Paper in the process of submission. It summarizes the main content of the book How Do Proper Names Really Work (De Gruyter 2023)
SEMANTIC
COGNITIVISM:
OVERTHROWING THE STALEMATE BETWEEN DESCRIPTIVISM AND REFERENTIALISM
Abstract:
This article summarizes the principal stages of a
sustained inquiry into theories of reference and advances several corrective
proposals. It develops a cognitivist account of reference grounded in implicit
criterial rules, intended as a replacement for cluster descriptivism, particularly
in the form defended by John Searle. The central difficulty with cluster
descriptivism lies in its indeterminacy, which renders it vulnerable to the
familiar objections raised by referentialist externalists such as Saul Kripke,
Keith Donnellan, Hilary Putnam, and David Kaplan. This indeterminacy stems from
the absence of a satisfactory structure for the cluster: we lack both a
principled account of which descriptions are relevant and a method for
assessing their relative weight. The
article develops a higher-order rule schema – sufficiently flexible to admit
variants – capable of organizing clusters of descriptions and establishing
reliable measures of application. Within this framework, each proper name can
be associated with an identification rule robust enough to warrant rigid
designation and to withstand the objections advanced by externalist theories.
Moreover, the framework is readily generalizable to other singular and general
terms beyond proper names. Its significance lies in its capacity to move beyond
the entrenched opposition between descriptivism and referentialism, offering a
synthesis that preserves the strengths of each while avoiding their
characteristic weaknesses.
Keywords: reference, proper names, general terms, meaning, externalism
Since 2007, I
have been developing a form of neodescriptivism in theories of reference I am
now inclined to call semantic cognitivism. This work has
resulted in several articles and at least one book. My current article
revisits the central ideas of the theory, incorporating corrections and
refinements. The motivation is clear: I believe that semantic cognitivism
offers a framework that surpasses both referentialism (as advanced by
philosophers such as Kripke, Donnellan, Putnam, and Kaplan) and descriptivism
(as sustained by Searle, Strawson, Wittgenstein, Russell, and Frege). In doing so, it creates
an explanatorily more powerful theory of reference.
I begin with a metaphilosophical remark that
highlights a fundamental difference between my approach to the problem of
reference and that of philosophers working within a deductivist formalist
orientation. The difference may be captured by the contrast between digital and
analog computation. Digital computation operates with discrete symbol combinations,
whereas analog computation processes data continuously. The human brain is more
like an analogical system because human cognition relies heavily on continuous
processing and pattern recognition. Classical logic, by contrast, is digital,
while inductive logic is analog, insofar as the strength of inductive arguments
cannot be determined in absolute terms.
This observation suggests that a theory of
reference can adequately account for the referential employment of our terms
only if it incorporates the analogical dimension. The theory advanced here does
precisely this. It shows that a proper name becomes a rigid designator when it
is associated with an identification rule that integrates the analogical aspects
of reference. Such a rule renders the name sufficiently flexible to be
translated into descriptive terms without forfeiting its status as a rigid
designator.
1
The referentialist externalism was of undeniable
importance: imaginative in conception, rigorous in execution, and
intellectually provocative. Nevertheless, I take it to be ultimately untenable.
By contrast, the descriptivist internalism appears to me fundamentally sound
and intuitively compelling, though insufficiently elaborated in its theoretical
articulation. At their best, both traditions offer genuine contributions to the
advancement of a cognitivist theory of reference – a framework I shall endeavor
to articulate in what follows.
The account I
propose may be described as a neo-descriptivist and tacitly cognitivist theory
of reference, one that is internalist in its explanation of how proper names
and related expressions secure their referents. Its central claim is that the
traditional opposition between descriptivism and referentialism can be
dissolved, not by siding with one camp against the other, but by reconfiguring
the problem of reference itself. The stalemate between descriptivist and
referentialist accounts has persisted for more than four decades, and I can
discern no alternative strategy capable of overcoming it, except the view to be
presented here. On this view, reference is best understood as a process
susceptible to refinement through successive approximations, in a manner
continuous with scientific inquiry. Properly articulated, such a theory is not
merely of philosophical interest: it is sufficiently strong that, if
implemented within a computational framework and supplied with the requisite
data, it could in principle identify proper names’ owners.
I
want to begin by considering Michael Dummett’s interpretation of Fregean sense in
terms of criterial rules. As he wrote:
For Frege, the sense of an expression always consists in
a rule which, taken together with the constitutive rules of the senses of other
words, determines the condition for the truth of the sentence in which the
expression occurs. (1981: 194, my italics)
And further,
to know the sense of a proper name is to possess a
criterion for recognizing, for any given object, whether or not it is the
bearer (referent) of that name; to know the sense of a predicate is to have a
criterion for deciding, for any given object, whether or not the predicate
applies to that object; and to know the sense of a relational expression is to
have a criterion for deciding, given any two objects in a particular order,
whether or not the relation they represent holds between the first object and
the second (1981: 229).
These ideas strongly echo Wittgenstein’s earlier
suggestion that criteria – or better, criterial rules – are: “that which
gives our words their common meanings” (1958: 57; cf. G. P. Baker 1986). Or
again: “There is a correspondence between the words ‘meaning’ and ‘rule’” (1969,
sec. 62).
The same intuition
has found a more explicit formulation in Ernst Tugendhat’s speculative proposal
(1983: ch. 13.4; 1976: 258–263), according to which the singular terms must be
governed by rules of identification (Identifikationsregeln), while
general terms must be governed by rules of application (Applikationsregeln),
and, that the application rule based on the previous application of the
identification rule in a singular predicative statement form its verification
rule (Verifikationsregel).[1]
My point of departure is the analysis of proper names,
long regarded as the central focus of theories of reference, with inevitable implications
extending to other terms. The aim has been to identify general forms or schemata
of rules for the identification of proper names. On the basis of the
working hypothesis outlined above, the initial task was to formulate criterial
rules capable of determining the referents of proper names – rules that, within
a neo-descriptivist framework, vindicate Kripke’s compelling claim that proper
names function as rigid designators. With such rules in place, I argue that a
more refined version of neo-descriptivism can successfully address the
counterexamples advanced by Saul Kripke and Keith Donnellan (... 2023, chap.
IV), as well as the Fregean identity paradox (... 2023, chap. V; Frege 1892).
How, then, are
we to arrive at a rule of identification for a proper name? To begin, it is
worth noting that the definite descriptions invoked by defenders of the cluster
theory are best understood as expressions of rules guiding the
identification of the bearer of a name. We can term them descriptions-rules.
The initial task, therefore, is to isolate the most fundamental descriptions-rules
for identifying the bearer of a proper name. My strategy has been to focus on
what I call privileged users of names – those individuals who are best
positioned to know the sources that secure the veracity of fundamental
descriptions (experts, witnesses, family members, and so forth). This focus is
chosen because it offers the best prospect of regulating the disordered clusters
of descriptions of proper names inherited from historical descriptivism (cf.
Searle 1958).
The method
employed to uncover the most fundamental descriptions-rules known to privileged
users is inspired by J. L. Austin’s pragmatic methodology (1961), characteristic
of “ordinary language philosophy”. Austin’s guiding principle was to begin with
what dictionaries record, thereby producing pre-philosophical mappings of
relevant meanings to serve as points of departure. Ordinary language, as Austin
acknowledged, may not constitute the end of philosophical inquiry, but it may
well provide its beginning. The difficulty, of course, is that proper names are
rarely listed in dictionaries. Yet this difficulty can be circumvented, since
many names are at least encyclopaedised. There we find what we mean by using a certain
proper name.
Consider, for example, the name ‘Aristotle’. A reliable
encyclopaedia, I chose a particularly concise one, the Penguin Dictionary of
Philosophy, provides the following compact account:
Aristotle was born in Stagira in 384 BCE, the son of
Amyntas III’s physician. At seventeen, he went to Athens, where he studied
under Plato for twenty years. Following Plato’s death, he moved to Assos, spent
two years conducting biological research in Lesbos, returned to the Macedonian
court for a time, and later, during Alexander’s domination of Greece, lived for
12 years in Athens. After Alexander’s death, he fled to Chalcis, where he died
in 322 BCE at the age of sixty-two. The entry is followed by a list of his
principal works – the Organon, Physics, Metaphysics, Nicomachean
Ethics, Poetics, among others – together with concise explanations
of their contents.
Two types of descriptive rules can be readily abstracted
from entries on proper names in sources such as the Penguin Dictionary
and, in fact, from any other reliable encyclopaedia. These rules yield two conditions that may, as later will
be seen, appropriately be called foundational:
1. Localizing condition: This condition specifies
the place and time of the bearer’s emergence, together with their origins,
their spatio-temporal trajectory, and, where applicable, the place and time of
their disappearance.
2. Characterizing condition: This condition
specifies the principal reason – or set of reasons – for which the proper name
is employed.
Let us consider the name ‘Aristotle’ more closely. Its
locating condition may be summarized in terms of origin (the son of the court
physician of Amyntas III), place and date of birth (Stagira, 384 BCE),
spatio-temporal trajectory (he crossed the Aegean at the age of seventeen to
study with Plato, remained at the Academy for twenty years, and after Plato’s
death moved to Assos, among other episodes...), and eventual disappearance (his
death in Chalcis in 322 BCE). A parallel account can be given for the characterizing
condition. Aristotle is remembered above all as the author of the Aristotelian
corpus, the most comprehensive philosophical system of antiquity, whose
influence endures to the present day.
All of this and much more, as is well known, is familiar to
the privileged users of the name, in this case, interpreters and historians, in
its minutest details. As I noted, it is reasonable to begin by concentrating on
the study of the proper name as it is employed by such users, postponing
consideration of those who possess only limited knowledge, since – as will
become clear – the mechanism of reference in the last case is of a very different
kind.
This raises the question of how the discovery of these
two fundamental conditions is philosophically helpful. Kripke was correct to
observe that, in a possible world, Aristotle might have been born many years
later or might have left no writings at all (1980, 62). That much is true. Yet
in another respect, the foundational rule-descriptions are indispensable: it is
impossible to conceive of a situation in which none of the demands they
impose are satisfied. Our historical Aristotle could not have been a newly
enrolled student at the University of Athens, nor the twentieth-century Greek
shipowner Aristotle Onassis, nor, as Searle once noted (1967, 490), an
illiterate homonymous fishmonger living in Venice during the late Renaissance.
The greater importance of the foundational
conditions becomes evident when they are contrasted with what may be termed auxiliary
rule-descriptions, which are generally accidental. Consider, for example, metaphorical
descriptions such as Dante’s epithet ‘the master of those who know,’ or familiar
but contingent descriptions such as ‘the tutor of Alexander’ or ‘the
founder of the Lyceum.’ One might also quote obscure contingent descriptions
such as ‘the grandson of Achaeon’ or ‘the husband of Pythias,’ or adventitious
descriptions such as ‘the philosopher mentioned by the professor in the
last class,’ which are indexical and typically short-lived. All
of these definite descriptions may be absent without undermining Aristotle's
existence. By contrast, it is
inconceivable that any historical Aristotle could satisfy all of these
auxiliary descriptions while failing to satisfy any foundational condition.
Imagine, for instance, a director of a lyceum in Tessalonica named Aristotle,
whose wife was called Pythias, who had a student named Alexander, and who was
esteemed by his colleagues as a true “master of those who know.” However
striking the coincidences, such a figure could never be identified with our
historical Aristotle.
There
is, then, a significant distinction that warrants closer examination. Auxiliary
descriptions, however promising, do not ultimately prevail. Indeed, they have
misled descriptivist philosophers in the past. Frege, for example, substituted
the name Aristotle with ‘the pupil of Plato and the tutor of Alexander the
Great’ (1892, 28n.), while Wittgenstein replaced Moses with ‘the man who, as a
child, was taken from the Nile by the Pharaoh’s daughter’ (2009, I, §79). Both
are well-known accidental auxiliary descriptions. Although such descriptions
undeniably guide us toward reference and form part of the descriptive cluster, they
remain dispensable.
A particularly instructive case of what may
be termed an accidental auxiliary definite description is the form ‘the
bearer of the name N.’ Consider, for instance, the description ‘the bearer of
the name Aristotle,’ which has played a central role in the development of
metalinguistic theories of proper names. Yet there is nothing fundamental about
this description itself. To see this, imagine a possible world closely
resembling our own. In 238 BC, the court physician of Amyntas III fathered
twins. One is baptized ‘Aristotle,’ the other ‘Pitacus.’ Suppose further that
Pitacus lives out precisely the spatio-temporal career of our Aristotle: he
studies with Plato, composes the Pitacusan corpus (identical in every respect
to the Aristotelian corpus familiar to us), and dies in Chalcis in 322 BC. Meanwhile,
the twin baptized ‘Aristotle’ becomes a physician like his father, joins
Alexander’s forces, and perishes of hunger and thirst upon returning from
India. In such a scenario, we would unhesitatingly identify our Aristotle with
Pitacus, and not with the unfortunate physician who bore the name ‘Aristotle.’
The lesson is clear: the description ‘the bearer of the name N’ does not fix
reference in any philosophically fundamental way. It is, rather, an accidental
device, one whose apparent significance derives from contingent linguistic
practices rather than from any deep metaphysical or semantic necessity.
We may consider
a wide range of options in attempting to organize the foundational conditions
into one or more rule-schemes that could serve as determinants of the rigidity
of proper names. In general, the inclusive disjunction of such
rule-descriptions suffices: the complete satisfaction of both disjuncts,
or even of a single disjunct, is not required. Yet insufficient
satisfaction can be equally inadequate, though one or the other disjunct can
remain alternatively fully unsatisfied insofar as this is compensated by the
satisfaction of the other. Suppose, for example, that in a possible world the
only son of the court physician of Amyntas III, born in 384, was an
anencephalic fetus who survived less than a week and was never baptized. In that world, with no Aristotelian works in existence,
we would hesitate to recognize him as our Aristotle. Nor would we identify as our Aristotle someone who lived
in a different time and place, but had written only the Magna Moralia, a
work of some historical interest but now widely regarded as spurious.
Moreover, we
admit only that a single bearer sufficiently satisfies the foundational
conditions sufficiently, and this is secured by adding the further requirement
that the bearer who best satisfies them must be chosen. Thus, if in a possible
world all that were known is that two individuals lived in Greece during
Aristotle’s time – one the author of the Metaphysics, the other of the Magna
Moralia – we would prefer to recognize as the better competitor the author of
the Metaphysics, and thereby our Aristotle.
The result, in
summary, is the following rule-scheme: the referent of a proper name is fixed
by the bearer who best satisfies the relevant foundational conditions, provided
that such satisfaction is sufficient, though not necessarily complete. In other words:
RC: A proper name N (or an equivalent name) has a bearer
see
(i-a) satisfies its location condition and/or
(i-b) satisfies its characterization condition
(ii) in a sufficient manner as a whole and
(iii) better than any other competitor.
The name 'Aristotle' satisfies this rule-scheme through
the identification rule summarised below:
RI: 'Aristotle'
(or an equivalent name) has a bearer
See
(i-a) satisfies the localizing condition summarised
above, and/or
(i-b) satisfies the characterising condition of having
written the main works of the Aristotelian opus.
(ii) he satisfies these conditions sufficiently as a
whole.
(iii) he satisfies these conditions better than any other
competitor.
The RI scheme may be taken as canonical, though it is by
no means exhaustive. Its application is most immediate in the case of proper
names such as Paris, the Mona Lisa, the Eiffel Tower, the Taj Mahal, Everest,
the Amazon River, Death Valley, where the referent is fixed by a standard
identifying condition. Yet the scheme admits of significant variation. In
certain instances, characterization assumes primacy; Aristotle himself provides
the paradigmatic case, where the identifying condition is constituted by a set
of descriptive features. In other cases, spatial or locational criteria
dominate, sometimes to the point of excluding all else. The planet Venus
illustrates this latter mode: its reference is determined by its orbital
position between Earth and Mercury, a condition that has held since its
earliest recognition. That Venus is a planet rather than an asteroid is not
irrelevant, but even if it lost its extremely dense, hot, and reflective
atmosphere, responsible for its brightness, it would remain Venus.
There are, moreover, proper names whose reference is
secured by a single foundational condition—what may aptly be termed
“one-footed” names. The name universe, for example, is fixed by the condition
“everything that exists,” and in this respect it lacks any locational anchor.
Conversely, the name O, designating the circumcenter of a triangle, is
determined wholly by spatial position, without the need for further descriptive
characterization.
The RI scheme,
then, provides a general framework for the rigidity of proper names, but its
application varies depending on the balance between characterization and
location. In some cases, both conditions cooperate; in others, one
predominates. And in certain limiting cases, the satisfaction of a single
condition suffices.
What, then, is to be said of proper names
that have not undergone encyclopedic consolidation? Such names typically rest
upon a characterization far more diffuse and indeterminate than that associated
with figures like Napoleon. Nevertheless, their referents remain identifiable
through a network of contingent supports: personal history, documentary records
of identity, and the recollective testimony of family members, friends, and colleagues.
At this
juncture, one might object that the proposed scheme of regimenting rules for
bundles of descriptions is insufficiently determinate. How many foundational
descriptive details must be supplied? What, precisely, constitutes “sufficiency
as a whole”? Or again, what is meant by “the absence of a competitor of equal
standing”? My response is both Aristotelian and Wittgensteinian. Aristotle
reminds us in the Nicomachean Ethics that “it is the mark of an educated man to
look for precision in each class of things just so far as the nature of the
subject admits” (1984: I.3, 24–25). Wittgenstein, in the Philosophical
Investigations, likewise emphasizes that natural language is not a rigid
formal calculus but a living instrument, whose vagueness is not a defect but a
constitutive nd protective feature. To demand greater precision than the
subject matter allows is, in both cases, to misconstrue the nature of
linguistic practice.
The vagueness of natural language is, in the
end, irreducible. In this respect, our classificatory practices may be likened
to those of meteorologists, who sort clouds by shape in a sufficiently vague
way, while fully aware of their innumerable particular forms. What matters, for
present purposes, is that proper names succeed in their function under the
ordinary circumstances of their employment. Criterial rules inevitably falter
at points of indeterminacy—where boundaries blur or criteria conflict. Our cognitive
apparatus operates more like an analog than a digital computer, tolerating
gradations rather than demanding strict demarcations. By contrast, the very
notion of rigid designation, as introduced by Kripke, presupposes an additional
condition: that the bearer of the name be definitively identifiable
across all possible worlds. In our case, the corresponding condition is that
the identification rule must be definitely applicable.
The proposed scheme of rules already proves
serviceable in addressing a wide range of familiar puzzles. Consider, for
instance, the ancient paradox of the Ship of Theseus, which we may designate as
‘Calibdus’. Theseus sailed upon this vessel for many years, gradually replacing
its worn parts. Eventually, every
component was replaced, with the originals stored in a warehouse. Suppose now that these discarded parts are reassembled
into a ship structurally identical to the one Theseus continued to sail. Which, then, is the genuine Ship of Theseus? No definite
answer seems available.
The
paradox admits resolution once we distinguish between two competing conditions
of identity. The first, the locating condition, is satisfied by the vessel
originally constructed for Theseus and persisting through its career in space
and time. The second, the characterizing condition, is satisfied by the vessel
that retains some portion of the original material. The difficulty, then, lies
in the conflict between these criteria. Yet most persons would be inclined to
identify the genuine ship of Theseus with the vessel he in fact sailed, on the
grounds that the locating condition is more adequately met. Although the
material has not remained constant, the ship's structural continuity over such
a long time provides a more compelling basis for identity than the mere
retention of parts.
The difficulty
is not dispelled, for the example admits of modification. Suppose that, shortly
after its completion and christening, the Calibdus remains at the shipyard. Its
parts are then rapidly replaced by identical ones, so that within a week
another vessel, materially continuous with the first, emerges alongside it. In
this case, since the spatio-temporal career of the original ship was virtually
nonexistent, we are inclined to say that Theseus’ ship has merely changed
places, and it is now the second vessel rather than the first. The rationale is
that the characterizing condition is fully satisfied by the second ship,
whereas the locating condition had scarcely begun to be satisfied by the first
one.
What I have sketched thus far is a theory of
the referential meaning of proper names. The theory is articulated through a
set of description-rules, at the center of which lies the rule of
identification, surrounded by a constellation of auxiliary description-rules
that, in general, serve a directive function. Contrary to Kripke’s suggestion
that proper names are devoid of meaning, I contend that they are, in fact,
saturated with meaning when considered from the standpoint of their privileged
users. The difficulty, however, is that the cognitively active meanings of
proper names vary widely and remain opaque to many speakers. Because we
typically employ them with only one or the other specific meaning in view—and because
most of us are not privileged users—we are led to suppose that proper names
lack meaning altogether. This impression is reinforced when we contrast them
with general terms, whose meanings are stable, repeatable, and uniform across
contexts.
What I have
presented so far are the outlines of a theory of the referential meaning of
proper names. It consists of a set of rules-descriptions whose core is formed
by the rule of identification and a halo composed of auxiliary
rules-descriptions that are generally capable of performing a directing function.
Proper names are not meaningless terms, as Kripke would like to believe. On the
contrary, they are overly rich in meaning for their privileged users. It is
just that their cognitively active meanings vary greatly and are unknown to
many speakers. As we often use them with only one or the other specific meaning
in mind and are almost never privileged users, we have the impression of a lack
of meaning when we compare them with general terms, whose meanings are repeated
and always the same.
2
I now wish to consider how those who are not privileged
users—whom I shall call indigent users of proper names, and who in fact
constitute the vast majority—are nonetheless able to employ such names
referentially. Take, for instance, the case of someone whose sole knowledge of
Aristotle is that he served as Alexander’s tutor, gleaned from a film in which
an elderly philosopher instructs the young prince. Or consider the passerby
who, when asked who Aristotle was, replies only that he was some great ancient philosopher,
without further detail. Kripke
maintains that in both cases the speaker succeeds in referring to Aristotle. My
own view is more cautious: I prefer to say that the indigent user succeeds in
inserting the name appropriately into discourse, and that reference does indeed
occur, but only by virtue of mechanisms akin to what Strawson termed borrowing
of reference (1963, p. 185) and to what Searle later described as
parasitic reference (1983, chap. 9).
The uninformed user, let us suppose, possesses only an
auxiliary rule-description or some generic fragment of the identification rule.
He has, at best, a tacit grasp of the general structure of identification-rule
schemes, together with the awareness that there exist privileged users who
genuinely command them. Consequently, he recognizes that his own knowledge is
insufficient to sustain a determinate reference grounded in full acquaintance
with the relevant facts. Conscious of this limitation, he nevertheless succeeds
in inserting the name into discursive contexts that are deliberately vague, yet
sufficiently determinate to allow interlocutors to identify the referent
without lapsing into outright error. These interlocutors may know as much,
less, or more than he does about Aristotle, but the communicative act succeeds
insofar as the intended referent is grasped. Should further precision be
desired, one need only pursue the chain of references until reaching those
privileged experts—or, in contemporary circumstances, consult an AI repository
that effectively embodies their expertise.
The uninformed
user, let us suppose, possesses only an auxiliary rule-description or some
generic fragment of the identification rule. He has, at best, a tacit grasp of
the general structure of identification-rule schemes for proper names, together
with the awareness that there exist privileged users who genuinely command
them. Consequently, he recognizes that his own knowledge is insufficient to
sustain a determinate reference grounded in full acquaintance with the relevant
facts. Conscious of this limitation, he nevertheless succeeds in inserting the
name into discursive contexts that are deliberately vague, yet sufficiently
determinate to allow interlocutors to identify the referent without lapsing
into outright error. These interlocutors may know as much, less, or more than
he does about Aristotle, but the communicative act succeeds insofar as the
intended referent is grasped. Should further precision be desired, one need
only pursue the chain of references until reaching those privileged experts—or,
in contemporary circumstances, consult an AI repository that effectively
embodies their expertise.
For a parasitic reference to succeed, it
must exhibit convergence. Convergence obtains when two conditions are satisfied:
first, the bearer of the name must be classified in an acceptable manner;
second, the speaker must possess a tacit knowledge of the language,
specifically the awareness that the identification rule is not fully known—an
awareness whose absence would risk embarrassment. Such parasitic use, or
borrowing of a proper name, constitutes an indirect mode of reference,
sustained by the epistemic resources of the linguistic community as a whole. It
is possible only because the indigent user recognizes that privileged users
exist who command the identification rule with sufficient thoroughness to
secure the individuation of Aristotle.
Suppose, by way
of illustration, that in the aftermath of a third world war a survivor
belonging to a community wholly ignorant of European history were to discover a
scrap of paper bearing the name ‘Aristotle’, accompanied by the description
‘the author of the Aristotelian corpus.’ It would be mistaken to claim that,
armed with this description, the survivor could thereby succeed in referring to
Aristotle. For reference requires insertion into a linguistic practice in which
the name can be sustained by a chain of transmission: speakers borrow the term
from others, who in turn borrow it from yet others, until the chain terminates
in those privileged users whose employment of the name is grounded in effective
knowledge of its bearer. Absent such a community, the survivor’s use of
“Aristotle” would lack the requisite anchoring in a practice of reference.
This analysis
may be extended to two of Kripke’s well-known examples against descriptivism.
Kripke observes that one might refer to Richard Feynman solely under the
indefinite description ‘an American physicist’ (1980: 81), or to Einstein under
the erroneous definite description ‘the inventor of the atomic bomb’ (1980:
85). In both cases, however, the
speaker nonetheless succeeds in referring to the intended bearer of the name.
The success lies in the fact that the name is inserted convergently into
discourse: Feynman is correctly classified within the broader kind of
physicists, and Einstein within the kind of inventors. In this way, the speaker
refers parasitically, relying upon the established linguistic practice. By
contrast, if Feynman were identified with a brand of perfume or Einstein with a
precious stone once owned by Queen Victoria, the descriptions would be
divergent. In such cases, we would deny that the way of reference had been
achieved. The speaker would fail, and indeed fail miserably, to insert the
names parasitically into discourse, notwithstanding the possibility that others
might later clarify the intended referent.
3
A central issue concerns the cognitivist account of the
contrast, emphasized by Kripke, between rigid and accidental or flaccid
designators. Kripke maintains that proper names function as rigid designators:
they pick out the same object in every possible world in which that object
exists. By contrast, definite descriptions lack this rigidity. They may
designate different objects across possible worlds, or fail to designate
altogether. For this reason, Kripke characterizes them as accidental
designators (1980, p. 48).
Kripke illustrates the point with the case
of Benjamin Franklin. The proper name ‘Benjamin Franklin’ is rigid: it designates
the same individual in every possible world in which that individual exists. By contrast, the definite description ‘the inventor of
bifocals’ is non‑rigid. In
the actual world, the description happens to apply to Franklin. Yet we can
readily conceive of a world in which John Smith, rather than Franklin, invented
bifocals, or even of a world in which Franklin existed but no one invented
bifocals at all (Kripke 1980: 98, 145).
Kripke’s celebrated distinction between
rigid and accidental designators is presented as a metaphysical discovery: the
semantic behavior of proper names, he argues, is fundamentally unlike that of
descriptions, and this difference underwrites the falsity of descriptivist
accounts of naming. Yet this claim, far from establishing a metaphysical
insight, seems to rest upon what can only be described as a metaphysical
superstition. In what follows, I shall argue that the most sophisticated theory
of reference available to us not only dispels this supposed discovery but also
provides decisive reasons for rejecting the metaphysical pretensions that
Kripke attaches to it.
The explanation I propose is that the
identification rule governing proper names functions as a rigid designator
precisely because of its vagueness and flexibility: in any possible world in
which the bearer indisputably exists, the rule applies. This, however, does not entail that the auxiliary
descriptive conditions—or even particular components of the identification rule
itself—must necessarily apply. To illustrate, consider the identification rule
for Benjamin Franklin as it might be distilled from any standard encyclopedia: “Franklin
is identified as the American statesman, inventor, and scientist, born in
Boston in 1706, a central figure in the drafting of the Declaration of
Independence, and renowned for his experiments with electricity.“ Such a rule
provides a cluster of descriptions that individually may serve to fix
reference, but they are neither essential to the name’s rigidity nor
metaphysically constitutive of its referential force.
The explanation I have is that the proper
name identification rule is a rigid designator because it is so vague and
flexible that in any possible world in which its bearer definitely exists, it
applies. But that does not mean that the auxiliary description rules, or even
parts of the identification rule itself, necessarily need to apply. Here is the
identification rule for Benjamin Franklin, with contents that can be picked out
from any encyclopedia:
RI 'Benjamin Franklin' (or equivalent name) has a bearer
See
(i-a) satisfies the localization condition of being a
person born in Boston on 17 January 1706 and died in Philadelphia on 17 April
1790 (...)
and/or
(i-b) satisfies the characterising condition of having
been a polymath, considered one of the founding fathers of the United States,
one of the leaders of the American Revolution, known for his inventions and
experiments with electricity (...)
(ii) This combination must, on the whole, be sufficiently
satisfied.
(iii) It must be satisfied better than by any other
competitor.
Before following, it is worth introducing a technique
that facilitates the explanation, which consists of laying down the
identification rule in order to transform it into a abbreviated defined description of the proper name (the individualizing
description or ID), with the characteristic of being a rigid designator, since
it says the same thing. Here is how it looks:
Benjamin Franklin's DI (or equivalent name) = The
person who (iii) best satisfies, better than any other, and (ii) overall
sufficiently satisfies the inclusive disjunction between (i-a) being born in
Boston on 17 January 1706 (...) or (i-b) being an American polymath considered
one of the founding fathers of the United States (...).
This is a very extense definite definition, but that does
not make it any less of a definite description. What is important is that it
makes the name in question a rigid designator, because, due to its vagueness
and extraordinary flexibility, it applies to all possible worlds in which
Benjamin Franklin clearly existed, unlike auxiliary descriptions or even parts
of the identification rule, which, in counterfactual situations, may not apply.
It is this possible mismatch between the reference of the identification rule
and that of the components of the cluster that makes partial and auxiliary
descriptions accidental, with nothing metaphysical about them.
There is a
compelling way to prove that my explanation is correct and that Kripke's is
illusory. We need only consider definite descriptions that are not part of the cluster
of any proper name. Since there can no longer be a mismatch between their
references and the reference of the proper name to which they belong, they
become rigid designators, applying in any possible world in which the object
they refer to exists. Examples are
rare, but they can be found. Here are a few:
1. The last ice age,
2. The assassination of Austrian Duke Franz Ferdinand in
Sarajevo in 1914,
3. The Rafflesia discovered by Dr Joseph Arnold on 20 May
1818,
4. The least rapidly converging series.
5. The 52nd English Infantry Regiment,
It does not matter when, in other possible worlds, the
last ice age ended, nor whether in those worlds the assassination of the duke
caused the First World War... All these descriptions are rigid designators,
since they cannot fail to refer to any possible world in which the process,
event, or object they refer to exists. The reason for this is that they are not
associated with any proper name with which their reference can be contrasted. I call these definite descriptions autonomous.
It is curious
that the identification rules for autonomous descriptions are exactly the same
as those we use to identify the bearers of proper names. Consider the
autonomous definite description (5): 'the 52nd English Infantry Regiment'. The
identification rule that makes this description rigid can be summarised as:
DI of ‘The 52nd English Infantry Regiment’ = The
regiment that (ii) satisfies in a wholly sufficient manner and (iii) more than
any other the inclusive disjunction between the locating condition of (i-a)
having existed from 1757 to 1881, being stationed in Oxfordshire, having acted
in the American War of Independence, the Anglo-Mysore War in India and the
Napoleonic Wars, or the characterising condition of (i-b) consisting of one or
two light infantry battalions each with approximately 1000 men recruited in
Oxfordshire.
This autonomous description is a rigid designator, as it is
sufficiently flexible to apply to any possible world in which that ancient
bastion of the empire existed.
Let us now
contrast this with examples of definite descriptions that belong to the halo or
even the semantic core of proper names’ clusters:
1. The eagle of The Hague.
2. The Iron Marshal.
3. The founder of the Lyceum.
4. The first Roman emperor.
5. The City of Light.
These descriptive predicates are accidental inasmuch as they
invite a potential referential mismatch with the identificatory rule that
secures the proper name across counterfactual circumstances. Consider, for
example, the following scenarios: Rui Barbosa’s ship might have sunk before
reaching The Hague; Floriano Peixoto might have died prior to assuming the
presidency; Aristotle might never have returned to Athens to establish the
Lyceum; Caesar might have failed to cross the Rubicon; and Paris might have
been destroyed before acquiring its reputation as the City of Light. In each
case, the auxiliary description is falsified, yet the proper name continues to
designate its bearer. The lesson is that such descriptive components are
contingent and accidental, lacking any metaphysical necessity. The persistence
of reference across these counterfactuals underscores the independence of
proper names from the descriptive clusters that may contingently accompany
them.
4
Let us now turn to the brilliant counterexamples that
Kripke and Donnellan presented against descriptivism. All of them were easily
explained in the texts. I
will choose just two here.
Let us begin
with the canonical case: Gödel–Schmidt (Kripke 1980: 83–84). Kripke asks us to
suppose that Gödel had a colleague, Schmidt, who in fact discovered the
incompleteness theorem. Shortly after completing the proof, Schmidt dies under
suspicious circumstances. Gödel then appropriates Schmidt’s manuscript and publishes
it under his own name. Now consider Maria, who knows only the theorem and
associates the name “Gödel” with the description “the discoverer of the
incompleteness theorem“. Imagine that it later becomes public knowledge that
Schmidt, not Gödel, was the genuine discoverer. On a descriptivist account,
Maria should conclude that Schmidt is the referent of “Gödel.” Yet this is not
what occurs. For Maria, Gödel remains Gödel.
The cognitivist
theory advanced here explains this phenomenon from an internalist standpoint.
The reference of “Gödel” is not secured by contingent descriptive associations
but by an identification rule that governs the use of the name. Gödel’s
identification rule can be formulated in the following definite description:
DI: ‘Gödel’ (or equivalent name) = the person who
satisfies (ii) as a whole sufficiently and (iii) better than any other
candidate, also satisfying the inclusive disjunction between (i-a) the locating
condition of having been born in Brünn in 1906, studied at the University of
Vienna, emigrated to the USA via the Trans-Siberian Railway in 1939 and worked
at Princeton until 1978, the year of his death (...), or (i-b) the
characterising condition of having been the mathematician who discovered and
published the incompleteness theorem, in addition to other important works
(...).
Any privileged user of the name Gödel will not
feel compelled to replace it with the name Schmidt just because they have
discovered that it was Schmidt who developed the incompleteness theorem, since
they know that, despite this, the name Gödel still satisfies the identification
rule much more than Schmidt. It
fully satisfies the location condition and part of the characterization
condition, since Gödel published additional notable works later.
What about
Maria, assuming she is an unskilled user who only knows the name of the
discoverer of the incompleteness theorem she studied? Well, assuming Maria is a
competent user of language, she knows that she does not know. She knows
that she is borrowing the proper name. Therefore, she will not change the name's bearer until she obtains more
information. And when she obtains it, then she will have the best reasons not
to change the bearer of the name.
I want to examine here just one of Keith Donnellan's
excellent counterexamples for its instructive character. It concerns the case
of the philosopher Thales (1972: 377). All we know of relevance about Thales,
Donnellan argues, is the definite description 'the philosopher who said that
everything is water'. He then supposes that a historian who knew very little of
the local dialect encountered an intelligent well-digger named Thales who,
tired of his profession, said: 'I wish everything were water so I wouldn't have
to dig these damn wells'. Mistaking the well digger for a philosopher, the
historian passed on the message that everything is water to others, reaching
Aristotle and the doxography we have inherited. However, Donnellan imagines
that, coincidentally, in ancient times, a hermit philosopher did indeed claim
that everything is water. If descriptivism were correct, then people should
conclude that the hermit was Thales. But the fact is that no one can deny that in this case, Thales was indeed
the clever well digger.
The answer is
easy. While I deny the explanatory relevance of the causal-historical chain, I
do not thereby deny its existence. What I emphasize instead
is what I call causal history. By this, I mean the cognitive
counterparts of nodal points within the external causal-historical chain that
have entered the public domain, thereby becoming part of the generally
recognized cluster of descriptions. This
dimension of causal history is particularly salient in the case of Thales of
Miletus, a figure whose significance is primarily historical rather than
philosophical. Here is a summary of
the characterization of Thales of Miletus, considering the doxographic elements
that belong to the causal history:
The philosopher who is referred to second-hand in
Aristotle's doxography as having been the first Greek philosopher, who in his
cosmology stated that water is the originator of the world, that it permeates
all things, that the Earth rests on water, and that all things are filled with
gods. He was also an astronomer who, according to Herodotus, Diogenes Laertius,
and Dercyllides, predicted a solar eclipse, and a geometer who, according to
Laertius, based on Hieronymus, measured the pyramids by their shadows. (Kirk
& Raven, 1971)
The insertion of elements from causal history into the
description suffices to show that Donnellan’s hermit cannot plausibly be
identified with Thales, since Thales himself is absent from the relevant
doxographical record. To be sure, the doxographers did attribute falsehoods to
him, but the falsity of those attributions does not undermine the fact that
they truly said those things. The description, therefore, remains intact.
Consequently, the characterization retains its truth, even under the
counterfactual supposition that Thales was merely an intelligent well-digger
who maintained that all things are water.
The localizing
condition settles the matter, as it remains fully satisfied by Thales, both in
form and content. He is described as:
The person who, according to the doxography of Diogenes
Laertius, was born in 640 BC and died in 549–546 BC. According to Laertius and
Herodotus, Thales was a Milesian, and Laertius wrote that he was the son of
Examynes and Cleobulina. According to Laertius
and Proclus, he visited Egypt, the traditional source of Greek science, at some
point in his life (Kirk & Raven, 1971).
Nothing in this descriptive record of causal history is
satisfied by Donnellan's hermit, who lived in ancient times and was not even
associated with the city of Miletus, and who is referred to by none of the
historical sources included in the description. Even if it were discovered that
Thales of Miletus had been a well-digger, privileged users would agree that he
still completely satisfies the locating description. Lucid indigent users, on the other hand, would suspend judgment
until they are better informed.
It is interesting to note the similarity of
our account of reference with Duns Scotus problem of individuation. Aristotle’s
distinction between form and matter was intended to secure a twofold mode of
identification: indexical reference to the individual through its matter, and
essential classification of the species and genus through its form. Yet, as
Duns Scotus perceptively observed, neither form nor matter suffices for
individuation, since both are instantiated across a multiplicity of beings. What
is lacking, Scotus argued, is the differentia individualis, the haecceity
that confers upon a particular its irreducible singularity. It is this haecceity
that renders that man Aristotle, and not Theophrastus. The theory I have sketched
above endeavors to capture precisely the individualizing characteristic sought
by Scotus: the features that underwrite the semantic integrity of proper names
or the autonomous definite descriptions.
5
I would now like to apply this theory to Frege's identity
puzzle. The puzzle is well known: why are sentences of the form a = a
uninformative, in contrast to those of the form a = b?
We can paraphrase the enigma as a problem of “identity in
difference”: how is it possible that sentences of the type a = b express
identities while at the same time expressing differences? Frege’s answer is
that a = b affirms a numerical identity in the objective reference (Bedeutung),
but a difference in the meanings, in the modes of presentation of the object.
Consider, for example, the statement "Phosphorus = Hesperus"
("The Morning Star is the Evening Star"). For Frege, this would be an
a posteriori and contingent statement, as it could be false. For Kripke,
however, it is a necessary a posteriori statement, because even if obtained
through experience, the designators ‘Phosphoros’ and ‘Hesperus’ are rigid and
must identify the same thing in any possible world.
To answer the
question, we must first consider the identity statement "Venus =
Venus." This statement identifies two rules of identification (meanings)
for the proper name 'Venus', which appear in square brackets. They can be
presented as:
Venus [[2]The planet specified as satisfying sufficiently and more than any other the
condition of being the second planet in the Solar System orbiting the Sun
between Mercury and Earth at least since some time after being so named] =
Venus [the planet specified as sufficiently and more than any other satisfying
the condition of being the second planet in the Solar System orbiting the Sun
between Mercury and Earth at least since some time after it was so named[3]].
Let us now look at the identification rule for
‘Hesperus’:
DI of 'Hesperus(-Venus)' = the celestial body usually seen
at dusk as the brightest after the Moon and considered to satisfy
sufficiently and more than any other the condition of being the second planet
in the Solar System, orbiting the Sun between Mercury and Earth since at least
some time after it was named.
I call it Hesperus(-Venus) because the meaning of
'Hesperus' emphasized by the name and underlined by me implies the
identification of Venus, since it is well known that Hesperus is also Venus.
Let us now look at the rule for identifying 'Phosphorus':
RI-‘Phosphorus-(Venus)’ = the celestial body usually seen
in the dawn as the brightest after the Moon, which sufficiently and more
than any other satisfies the condition of being the second planet in the Solar
System to orbit the Sun between Mercury and Earth since at least some time
after it was named.
Note that both Hesperus (-Venus) and Phosphorus (-Venus)
implicitly express the same rule for identifying the planet Venus, differing
only in the addition of subrules constituting their different meanings or modes
of presentation, which the diversity of names emphasizes and which I have
therefore underlined. With this unpacking of semantic rules, we can now expose
the difference between Phosphorus and Hesperus found in the different Fregean
meanings of the two names, while at the same time exposing what they have in
common: the same rules for identifying the planet Venus embedded in each name.
Let us therefore analyze the statement “Phosphorus = Hesperus”:
“Phosphorus(-Venus) = Hesperus(-Venus)”
Phosphorus [A celestial body seen in the morning as the brightest
celestial body after the Moon, considered as sufficiently and more than any
other to satisfy the condition of being the second planet of the Solar System,
which orbits between Mercury and Earth since at least some time after having
been named] = Hesperus [A celestial body seen in the evening as the
brightest after the Moon, considered as sufficiently and more than any
other to satisfy the condition of being the second planet of the Solar System
that orbits the Sun between Mercury and Earth since at least some time after
having been named.]
What is different in this sentence of the type a = b is
what is underlined, that is, the respective senses or modes of presentation of
‘Phosphorus’ and ‘Hesperus’ that flank the equality sign. What is identical in
this same sentence of the type a = b is what is not underlined, namely, the
rule of identification of Venus, which also appears flanking the equality sign.
We now have the explanation for identity in difference: the identity is that
of two implicit rules, while the difference lies in the sub-rule that makes
explicit the different senses. In considering the informative diversity of senses, Frege emphasized what
is being underlined.
We can now
identify Kripke’s error in considering this identity, as well as an error
committed by Frege himself. I want to begin by considering how
Kripke conceives of the identity presented above. For him, the statement
“Phosphorus = Hesperus” is understood by cutting off or excluding the
explicitation of the diverse senses and substituting the rules of
identification of Venus with the causal-historical chain. “Phosphorus =
Hesperus” thus becomes:
“Phosphorus(-Venus)
= Hesperus(-Venus)”
Phosphorus
[a celestial body seen in the morning as the brightest after the Moon
and (considered as sufficiently and more than any other satisfying the
condition of being the second planet of the Solar System, orbiting between
Mercury and Earth since at least some time after having been named)] = Hesperus
[the celestial body seen at dusk as the brightest after the Moon and
(considered as sufficiently and more than any other satisfying the condition of
being the second planet of the Solar System that orbits the Sun between Mercury
and Earth since at least some time after having been named)].
Of course, when considered in this way, this identity is
tautological, containing two rigid designators. Kripke’s mistake is to
disregard the difference – the part that had been emphasized – and to consider
only what was not emphasized in the whole sense of the sentence
“Phosphorus(–Venus) = Hesperus(–Venus).” This leads him to see it
as a necessary identity, though a posteriori. What he actually does is merely to
emphasize the rule of identification of Venus, which in his view must be
replaced by an assumed causal history and repeated alongside the identity. In
this respect, the identity is necessary, since it is numerical. But he does not
pay attention to the fact that the addition of the different senses makes the
identity inevitably a posteriori and contingent. (Just imagine a strange cloud
of dark substance interposing itself between Venus and us, preventing us from
seeing it.) This is how he produces what, at least in this case, is a clear
aberration: the necessary a posteriori.
Now let us see how Frege perceives the
statement “Phosphorus = Hesperus”:
Phosphorus
(Venus) = Hesperus (Venus)
Phosphorus
[a celestial body seen in the dawn as the brightest after the Moon and
considered as sufficiently, and more than any other, satisfying the condition
of being the second planet of the Solar System, which orbits between Mercury
and Earth since at least some time after having been named] = Hesperus [the
celestial body seen in the evening as the brightest after the Moon and considered
as sufficiently, and more than any other, satisfying the condition of being
the second planet of the Solar System that orbits the Sun between Mercury and
Earth since at least some time after having been named.]
Frege
ignores the implicit rules for identifying Venus. We see, therefore, that there
is also an error on his part here, which is to consider only what
differentiates the rules for identifying Phosphorus(-Venus) and
Hesperus(-Venus), ignoring what is cut off by believing that the reference (the
Bedeutung = the meaning) is the planet itself, without realising that
the same rule for identifying Venus is embedded on both sides of the identity.
However, since the subrule explaining the difference in meaning is grounded in
experience, Frege is right to regard this identity as a posteriori. There is much more to say about identity sentences, but
my intention here is to give the gist of a more inclusive way to see identity
sentences.
6
What about indexicals? In this case, we need to establish
a specific characterization, since the spatio-temporal location is already
given. But we only have epistemic access to this location because we have cognitively
internalized it. Here, I follow Frege (1918: 76): the thought (Gedanke)
expressed in the indexical utterance is independent of context, as shown by the
fact that the sentence (Satz) can be detached from context by including
the time of utterance, thereby expressing the complete thought.
Perry challenged
Frege’s view. His suggestion becomes clear if we recall his main example. He
was pushing a shopping trolley when he noticed a trail of sugar in front of
him. He went around the shelf, looking to warn the person that they were making
a mess, only to discover that the trail came from his own trolley. He reacts by
thinking: “I am making this mess” (1979: 3). According to Perry, it is
impossible to paraphrase this statement in an eternal sentence that makes the
concrete space-time function of the indexical ‘I’ disappear, since it is not
replaceable in any context. It would not be possible for him to say “Perry is
making a mess” in the case he were in the early stages of dementia and could
not remember his own name... Notwithstanding, I think it is possible to
paraphrase Perry’s statement “I am making a mess” in a Fregean manner using an
eternal sentence such as the following:
(E): At 10 o'clock in the morning on 26 March 1968, in
the food section of the Fleury supermarket in Berkeley, Perry notices sugar
falling out of his shopping trolley and thinks he is making a mess (or else
thinks, ‘I am making a mess’), and, in fact (it is true that), he is making a
mess.
It is true that the indexicals ‘he’ and ‘I’ remain. But in
line with Frege, they now refer to their meanings rather than to something out
there in the world. ‘He’ comes after a propositional attitude verb in a
complementary clause, in which it now refers to its meaning, which also occurs
with ‘I’ in the sentence in quotation marks. Here, the indexical has been
hijacked into the eternal sentence in order to express a thought independent of
context. I examined other examples from Perry with the same result (2014: ch. IV).
One objection to
my analysis is that Perry did not consider everything in sentence (E). The answer is that this is unnecessary. If this
information were given to Perry, he would tend to agree... Other people, when
commenting on the incident, may remember things like the time of day it
happened and the name of the supermarket... It is true that the private phenomenal content Perry
experienced at that moment is missing. But it is neither necessary nor possible
for that content to be contained in the indexical sentence, which is incapable
of communicating it. The important thing
about the paraphrase is that it dispels the myth of the essential indexical by
demonstrating that it does not eliminate the entirely cognitivist character of
the propositional content and, in this respect, is contextually independent of
our perceptual recognitions.
7
The proposed theory I am summarizing here can easily be
extended to general terms, except that the condition of space-time location
inevitably disappears, since predicates are instantiated across a wide variety
of locations. But the method remains Austinian: start by examining what
dictionaries have to say.
As
is well known, there is no classification of general terms that does not
overlap with others. What I did was to introduce a classification symmetrical
to that of singular terms, in which general terms are divided into: indexers,
descriptors, and nominators.
Indexers are
terms such as 'red' and 'round' that designate what is phenomenally given to
the senses. They gain their phenomenal meaning through acquaintance. The
rules for attributing these terms are learned interpersonally through positive
and negative examples in indexical situations, which makes it impossible for a
blind person to learn the phenomenal meaning of the word 'red'.
Descriptors are analogous to definite descriptions, but
with a classifying rather than an identifying function. They can take the form of indefinite descriptions, as in
‘a gold digger’ and ‘a bridgehead’. They
are more stable than indexers, just as definite descriptions are more stable
than indexicals.
Finally, there
are the nominators, traditionally called 'general names', such as 'tiger',
'water', and 'chair'. They are not only stable, like definite descriptions, but
also more flexible, like proper names. And just as there is a genetic
progression that tends to go from contextual learning given by indexicals to
the learning of definite descriptions and, finally, to proper names, there is
also a genetic progression that begins with indexers, then moves on to descriptors,
and ends with nominators. I would like to analyze two cases of nominators to
show how the proposed scheme for proper names can be adapted to them.
Consider how
Kripke considered the nominator ‘tiger’. Against descriptivism, he noted that
the description of the tiger as:
Large and fierce Asian carnivorous quadruped animal with
yellowish-brown fur, transverse stripes, and a white belly (1980: 119),
It is neither necessary nor sufficient. It is not
necessary because we can, in a possible world, find animals that satisfy Dt but
that have a genetic layout closer to reptiles than to felines; it is not
sufficient because we can imagine that evolution produces something like the
shame of the species: small herbivorous tigers that walk on their hind legs and
are as tame as rabbits.
The problem lies
in the simplified definition, which relies solely on superficial traits. Here
is a rule of attribution (Tugendhat’s Verwendungsregel) more suitably
associated with the nominative term ‘tiger,’ capable of making it a rigid
designator in the sense of being applicable in any possible world in which
tigers exist:
RA – ‘tiger’:
We attribute the concept of tiger to a specimen as follows:
(i) It is a large carnivorous mammal belonging to the family Felidae
and the genus Panthera. It is the largest cat species, native to Asia.
Its essential mark is the possession of the same mitochondrial DNA, which
enables it to interbreed with its subspecies and produce fertile offspring. It
is usually recognized externally by its dark transverse stripes, orange coat,
and white belly.
(ii) It sufficiently satisfies (i).
(iii) It satisfies (i) more than any other specimen of the genus Panthera.
Note that there is no longer any locating condition here,
since general terms can apply in a wide range of places and times.
Here we also find privileged and indigent users, as well
as the application of borrowed reference. In a classic example, a person may
think that a whale is a fish when, in fact, it is an aquatic mammal, which
still leads to a convergent use of the term. But if a child thinks that “whale”
is the name of a mountain in the Appalachians, they do not even get the general
class of aquatic animals right, so we cannot even admit that they referred to
whales in a parasitic way.
Note that there
is no longer a locative condition here, since general terms can apply in a wide
variety of places and times.
8
Although I have examined a considerable number of
conceptual terms, I now want to focus on the general term “water”, perhaps the
most discussed in theories of reference. It is essential to note that our
understanding of the word “water” underwent a revolutionary change with the
development of chemistry. From this resulted two uses or cores of meaning for
the word: the popular and the scientific.
The popular core
is the one that has been with us for thousands of years: that of an “aqueous
liquid,” transparent, odorless, and tasteless, which quenches thirst,
extinguishes fire, serves for washing, falls as rain, and covers rivers, lakes,
and seas.
It remained so
until the late eighteenth and early nineteenth centuries, when what I call the
scientific use or sense of the term emerged (Ball 2000). As Avrum Stroll
observed (1998: 71), good modern dictionaries also account for this sense,
which is summarized by the formula H₂O, but which is inferentially much more
complex when considered by its privileged users, namely chemists. They know,
for example, that the slightly negative charge of the oxygen atom is attracted
to the slightly positive charge of the hydrogen atoms of other molecules,
forming chains responsible for its higher surface tension, or that near
absolute zero water liquefies again...
It is also clear that, depending on the context of use,
one of these meanings is emphasized while the other is forgotten. For example, in a chemistry laboratory where students are
conducting electrolysis experiments, the term's scientific meaning is
emphasized. But if we consider the context of use in an Indigenous community
whose sole interest is drinking and washing, the popular meaning is emphasized.
The final attribution rule that I proposed as being well-known
to privileged users is:
RA-“water”:
We assign the word “water” to a sample if:
(i-a) the sample satisfies the constitutive characterization condition of
its popular semantic core, and/or
(i-b) it satisfies the constitutive characterization condition of its
scientific semantic core;
(ii) it satisfies these characterizing conditions to a sufficiently
complete degree;
(iii) it satisfies them more than any other inorganic compound.
This inclusive disjunction must not be confused with that
of proper names. It merely indicates that we can speak of water by reference
either to its popular core or to its scientific core of meaning. It is not
required that ordinary users know the chemical composition of water in order
for us to admit that they are capable of referring to it, nor that they possess
exhaustive knowledge of the popular core.
Thus, the old conceptual conflict between descriptivists
such as A. J. Ayer (1984) and Avrum Stroll (1998), and referentialists such as
Kripke (1980: 128) and Putnam (1975), concerning the concept of water, is
dissolved. The former admitted only the superficial popular descriptive core of
the word’s meaning, as though descriptivism were inevitably committed to
surface-level descriptions, while the latter admitted only the scientific core.
Both failed to recognize that the attribution rule accommodates both cases.
I now wish to
demonstrate the explanatory power of the attribution rule by refuting Putnam’s
famous externalist argument of the Twin Earth (Putnam 1975: 224). According to
Putnam, Oscar-1 is on Earth, and Oscar-2 is on Twin Earth, which is practically
identical to our Earth, including its history, except for the fact that the
“watery liquid” that quenches thirst, extinguishes fire, and falls as rain has,
on Twin Earth, the composition XYZ rather than H₂O. Thus, faced with a rainfall in 1750, Oscar-1 on Earth
says: “This is water!” And Oscar-2 on Twin Earth likewise says: “This is
water!” Since around 1750, the chemical composition of water was not yet known,
all that the Oscars could have in their minds was the same idea of a
transparent, odorless liquid falling as rain. Nevertheless, they referred to
different things: Oscar-1 referred to H₂O, and Oscar-2 referred to XYZ, even
though they did not know it. For Putnam, the difference is semantic, since they
“meant” different things. Hence, Putnam concluded that Oscar-1 and Oscar-2,
without knowing it, attributed different meanings to water in 1750: for the
former, water meant samples of H₂O, and for the latter, samples of XYZ.
Conclusion: meaning is not in the head! For in their heads, they had only the
same idea of the watery liquid. Meaning belongs essentially to the external
world.[4] It did not take long for a
philosopher named John McDowell (1992) to carry this reasoning to an extreme,
concluding that, since the locus of meaning is the mind, the mind itself lies
outside the head.
Keeping in view the evolution of
the attribution rule for the nominative term “water”, it becomes easy to refute
Putnam’s argument (Costa 2022: ch. 8). To begin with, it must be recalled that
to mean in English signifies both “to intend to say” (to signify) and “to point
to,” as when someone says: “I mean this table and not that chair.” What Putnam
did was to employ the word in the first sense within a context in which the
second sense ought to have prevailed. What the Oscars “meant” differently was
the referent they pointed to, not what they had in mind.
However, the sense of ‘to mean‘ as “signification,” as “what one intends
to say,” was in 1750 restricted to the popular meaning of the word, which was
in itself sufficient to sustain its application. Thus, both Oscars possessed
only this popular semantic core of water—namely, the “watery liquid” (dismissed
by Putnam) – in their heads. On the other hand, it is evident that, when we
consider “meant” in the sense of “to point to,” the Oscars were unknowingly
pointing to H₂O on Earth and to XYZ on Twin Earth.
Considering these distinctions, it becomes clear how the
semantic trick operates: when Putnam tells us that in 1750 the two Oscars
“meant” different things by the word water, he is merely inducing us,
surreptitiously, to use Oscar-1 and Oscar-2 as indexical instruments for what
we ourselves intend to say (meant in the sense of “signify”) with the word ‘water‘
in each case. After all, it is evident that almost all of us today take the
scientific core into account. We presently understand water as the aqueous
liquid H₂O when considering what Oscar-1 is pointing to, and we regard the
aqueous liquid pointed to by Oscar-2 as XYZ (whatever that may be), making it
clear that what we have in mind is different.
By a covert
appeal to our distinct mental states applied to the objects referred to by the
Oscars, Putnam managed to persuade many, as well as himself, that his Oscars of
1750 really meant water by attributing different meanings to their
referents. It is therefore unsurprising that, many years later, he confessed to
Searle that he had ceased to believe in his Twin Earth fantasy.[5]
Semantic
externalism has played a crucial role in the development of contemporary
philosophy of language. Its arguments are often profound, and without them the
dialectical trajectory of the theory under consideration could scarcely have
taken flight. Yet the position rests on a genetic fallacy: because meaning is
traced to external origins, it comes to be mistakenly conceived as possessing
an external component. The fact that the genetic fallacy admits of ever more
refined formulations does not absolve it of being what it is (cf. Kallestrup
2012). In an extended sense, “meaning” may substitute for “importance,” as when
we say that the conquest of the Americas was a “significant” external event. In
its primary sense, however, meaning is internal. It consists of rules, or
combinations of linguistic rules, that are instantiated cognitively in our
minds, and they do so continuously, independently of what happens out there.
REFERENCES:
Austin, John Langshaw. 1961. Philosophical
Papers. Oxford: Oxford University Press.
Aristotle. 1984. Nicomachean
Ethics. In Jonathan Barnes, ed., The Complete Works of Aristotle,
vol. II. Princeton: Princeton University Press.
Ayer, Alfred Jules. 1984.
Philosophy in the Twentieth Century. New York: Vintage Books.
Baker, Gordon Park. 1986.
“Criteria: A New Foundation for Semantics.” In S. Shanker, ed., Ludwig
Wittgenstein: Critical Assessments, vol. 2. London: Croom Helm.
Ball, Philip. 2000. Life’s
Matrix: Biography of Water. New York: Farrar, Straus & Giroux.
Burge, Tyler. 1973. “Reference
and Proper Names.” Journal of Philosophy 70, no. 14: 425–439.
Costa, Claudio. 2023. How
Do Proper Names Really Work? Berlin: De Gruyter.
Costa, Claudio. 2014. “A
Fregean Answer to the Problem of the Essential Indexical”, in Lines of
Thought: Rethinking Philosophical Assumptions (Cambridge Scholars
Publishing 2014) pp. 101-112.
Devitt, Michael. 1990.
“Meanings Just Ain’t in the Head.” In George Boolos, ed., Method, Reason, and
Language: Essays in Honor of Hilary Putnam, 79–104. Cambridge: Cambridge
University Press.
Donnellan, Keith. 1972.
“Proper Names and Identifying Descriptions.” In Donald Davidson and Gilbert
Harman, eds., Semantics of Natural Language. Boston: D. Reidel. Originally
published in Synthese 21 (1970): 335–358.
Dummett, Michael. 1981.
Frege: Philosophy of Language. London: Duckworth.
Ferreira-Costa, Claudio. 2023.
How do Proper Names Really Work? Berlin: De Gruyter.
Ferreira-Costa, Claudio. 2018.
Philosophical Semantics: Reintegrating Theoretical Philosophy. Newcastle
upon Tyne: Cambridge Scholars Publishing.
Ferreira-Costa, Claudio. 2014.
Lines of Thought: Rethinking Philosophical Assumptions. Newcastle upon
Tyne: Cambridge Scholars Publishing.
Frege, Gottlob. 1892. “Über
Sinn und Bedeutung.” Zeitschrift für Philosophie und philosophische Kritik 100:
25–50.
Frege, Gottlob. 1918. “Der
Gedanke: eine logische Untersuchung.” Beiträge zur Philosophie des Deutschen
Idealismus 2 (1918–1919): 58–77.
Mautner, Thomas. 2005. The
Penguin Dictionary of Philosophy. London: Penguin.
McDowell, John. 1992. “Putnam
on Mind and Meaning.” In A. Pessin and S. Goldberg, eds., The Twin Earth
Chronicles: Twenty Years of Reflection on Hilary Putnam’s ‘The Meaning of
‘Meaning’.’ New York: M. E. Sharpe.
Perry, John. 1979. “The
Problem of the Essential Indexical.” Nous 13: 3–21.
Putnam, Hilary. 1975. “The
Meaning of ‘Meaning.’” In Mind, Language, and Reality, vol. 2. Cambridge:
Cambridge University Press.
Kallestrup, Jesper. 2012.
Semantic Externalism. London: Routledge.
Kaplan, David. 1989.
“Demonstratives.” In Joseph Almog, John Perry, and Howard Wettstein, eds., Themes
from Kaplan. Oxford: Oxford University Press.
Kirk, G. S., and G. E. Raven.
1971. The Presocratic Philosophers. Cambridge: Cambridge University
Press.
Kripke, Saul. 1980. Naming
and Necessity. Cambridge, MA: Harvard University Press.
Kripke, Saul. 1971. “Identity
and Necessity.” In Milton Munitz, ed., Identity and Individuation, 135–164. New
York: New York University Press.
Recanati, François. 2022.
“Mental Files.” In Piotr Stalmaszczyk, ed., The Cambridge Handbook of
Philosophy of Language. Cambridge: Cambridge University Press.
Recanati, François. 2012. Mental
Files. Oxford: Oxford University Press.
Salmon, Merrilee. 2002. Introduction
to Logical and Critical Thinking. Belmont, CA: Wadsworth.
Searle, John. 1983. Intentionality:
An Essay in the Philosophy of Mind. Cambridge: Cambridge University Press.
Searle, John. 1967. “Proper
Names and Descriptions.” In Paul Edwards, ed., The Encyclopedia of Philosophy,
vol. VI. London: Macmillan Publishers.
Searle, John. 1958. “Proper
Names.” Mind 266: 166–173.
Strawson, Peter Frederick.
1963. Individuals: An Essay in Descriptive Metaphysics. New York: Anchor
Books.
Stroll, Avrum. 1998. Sketches
and Landscapes. Cambridge, MA: MIT Press.
Tugendhat, Ernst. 1976. Vorlesungen
zur Einführung in die sprachanalytische Philosophie. Frankfurt: Suhrkamp
Verlag.
Tugendhat, Ernst. 1983.
Logisch-Semantik Propädeutik. Stuttgart: Reclam.
Wittgenstein, Ludwig. n.d.
Wittgenstein’s Lectures 1932–1935. Edited by Alice Ambrose. New York:
Prometheus Books.
Wittgenstein, Ludwig. 2009. Philosophische
Untersuchungen / Philosophical Investigations. Oxford: Basil Blackwell.
Wittgenstein, Ludwig. 1969. Über
Gewissheit. Oxford: Basil Blackwell.
Wittgenstein, Ludwig. 1958. The
Blue and Brown Books. Oxford: Basil Blackwell.
[1] It should be emphasized that the verification rule under
discussion does not need to bear any relation to the misconstruals effected by
the logical positivists, who transformed Wittgenstein’s original pragmatic suggestions
into a non sequitur. For relevant textual evidence about that, see Alice
Ambrose (ed.), Wittgenstein’s Lectures 1932–1935, §§24–25. See also ...
(2018, ch. V).
.
[2] I include in
brackets a description of what belongs to the meanings or senses (Sinne)
of the names in Frege’s sense.
[3] With 'at least
for some time...' I want to block the objection that in another possible world
'Venus' may have later changed its orbit, been destroyed, etc., while still
identifiable.
[4] Putnam
admitted (1975: 270) that, in addition to this most important external
extensional component of the meaning of the word ‘water’, there are syntactic
markers (mass noun) and semantic markers (natural kind terms), as well
as psychological stereotypes (surface descriptions). This is correct, but it only further confuses the
credulous reader. (According to Searle, Putnam later admitted to him that he no
longer believed in his own externalism.)
[5] I took care to refute Putnam’s
other two examples in the article on Twin Earth (... 2023: 235–236), along with
some ingenious examples by Tyler Burge (1973; ... 2023: 237–239) and David
Kaplan (1989: 516–517, 531–532; ... 2023: 239–241), respectively concerning the
externality of thought and of indexicals.
Nenhum comentário:
Postar um comentário