while still contributing descriptive structure to the proposition. The rigid description and the simple name may have the same modal extension yet differ in how they signify their bearers. The modal profile alone cannot determine whether a term contributes an object or a property-mediated condition.

The reverse direction also requires care. Direct reference is commonly associated with a fixed object as semantic contribution. If an expression directly contributes a particular object to an evaluated proposition, that object is retained through ordinary counterfactual assessment. Yet contextual expressions such as I and that can change their actual referents across distinct utterance contexts. Their content can be rigid relative to a fixed context while their character yields different contents as the context changes. Changing context and varying circumstance must not be conflated.

These qualifications restore precision to a debate sometimes framed as one opposition between descriptivism and rigidity. One can accept de jure rigid names while defending descriptive information in the psychological explanation of understanding. One can accept a rigid description without endorsing direct reference. And one can accept contextual direct reference without claiming that a single token-independent referent belongs to every utterance of the same expression type.

The Modal Argument Against Descriptive Synonymy

Consider a name N whose actual bearer is a, and an identifying description the F that happens to apply uniquely to a. The descriptive synonymy thesis states that N and the F make equivalent semantic contributions. Kripke's modal argument asks whether the resulting expressions preserve the same truth conditions under counterfactual evaluation. If a could have lacked F, the name N still appears to designate a. The ordinary description the F, evaluated in that world, either designates someone else or fails to designate anyone. The terms therefore have different modal profiles.

The argument can be reconstructed through four premises. First, competent counterfactual use of a proper name tracks its established bearer. Second, descriptions generally track satisfaction of their descriptive conditions. Third, properties used in ordinary identifying descriptions are often contingent features of their bearers. Fourth, expressions that differ in truth-conditional behavior in admissible modal contexts are not universally interchangeable as semantic synonyms. It follows that ordinary names are not generally synonymous with the contingent descriptions users associate with them.

The argument is powerful precisely because it concerns the same individual under altered properties. It does not depend on the speculative claim that another possible world contains a perfect duplicate of our entire language. It asks us to hold the actual naming relation fixed while allowing the bearer to possess different characteristics. If someone else had taught Alexander, that person would not thereby have become Aristotle.

A descriptivist may respond with a rigidified description, for example the actual F, evaluated with an actuality operator anchored to the original context. That modification can reproduce a rigid modal profile. The resulting theory must then explain why the proposed descriptive content is semantically appropriate, how speakers grasp it, and whether it supports the same cognitive and epistemic behavior as the ordinary name. Rigidity by itself excludes simple nonrigid synonymy; it does not refute every theory incorporating descriptions into reference-fixing or into the explanation of cognitive significance (Soames, 2002).

The Epistemic and Semantic Arguments

The modal argument addresses how names behave in counterfactual evaluation. Kripke's epistemic arguments concern what speakers can know. If a name were synonymous with a description, one might expect competent speakers to know the associated description and to know its relevant identifying relation a priori. Ordinary speakers often identify famous individuals through a sparse, fallible, or borrowed network of information. They can use Einstein while knowing little beyond a broad association with physics. Competence with the name therefore cannot require possession of a unique and sufficiently precise description (Kripke, 1980).

The semantic or error argument adds a different pressure. Suppose that most members of a community credit Gödel with proving a theorem that was actually proved by someone else. Does their use of Gödel refer to the unknown true discoverer? The causal-historical intuition is that the name can still refer to Gödel despite the false description. The case shows how apparent explanatory success of descriptions in routine contexts may disappear in systematic cases of misinformation.

Neither argument proves that descriptive practices are dispensable. The users still require some historical route of linguistic participation, some ability to deploy the expression, and often contextual information to know what conversation is underway. What fails is a strong thesis about constitutive semantic identification. A speaker's incomplete or erroneous beliefs need not determine the reference of the name.

The strongest form of the anti-descriptivist case is cumulative. The modal argument distinguishes actual reference from descriptions under counterfactual variation. The epistemic argument weakens the assumption that successful use requires identifying knowledge. The semantic argument reveals the resilience of name reference under misinformation. Together, they favor an account in which the object is established by more than the speaker's present inventory of descriptions. They leave open the more difficult issue of what causal, conventional, and intentional relations are adequate to ground reference.

Necessary A Posteriori Identity

The identity of Hesperus and Phosphorus is the paradigm of necessary a posteriori knowledge. The two historical names were associated with appearances of a celestial body at different times; astronomical inquiry established that their actual bearer is the same planet, Venus. If both names rigidly designate Venus, an identity sentence relating them expresses an identity that could not have been false of that very object. Yet the truth of the identity was discovered empirically. The epistemic route to the truth and the truth's modal status diverge (Kripke, 1971, 1980).

The formal reasoning begins with numerical identity. An object is necessarily identical with itself under standard identity semantics. When two rigid designators a and b name the same object, replacing one with the other does not generate a possible distinction in the object itself. Consequently, if a = b is true, then □(a = b) is true within the relevant modal semantics. The contingent fact is that speakers might have encountered different objects, used different naming practices, or lacked evidence for the identity. Those epistemic possibilities do not entail that Venus itself might have been two objects.

This argument also corrects the identification of the a priori with the necessary. A proposition may be metaphysically necessary while its truth is available only through empirical inquiry. Conversely, an a priori proposition can concern a contingent matter when its truth derives from a reference-fixing convention under an actual-world stipulation. The two dimensions must be investigated independently.

A philosophical qualification concerns the existence of the named object. Some systems treat an identity involving an individual as necessary only in circumstances in which that individual exists, or interpret necessity using outer-domain semantics. The general insight survives the qualification: where two rigid expressions designate one and the same individual, they do not designate distinct individuals in alternative worlds merely because the object's contingent features vary. This insight reaches the interpretation of scientific identity claims, personal identity, and the metaphysics of natural kinds (Kripke, 1980).

Contingent A Priori Reference-Fixing

Kripke's example of the standard meter bar illustrates how a stipulated description can fix a reference without becoming synonymous with the resulting expression. A length standard established by a particular object under stated conditions can make it knowable by stipulation that the object has the designated length at the time of introduction. But the same physical object could have had another length under counterfactual conditions. The epistemic route through a stipulation need not make the measured physical condition metaphysically necessary (Kripke, 1980).

The philosophical lesson depends on the difference between the act that establishes a measure and the later modal evaluation of the object. A reference-fixing practice can select a real object through its actual descriptive properties; afterward, the object can be discussed independently of those properties. The fact that the initial use was licensed by a description does not establish a necessary identity between the referent and the described role.

This example challenges two opposite simplifications. It challenges the assumption that every a priori truth is necessary. It also challenges the assumption that any descriptive role in establishing a term must determine that term's modal semantics. A successful account of naming may explicitly contain a descriptive introduction and still treat the resulting name as rigid.

The illustration has generated debate about which propositions are genuinely known a priori when the introduction depends on empirical presuppositions. The speaker's stipulated rule can be known through linguistic competence, while the existence, uniqueness, and stability of the referent may require evidence. There is no philosophical gain in treating these different epistemic claims as interchangeable. The case is most illuminating when its conventions, presuppositions, and modal evaluation are specified separately.

Historical Chains and Causal Grounding

A causal-historical account explains name reference through an initial introduction and a chain of subsequent uses. A child is introduced to a person under a name; later speakers employ the name through testimony, teaching, documents, and social practice. Their success can depend upon the continuing historical relation of their uses to earlier uses rather than upon individually possessing uniquely identifying descriptions (Kripke, 1980).

The model explains the division of linguistic labor. Speakers may inherit a name from others who know more, just as experts can help sustain the reference of scientific and technical vocabulary. A name's referent is determined partly by participation in a community extending beyond any one speaker's current beliefs. This makes shared reference compatible with unequal knowledge.

However, an indiscriminate causal chain cannot suffice. People can hear a name and intentionally introduce a new use; they can appropriate a familiar name for another entity; they can confuse two traditions; and two separate individuals can share the same spoken name. Transmission must therefore include an appropriate historical and intentional orientation. The existence of causal influence alone does not decide which object a speaker means to discuss.

The causal tradition is best seen as a family of metasemantic approaches rather than a complete, uniform theory. It seeks to explain successful continuation of reference across changes in information. Its principal challenge is to distinguish reference-preserving chains from accidental associations and reference-changing episodes. Its most important achievement is the explanatory possibility it creates: a community can continue to speak about one object while correcting nearly everything its members believed about that object.

Evans and the Social Correction of Reference

Gareth Evans's The Causal Theory of Names, delivered in the 1973 Aristotelian Society supplementary symposium, assessed the prospects of a historical account and exposed difficult cases. The printed symposium includes a response by J. E. J. Altham, a bibliographic detail important when assigning arguments to their authors. Evans questioned whether reference could always be traced through an unbroken mechanical chain of borrowings. Social practices can change, and the dominant source of information for the use of a name may differ from its first introduction (Evans, 1973).

A classic type of challenge concerns historical transfer. A place name may originate with one landmark and eventually become associated with a different settlement through a living practice. A crude baptism-and-chain account could insist that every later use continues to concern the original object. But ordinary linguistic practice can reorient the name. This suggests that the structure of reference involves both history and the continuing dispositions, intentions, and informational practices of a linguistic community.

Evans's wider work on understanding and singular thought develops the requirement that a thinker be appropriately connected with the object of thought. Reference is a relation in which epistemic and informational conditions can matter without making a name simply synonymous with a descriptive cluster (Evans, 1982). This position offers a valuable intermediate perspective: descriptive information can play a role in maintaining object-directed cognition, while reference is not determined solely by a list of beliefs.

The practical conclusion is that the causal-historical program requires a theory of continuity and correction. One must ask whether a new use preserves the earlier object, reassigns the sign, or creates ambiguity. Actual language is historically dynamic. Rigidity remains a property of an expression under a settled interpretation, whereas the persistence of the interpretation over decades or centuries is a separate social-historical question.

Kaplan: Character, Content, and Demonstratives

David Kaplan's Demonstratives, developed in the 1970s and published in Themes from Kaplan in 1989, provided a sophisticated semantics for context-sensitive expressions. Kaplan distinguishes character—the rule connecting a linguistic expression and a context to its content—from content itself, what is expressed in the selected context. For the indexical I, character associates an utterance context with its agent. The resulting content contains that agent as the referent of I. This gives a systematic account of how one expression type can preserve a linguistic rule while referring to different people on different occasions (Kaplan, 1989, pp. 481–563; Braun, 2017).

Kaplan's account distinguishes the context of utterance from a circumstance of evaluation. The context fixes who is speaking and when; a modal evaluation varies the circumstances under which the proposition expressed is assessed. Holding context fixed, I continues to refer to the original speaker even in worlds where a different person occupies an analogous role. The utterance's content is rigid with respect to the relevant variations in circumstance, though the same word spoken by another person expresses a proposition about someone else.

Demonstratives introduce additional complexity. The referent of that or she can depend on an associated demonstration, intention, perceptual situation, or other contextual parameter. The semantic theory must say which contextual factors determine the bearer. Once an object is determined, a directly referential analysis allows the object itself to enter the proposition. The account therefore offers an explicit route from contextual rules to singular contents.

Kaplan also separates direct reference from the idea that every expression has a simple, context-free reference. A full semantics must model expression types, utterances, contexts, contents, and circumstances. The framework is a central foundation for contemporary indexical semantics because it explains variability of use without treating each resulting proposition as descriptively mediated. Its rigor also shows why the phrase directly referential requires a stated level of analysis.

Perry and the Essential Indexical

John Perry's Frege on Demonstratives (1977) and The Problem of the Essential Indexical (1979) investigate the cognitive and practical importance of perspective-dependent thought. A person can know a complete third-person description of their own situation and still fail to recognize the practical significance of the first-person assertion I am the one making the mess. The transition from impersonal knowledge to an indexical recognition can explain an action in a way that the bare third-person fact does not (Perry, 1977; 1979).

The significance for direct reference lies in the gap between an object as the referent and the subject's way of apprehending that object. A singular proposition about one person can be expressed in first-person, demonstrative, or name-based ways. Yet the cognitive role of those ways may differ dramatically. A theory that captures only the referent and a predicate can identify an important truth-conditional commonality while leaving open the explanation of changed behavior.

Perry's challenge is not exhausted by the observation that people have different beliefs. The contents needed for successful self-location and action seem to involve perspectival information that cannot always be replaced by a non-indexical description from the subject's point of view. Cognitive science and philosophy of mind can use this insight to analyze self-representation, spatial orientation, and planning. The philosophical problem arises at the interface between semantic content and the functional roles of representations.

A satisfactory direct-reference theory must therefore specify its ambition. If it accounts for truth-conditional content, its success need not automatically provide a complete psychology of thought. If it claims to explain rational action or linguistic understanding, it must address the explanatory work done by first-person and demonstrative modes. Perry's arguments make that distinction unavoidable.

Singular Propositions and Their Limits

A singular proposition is commonly characterized as a structured proposition containing an individual object as a constituent. On one direct-reference model, the proposition expressed by Alice is tired includes Alice herself together with the property of being tired. This contrasts with a general proposition whose structure contains a condition that someone may satisfy, such as whoever holds a specified office is tired (Fitch & Nelson, 2024; King, 2024).

The appeal of singular propositions is that they articulate objective aboutness. The assertion concerns a particular object, and its truth depends on that object's possessing a property. The object's identity is not supplied by a description within the proposition. Where two names rigidly designate one object, the associated basic Russellian singular propositions can be identical despite the cognitive difference in the sentences that express them.

This result creates a familiar difficulty. If Hesperus is bright and Phosphorus is bright contribute the same object and predicate, a simple model assigns them one singular proposition. A competent speaker nevertheless may believe one and doubt the other. If the singular proposition is taken to be the entire content relevant to belief ascription and rational inference, the model is too coarse-grained to preserve the observed difference.

Three families of response are available. One treats the singular proposition as semantic content while assigning associated modes of presentation a distinct cognitive or pragmatic role. Another enriches the structure of represented content with cognitive or linguistic elements. A third introduces relations among representations so that different ways of using the same object can matter without adding descriptive constituents to every singular proposition. The debate concerns explanatory adequacy, not merely whether objects can figure in propositions. A successful theory must say which phenomena its propositional individuation is designed to explain.

Frege's Puzzle and Belief Reports

Frege's puzzle is the objection that referential equivalence appears insufficient to explain informativeness and the truth of attitude reports. Suppose a person has learned one astronomical identity but not another. The person can rationally accept a sentence with Hesperus and reject the corresponding sentence with Phosphorus, even though both names designate Venus. In extensional predication, substitution of co-referring terms generally preserves truth. Within belief reports, substitution can appear to change the truth of what is attributed.

One response is Fregean: senses differentiate the contents entertained. Another is Russellian: the semantic proposition can be the same while the subject's representational state, the vehicle of belief, or the contextual standards for reporting beliefs differ. Nathan Salmon develops a detailed treatment of Frege's puzzle in his 1986 book of that title, while Scott Soames explores semantic minimalism and pragmatics in Beyond Rigidity (Salmon, 1986; Soames, 2002).

Kripke's 1979 A Puzzle about Belief complicates the dispute further. Through the case of a speaker who learns a name through different linguistic routes, Kripke shows that intuitive principles governing translation, disquotation, and belief attribution can generate tensions even without taking a simple direct-reference theory as a premise. This demonstrates that the epistemology of belief reports cannot be settled by immediately choosing a referential or descriptive theory of names.

The philosophical stakes concern the level at which rationality is evaluated. Truth-conditional equivalence is an objective semantic relation. Acceptance, doubt, inference, and rational explanation depend on representational structure, informational history, and context. A coherent account of reference must connect these dimensions while preserving their distinct functions. The identity of a referent and the identity of a cognitive route to that referent are different determinations.

Speaker Reference and Semantic Reference

In Speaker's Reference and Semantic Reference (1977), Kripke distinguishes an expression's established semantic referent from the person or thing a speaker intends to discuss on a particular occasion. A speaker might mistakenly describe a person in the room while clearly indicating that person to the audience. The success of the intended communication need not show that the conventional semantics of the description has changed. This distinction directly engages Donnellan's analysis of referential uses of descriptions (Kripke, 1977, pp. 255–276).

The distinction matters because semantics and pragmatics answer different explanatory questions. Semantic reference concerns the referent or truth-conditional contribution assigned through a linguistic rule or interpretation. Speaker reference concerns what object the utterer is attempting to identify by means of a particular speech act. Conversation can preserve an intended target despite factual mistakes in the expressions employed.

The distinction also resists a facile movement from communication to content. A listener can infer a speaker's intended subject through contextual evidence even when the utterance, interpreted literally, expresses a false proposition. The communicative achievement is real, but its explanation may involve inference, coordination, and intention rather than a revision of the expression's conventional semantic content.

Nevertheless, context-sensitive terms genuinely require contextual input to determine their semantic content. It would be equally mistaken to say that all successful referential interpretation is merely post-semantic pragmatics. The proper investigation asks which contextual parameters are licensed by the expression's character and which effects result from broader communicative inference. Kaplan's framework provides resources for the former; Kripke's pragmatic distinction illuminates the latter. The boundary is theoretical and must be drawn with examples rather than assumed from the outset.

Empty Names, Fiction, and Failure

The apparent meaningfulness of names without actual bearers is a challenge for simple direct-reference semantics. A person can discuss Sherlock Holmes, hypothesize about Vulcan, or ask whether an alleged individual exists. If the semantic contribution of a name is simply its referent, what is contributed by a name that lacks one? A theory must distinguish failed reference, fictional discourse, historical error, and deliberate pretense.

Russellian analyses often treat apparent names as descriptive constructions, preserving meaningful predication without a corresponding object. Fregean approaches can distinguish sense from reference and hence explain a mode of presentation where denotation is absent. Direct-reference theories may adopt negative free logic, gappy propositions, fictional objects, metalinguistic interpretation, or pragmatic accounts of pretense. These options have different commitments and should not be combined without stating their costs (Russell, 1905; Gratzl et al., 2025).

Free logic formalizes the possibility that a singular term does not denote any object in the relevant domain. Under some systems an atomic predicate containing such a term is false; under others it lacks a truth value; under still others an expanded semantic interpretation allows additional objects. Each treatment determines the behavior of negation, quantification, and existence claims in its own way.

Empty names therefore reveal a boundary condition on the direct-reference thesis. The theory can explain the semantics of successful singular reference while still needing a supplementary account of non-referring uses. The important question is how the supplementary account connects with the same linguistic practice. It is theoretically weak to treat failure cases as irrelevant when ordinary names can be introduced under mistaken assumptions and can lose their supposed bearers through historical correction. A mature account must state which constructions genuinely require a semantic referent and which remain intelligible through other mechanisms.

Natural Kinds and Theoretical Identities

The expansion from proper names to natural-kind terms connects semantics to scientific classification. Hilary Putnam's Twin Earth argument invites us to imagine a liquid resembling terrestrial water in ordinary appearances but differing in chemical composition. The case challenges the claim that shared superficial descriptions or internal psychological states alone determine the extension of water. Putnam emphasizes an environment-dependent relation linking a linguistic community to the substance it names (Putnam, 1975).

Kripke similarly argues that the reference of a kind term can be secured through an actual sample while scientific investigation discovers the underlying constitution of the kind. The proposition that water is H2O has therefore served as an example of an empirical identification with purported necessary consequences. Its modal force depends upon the term's reference being held to the same chemical kind across counterfactual evaluation (Kripke, 1980; Bird & Tobin, 2025).

The extension is philosophically contested. Natural kinds are not ordinary individuals. Their identity conditions can depend upon scientific theory, explanatory interests, population structures, historical descent, and whether a kind is treated as a substance, a property, or a category. A species term may behave differently from a chemical substance term. Rigidity for kind predicates also requires semantic interpretation that is more complex than assigning the same object to a proper name.

The lasting contribution is methodological. Scientific discovery may revise a community's understanding of what it has been referring to without simply creating a new referent whenever a theory changes. But a realistic account of science must distinguish reference preservation from conceptual revision. It must explain which discoveries disclose the constitution of an established kind and which discoveries warrant replacement of the original classification. Philosophical arguments about natural kinds thus extend rather than mechanically repeat the logic of proper names.

The Metaphysics of Objects and Modal Evaluation

Rigidity creates an interface between semantics and metaphysics. To say that a name denotes the same object in different possible worlds presupposes an intelligible practice of evaluating a single individual under varied circumstances. Kripke argues that familiar counterfactual questions already involve such an individual: could this person have lived elsewhere, or could this table have been made from different material? The metaphysical inquiry concerns the possible properties of a given entity (Kripke, 1980).

This perspective challenges an approach that requires a complete qualitative criterion before an individual can be considered across possible worlds. A question about Aristotle can concern Aristotle himself even when descriptive features that actually identify him are allowed to vary. A modal semantics can encode cross-world identity without specifying that every qualitative feature is essential to the individual.

But semantics does not automatically decide which features really are essential. Once reference is fixed, metaphysical arguments must examine whether an individual could have had a different origin, material, or kind. The conclusion that names are rigid is not by itself a proof of origin essentialism, mind–body identity, or any particular theory of individual essences. Additional premises about identity conditions and the nature of objects are needed.

The distinction is especially important with temporal change and constitution. A rigid name may track a continuing individual, but philosophical disputes about what constitutes that individual's persistence remain. The semantics explains how a proposition holds a subject fixed while assessing alternative circumstances; metaphysics determines which alternatives count as genuine possibilities for that subject. The two fields cooperate without collapsing into one another.

Two-Dimensional and Relationist Responses

Two-dimensional semantic theories distinguish two dimensions of evaluation, often associated with considering a scenario as actual and evaluating what would hold under a counterfactual circumstance. They attempt to explain how a speaker's understanding can determine an epistemic profile while reference, once established, remains fixed across modal evaluation. Different versions of the framework assign different theoretical roles to the two dimensions; the machinery alone does not commit one to a particular philosophical account of meaning (Schroeter, 2021).

The framework responds to a familiar tension. A competent speaker can entertain an epistemic scenario in which a familiar word turns out to refer differently from what it actually refers to, while still accepting that the word's established reference is rigid in metaphysical counterfactual evaluation. By distinguishing epistemic possibilities from metaphysical alternatives, the theory aims to preserve both the significance of discovery and the necessity of certain established identities.

Relationist approaches take a different route. Kit Fine's Semantic Relationism (2007) argues that representational relations among expressions can matter even when their individual semantic contributions do not contain distinct Fregean senses. The possibility that distinct occurrences stand in different coordination relations gives a means to explain cognitive differences without simply assigning descriptive properties to each individual name.

The critical issue for both approaches is explanatory burden. A second semantic dimension needs a principled interpretation of its epistemic evaluation. A relationist account must establish which relations among representations determine rational judgments, and why. Both develop important alternatives to the claim that objective reference and cognition must be captured by one undifferentiated semantic value. Their value is greatest when they specify the extra structure required for particular phenomena rather than multiplying semantic parameters without constraints.

Singular Thought and Cognitive Access

Singular thought concerns thought about a particular object as such. The central debate asks what conditions must obtain for a thinker to entertain an object-dependent thought and how those conditions relate to perception, memory, testimony, and linguistic communication. Russell's acquaintance restriction is one historical answer; later authors investigate less restrictive forms of object-directed cognition. Essays collected in New Essays on Singular Thought examine perceptual and nonperceptual routes to singular thought (Jeshion, 2010).

A thinker may possess substantial information about an entity without being able to coordinate that information with an established referent. Conversely, a thinker can inherit an effective name through testimony while knowing very little about the object. This asymmetry suggests that an adequate theory should distinguish the existence of a semantic anchor from the richness and reliability of the associated cognitive representation.

Imogen Dickie's Fixing Reference (2015) develops an account of the fixing of aboutness for thoughts and singular terms. This approach emphasizes the metasemantic conditions through which a mental representation becomes directed toward an object. It thereby targets a problem that pure modal rigidity does not solve: why is this token thought about this actual individual?

A promising synthesis treats direct reference as one possible answer to the question of propositional contribution while assigning independent explanatory roles to grounding, cognition, and communicative coordination. An object may be a constituent of what is asserted even while a speaker's mental file for it contains misconceptions. A description may help a subject find or recognize the object without becoming part of the proposition. The challenge is to explain how these levels interact so that the resulting theory remains predictive and empirically informed.

Comparative Evaluation and Contemporary Research

The historical progression yields several durable insights. Mill identifies a difference between naming an individual and describing attributes. Frege establishes the importance of cognitive value and the problem of informative identity. Russell demonstrates that apparent singular reference can conceal quantificational logical form. Donnellan discloses the flexibility of referential use in ordinary conversation. Kripke distinguishes the semantics of rigid names from the contingent descriptive information that speakers hold. Kaplan separates linguistic character, contextually fixed content, and circumstances of evaluation. Perry reveals the practical significance of indexical cognition.

No single line of argument establishes every component of a comprehensive direct-reference theory. Kripke's modal argument is strongest against the semantic identification of names with contingent descriptions. It does not entail that the referent alone explains all aspects of linguistic understanding. Kaplan gives a rigorous analysis of certain context-sensitive terms, but details of demonstrative determination remain disputed. Frege's puzzle persists whenever the individuation of propositions is coarser than the individuation of rational cognitive states.

Contemporary research focuses increasingly on metasemantic anchoring, singular thought, cognitive architecture, coordination, and the plurality of linguistic practices. Soames distinguishes semantic content from information pragmatically communicated by an utterance; Recanati develops a systematic defense of direct reference in relation to thought; Fine examines coordination relations; Dickie concentrates on the conditions fixing reference and aboutness. These projects disagree in significant ways but each recognizes that the object–word relation alone leaves explanatory work unfinished (Soames, 2002; Recanati, 1993; Fine, 2007; Dickie, 2015).

The comparative lesson is precise. Rigidity is a modal condition on designation; direct reference is a theory of content; descriptivism names multiple theses about meaning, knowledge, or reference determination; causal and social approaches explain grounding; cognitive accounts explain the perspective from which objects are represented. These questions admit overlapping but distinct answers. Their distinction supplies the conditions under which substantive disagreements can be resolved.

The Explanatory Architecture of Reference

The research supports a four-level architecture. At the level of grounding, a use of an expression is connected to an object through conventions, introduction, causal history, intentional selection, or appropriately structured social practice. At the level of semantic composition, the expression contributes a value to what is asserted; a direct-reference theory specifies that value as the object itself in appropriate constructions. At the level of modal evaluation, an established referent is held constant across admissible alternatives when the expression is rigid. At the level of cognition, thinkers deploy representations, recognition criteria, memory, testimony, and self-locating information that enable and constrain use.

These levels are jointly explanatory without being conceptually identical. A name can be grounded historically and contribute a singular object to a proposition. A descriptive term can acquire a constant modal extension while retaining descriptive structure. A demonstrative can have variable content across contexts and rigid content relative to one context. A belief about an object can possess a cognitive role that differs from another belief whose object-involving truth conditions are equivalent.

The architecture also clarifies the structure of philosophical disagreement. The question whether a name is rigid is settled by analyzing its behavior in counterfactual evaluation. The question whether it is directly referential requires a theory of semantic content. The question why it designates one bearer rather than another is metasemantic. The question why a speaker can rationally fail to recognize co-reference is cognitive and epistemological. An argument crossing these levels needs an explicit bridging premise. Without such a premise, it merely changes subjects.

This yields a disciplined conception of semantic objectivity. An individual can remain the public target of inquiry despite variation in the beliefs of speakers who use a name. Objectivity is sustained by reference-preserving relations and can be investigated independently of whether a given speaker has a correct identifying description. Yet cognitive and communal relations are indispensable to understanding how language achieves that objectivity. A complete account therefore explains the determination, contribution, evaluation, and uptake of reference in one coordinated but differentiated system.

Open Philosophical Questions

The first open question concerns individuation of semantic content. If singular propositions contain only objects and properties, what additional elements are necessary to explain informative identities and divergent attitudes? Theories of guises, coordination relations, and pragmatic enrichment offer competing answers. The challenge is to specify which elements are genuinely semantic, which are psychological, and which depend on conversational context.

The second concerns the historical stability of names. Causal chains account for continuing reference under imperfect knowledge, but they must also allow communal correction and reference change. There is no universal formula identifying the exact point at which a historically inherited term acquires a new bearer. Such cases require investigation of intention, convention, reliable information, and patterns of use.

The third concerns empty and fictional names. Theories must explain how apparently singular assertions remain meaningful without actual bearers and whether all such discourse has one logical form. Different kinds of nonreference may demand different mechanisms. This issue also determines how strongly direct-reference semantics must be restricted to successfully referring terms.

The fourth concerns extension to categories and theoretical terms. Rigidity for natural-kind vocabulary must be reconciled with scientific revision, classifications with fuzzy borders, and differing views about the ontology of kinds. The generalization from names of individuals to names of kinds is a research question, not a routine substitution.

The fifth concerns the relationship between external grounding and cognitively accessible content. A theory of communication needs to explain how two agents coordinate upon an object, how they can discover that two representational routes converge, and how this discovery transforms reasoning. Philosophical semantics, cognitive science, and social epistemology converge on this issue. The conceptual framework advanced here identifies where their evidence and arguments must meet.

Conclusion

Direct reference and rigid designation describe two distinct aspects of the ability of language to concern particular entities. Rigidity secures the stability of designation in modal evaluation. Direct reference identifies the object itself as a semantic constituent in theories of singular content. The first does not entail the second: some descriptive expressions designate rigidly. Nor does either by itself explain how names are introduced, transmitted, interpreted, and connected to the cognitive lives of their users.

The historical dispute advances through successive explanatory discoveries. Frege shows why co-reference does not exhaust cognitive significance. Russell distinguishes logical form from surface appearance. Donnellan and Kripke clarify the difference between conversational intention, conventional meaning, and cross-world designation. Kaplan and Perry demonstrate the importance of the separate dimensions of context, content, and perspective. Contemporary work on two-dimensional semantics, semantic relationism, and singular thought develops responses to difficulties left by simple referential theories.

The resulting philosophical synthesis identifies the central work of a comprehensive account of names: to specify the grounding of their referents; to explain their compositional contribution; to state their modal profiles; and to analyze the cognitive and communal conditions under which subjects successfully use them. These are interdependent determinations within one theory of reference. Their differentiation makes the enduring achievements of direct-reference semantics available without diminishing the explanatory force of Fregean cognitive problems, Russellian logical analysis, or the historical complexity of actual language.

A stable referent enables language to sustain inquiry into an object whose nature remains incompletely understood. The philosophical importance of rigid designation lies in this possibility of speaking and reasoning about the same thing while considering what it could have been. The importance of direct reference lies in explaining how an object can enter the content of an assertion. Together, they provide an essential foundation for the philosophy of language, modality, identity, and objective inquiry.