Chap31 Lynch a Functionalist Theory of Truth

29
This excerpt from The Nature of Truth. Michael P. Lynch, editor. © 2001 The MIT Press. is provided in screen-viewable form for personal use only by members of MIT CogNet. Unauthorized use or dissemination of this information is expressly forbidden. If you have any questions about this material, please contact [email protected].

description

fg

Transcript of Chap31 Lynch a Functionalist Theory of Truth

  • This excerpt from

    The Nature of Truth.Michael P. Lynch, editor. 2001 The MIT Press.

    is provided in screen-viewable form for personal use only by membersof MIT CogNet.

    Unauthorized use or dissemination of this information is expresslyforbidden.

    If you have any questions about this material, please [email protected].

  • 31A Functionalist Theory of Truth

    Michael P. Lynch

    To copy a reality is, indeed, one very important way of agreeing with it, but it isfar from being essential.

    William James

    The question is not whether [``true'' and ``false''] are in practice applied to ethicalstatements, but whether, if they are so applied, the point of doing so would be thesame as the point of applying them to statements of other kinds, and if not, inwhat ways it would be different.

    Michael Dummett

    The history of attempts to identify the property that all and only true

    propositions have in common has not been a happy one. Traditional

    theories that aim to provide us with the essence of truthcorrespondence,

    coherence, pragmatist, and so oneach face well-known difculties that

    prevent their advocates from striking their tents and declaring victory.

    Indeed, it is just these problems that deationists point to when arguing

    that it is best to simply abandon robust theories and admit that truth has

    no underlying nature.1

    This lack of success in explaining the nature of truth is doubtless due

    to a number of causes, not the least of which is the difculty of the sub-

    ject. But when I look back over the various attemptsmany of which are

    represented in this volumeI see a particular pattern of failure. Baldly

    put, that pattern goes something like this. A theory of truth is proposed

    and argued for by appeal to propositions of a certain domain, the truth

    of which the theory seems to explain quite nicely. The theory is then

    extended to cover propositions of every domain. But this extension runs

    up against counterexamples; that is, the theory does not explain how

    propositions from certain domains can be true.

  • Consider how this problem of scope arises for a popular form of the

    correspondence theory. That theory understands the correspondence

    relation in terms of the referential/causal properties of a proposition's

    constituents.2 The proposition that the book is on the table is true in

    virtue of its components being causally/referentially related to certain

    mind-independent objectsa particular book and a particular table. But

    as plausible as this account may be when applied to propositions about

    middle-sized dry goods, it is much less plausible when applied to propo-

    sitions about numbers, such as the proposition that the number six is

    even. Whatever else numbers might be, they presumably are not physical

    objects. No number is ever in causal contact with our thought. Thus

    our thought that the number six is even can't be true in virtue of a causal

    relationship between its components and the number six. Or consider

    propositions like those we hear on the nightly news about the status of

    the economy or the constitutionality of particular laws. Some of these

    propositions are true. But economies and laws are dubious candidates

    for physical, causally efcacious objects. Further, far from being mind-

    independent, economies and laws are paradigmatically conventional.

    Thus it is a puzzle how propositions about them can be understood in

    terms of correspondence with mind-independent objects.

    I bring up the causal theory of correspondence only as an example. In

    my view, similar problems of scope await other traditional theories of

    truth as well.3 Epistemic theories are notoriously unable to explain the

    truth of propositions about humanly inaccessible parts of the universe or

    about the past. A theory that denes truth in terms of what would be

    justiedly believed cannot explain how the proposition that the number

    of the stars in the universe right now is odd can even have a truth value.

    Even the deationary view of truth faces this problem of scope. This may

    be a bit surprising because one might think that deationism would not

    have a problem explaining the metaphysical nature of truth. After all, the

    core of the position is that truth has no nature. Yet this itself is a meta-

    physical position (see Devitt, chap. 25), and if deationism is to succeed

    as its advocates wish, it must be equally plausible in every domain. Yet as

    several essays in this volume reveal, deationary views face their own

    problems explaining our commitment to certain classes of propositions

    blind ascriptions of truth, generalizations involving the concept of

    724 Michael P. Lynch

  • truth, and indeterminately true propositions, to name just three examples

    (see Gupta, chap. 23, and Field 1997).4

    Proponents of both robust and deationary theories will protest that

    there are responses available to these problems. And, of course, there are.

    I shall not argue here that each set of counterexamples is insurmount-

    able. Rather, I note that two facts stand out when looking at all these

    problems together. First, each traditional theory of truth is more plausi-

    ble in some domains than in others. Second, all of the theories mentioned

    are assuming that the question ``What is truth?'' has a single answer. In

    other words, most of the players in the contemporary debate over truth

    share an unnoticed allegiance to a certain type of monism: truth has but

    one underlying natureif any nature at all.

    Once these facts are noticed, it is hard not to see them as related.

    Therefore, it seems relevant to reconsider our commitment to this alethic

    monism. At the very least, it seems important to think through possible

    alternatives. A further reason for pursuing this course is that the monism

    that comes so naturally to philosophers runs contrary to the way most

    other people think about truth. Most folks think that different sorts of

    propositions can be true without being true in the same way. Their intu-

    itive thought is that moral or legal propositions can be true all right; it is

    just that their truth is of a different kind than that of propositions like the

    cat is on the mat. Such considerations are not decisive, but they suggest

    that alethic monism is an artifact of philosophical theory, not a result of

    ordinary practice.

    Until recently, discussions that questioned alethic monism were absent

    from the contemporary literature. But the subject has been broached by

    several philosophers in the last few years, most notably by Crispin

    Wright: ``The proposal is simply that any predicate that exhibits certain

    very general features qualies, just on that account, as a truth predicate.

    That is quite consistent . . . with acknowledging that there is a prospect of

    pluralismthat the more there is to say may well vary from discourse to

    discourse'' (Wright, 1992, p. 38). The general features Wright is talking

    about here are what he calls ``platitudes,'' or a set of a priori principles,

    about the concept of truth. These include the principles that the propo-

    sition that p is true if and only if p, that truth is distinct from justication,

    that truth is timeless, and so on. In his original statement of the view,

    A Functionalist Theory of Truth 725

  • Wright argued that any predicate that satises a minimal set of these

    platitudes qualies as a truth predicate. But in some discourses, the truth

    predicate may have more robust content, depending on the types of

    additional platitudes about truth that the discourse brings along in its

    wake. In short, Wright's view allows for the possibility that there may be

    different kinds of truth.

    In recent essays, both Hilary Putnam and Terry Horgan have also

    entertained this idea (e.g., Putnam, chap. 30; Horgan, chap. 4).5 Most

    philosophers, however, continue to look askance at the suggestion that

    truth itself might come in different kinds. Different kinds of truths

    (propositions), yes; different kinds of truth, no. The prevailing thought

    seems to be that even if the nonphilosopher might sometimes talk as if

    she thought moral truth was of a different kind than physical truth, this is

    not a claim we should take seriously; such talk is just that: talk.

    One reason for this skepticism is that a plurality of kinds of truth

    seems to imply a plurality of truth concepts. And a plurality of truth

    concepts entails that the word ``true'' is ambiguous.

    The worry is a real one. The ambiguity of truth would have several

    ill consequences. First, it would undermine one of the most useful and

    important functions of the truth concept. As is often remarked, our

    concept of truth allows us to generalize ``blindly'' over propositions of all

    sorts, as in ``Everything Socrates said was true.'' If the word ``true'' were

    ambiguous in the way that ``bank'' or ``step'' happen to beif its meaning

    completely changes from context to contextthen attempts to ascribe

    truth to everything Socrates ever said, no matter what the context, would

    be impossible. Yet clearly we can generalize over propositions in this

    way, and hence truth must not be ambiguous.

    Second, the equivocality of truth would make validity and logical

    inference mysterious. Instances of valid argument forms whose premises

    are mixed, i.e., whose premises stem from different discourses, would end

    up equivocating (see Sainsbury 1996). Consider a simple argument:

    If violence causes pain, then it is wrong.

    Violence does cause pain.

    Therefore, violence is wrong.

    726 Michael P. Lynch

  • This is obviously valid; its premises are truth-preserving. Yet its second

    premise is a descriptive statement about cause and effect relations in the

    world, while its conclusion is a moral evaluation. Were the ambiguity

    thesis to hold with regard truth, these statements could be true in com-

    pletely different senses; they could literally have different properties, each

    expressible by the predicate ``true.'' The argument might no longer ``pre-

    serve'' a single property at all.

    In my view, these considerations suggest that our semantic or concep-

    tual account of truth must be uniform across context.6 But it does not

    imply, I shall argue, that our account of the deep nature of truth must be

    similarly uniform. We can be monists about the concept of truth while

    being pluralists about its underlying nature. The key is to see the concept

    of truth as the concept of a multiply realizable property.7

    1 Truth as a Functional Concept: A First Pass

    In the philosophy of mind, it is common to the point of dogma to say that

    mental states are functional states. A particular mental state such as pain,

    or the belief that it will rain, is the state that it is in virtue of the func-

    tional role it has within our overall cognitive system. But what realizes or

    plays that role in any particular organism may vary: mental states are

    therefore said to be ``multiply realizable.''

    The terms ``function'' and ``functional role'' mean different things to

    different people.8 In the general and intuitive sense I am concerned with,

    a function is a type of job. To occupy a functional role is therefore to

    satisfy a certain job description. Consequently, a functional concept is

    the concept of a property, state, or object that occupies or plays such a

    role. So for example, being a head of state, being a carburetor, and being

    a heart are all functional concepts in my sense. A heart is anything that

    has the function of pumping the blood throughout the body of an organ-

    ism, and various structures, organic and inorganic, might play or ``real-

    ize'' this role within some particular cardiovascular system. Indeed, the

    underlying nature of what plays the heart role in a given system can be

    quite different in different organisms. But it is the underlying nature of

    whatever plays that role that explains the performance of that function.

    A Functionalist Theory of Truth 727

  • As an intuition pump, consider one of our most basic platitudes about

    truth: a proposition is true when the world is as that proposition says

    that it is. This platitude (and its more formal brethren, the disquotational

    and equivalence schemata) captures something about the very essence of

    truth. Yet paradoxically, it tells us nothing specic about that essence.

    Robust theorists take platitudes of this sort as the starting point in any

    theory of truth, a minimum requirement that every theory must meet,

    while deationists take it to be the end of any such theory. Both attitudes

    overlook a subtly different way of understanding the main lesson of the

    platitude. Rather than attempting to read it as the key to truth's deep

    metaphysical nature (or lack thereof), we can understand our platitude

    as specifying truth's ``job'' in our conceptual schemeits functional role.

    We can understand it as telling us that our concept of truth is the concept

    of whatever property a proposition has when the world is as that prop-

    osition says that it is. Roughly speaking, ``saying it like it is'' is part of the

    functional role of true propositions, and propositions that do so (and

    also fulll various other conditions, as we shall see) have the property of

    truth.

    Consider another analogy.9 The position of head of state is found in

    almost every constituted government. It is held by presidents, prime

    ministers, kings, queens and even religious gures, all of whom are head

    of state in virtue of performing a certain job, namely, being the chief

    executive ofcer for the government. This job will vary from country to

    country in some respects (although it will remain constant in others), and

    individual heads of state may have properties that others do not have

    (such as being elected, which is true of presidents and popes but not of

    most kings). And yet when we say that both Fidel Castro and Bill Clinton

    are heads of state, we are not equivocating. They simply perform or realize

    the functional role of being a head of state differently. In the same sense,

    for the functionalist about truth, truth talk is not equivocal, because the

    concept of truth is everywhere understood as naming a particular func-

    tional role. Nonetheless, what realizes that role may vary from context to

    context.

    Much more needs to be said, of course, to make sense of this idea. But

    we can already see that a functionalist theory of truth will leave room

    for the possibility that in the case of propositions about middle-sized dry

    728 Michael P. Lynch

  • goods, the claim that a is F may be true in virtue of the referential prop-

    erties of that claim. Causally understood correspondence relations might

    realize the functional role of truth in such cases. Yet the functionalist

    account could also allow a different propertycoherence with other

    propositions, for instanceto occupy the role of saying how things are

    in other contexts. Functionalism about truth is therefore consistent with

    alethic pluralism at the deep metaphysical level: the underlying nature of

    what realizes truth in any particular mode of thought is still an open

    question.

    If the concept of truth can be seen as naming a functional role, we

    should be able to specify that role in an informative way. Doing so is the

    goal of the next section.

    2 The Alethic Network and the Role of Truth

    According to what is often called ``commonsense'' functionalism, mental

    concepts such as belief and desire come together in a package deal. Like

    the concepts of solider/army and teacher/student, they apply together or

    not at all (Armstrong 1999, 84). These concepts are not individuated one

    by one but by their place in the network of interrelated psychological

    generalizations that make up our commonsense psychology, or ``the psy-

    chological platitudes which are common knowledge among usevery

    one knows them, everyone knows that everyone else knows them, and so

    on'' (Lewis 1972, 208). These and an indenite number of other such

    platitudes jointly carve out the causal roles of our mental states. If we

    take the mental state of pain, for example, the relevant platitudes will

    include such chestnuts as ``People who are in great pain are usually not

    happy about it,'' ``One is typically aware of one's pain,'' and ``A threat of

    pain typically causes fear.'' String such platitudes together and we have a

    job description for pain, namely, that pain is a state that is typically

    causally related to certain inputs, outputs, and other mental states. The

    concept of pain will therefore apply to any property or state that ts this

    job description, that realizes this causal role. As Lewis says, a mental

    concept is ``the concept of a member of a system of states that together

    more or less realize the pattern of causal generalizations set forth in

    commonsense psychology'' (1980, 112). So the concept of pain, on this

    A Functionalist Theory of Truth 729

  • account, is the concept of the state that more or less realizes the role of

    pain in our psychology, and whatever state does realize that role is pain.

    A similar model helps us understand our concept of truth. Like our

    psychological concepts, what we might call our alethic concepts, or con-

    cepts like truth, fact, proposition, and reference, package-deal concepts.

    It has long been recognized that concepts like proposition and fact are

    denable only in terms of each other: a proposition is whatever is true or

    false, a fact is what makes propositions true, true propositions t the

    facts, and so on. The fact that these sorts of concepts form such a tight-

    knit family is part of what needs to be explained by any theory of truth.

    Adopting a functionalist perspective on these concepts does just that. Just

    as our psychological concepts are denable in terms of their roles in a

    network of interrelated psychological platitudes, so the common-sense

    set of principles and platitudes that together constitute our having a

    robust sense of the true and the real form the alethic network. Some of

    the most central of these principles are the following:

    . The proposition that p is true if and only if p.

    . The proposition that p is false if and only if it is not the case that p.

    . Propositions are what are true and false.

    . Every proposition has a negation.

    . A proposition can be justied but not true, and true but not justied.

    . True propositions represent, or correspond to, the facts, and false onesdo not.. Facts are what make propositions true.

    Crispin Wright (chap. 32) suggests that platitudes of this sort can be taken

    to compose what he calls an ``analytic theory'' of truth. On his account,

    these principles jointly determine the meaning of the word ``true'' without

    providing necessary and sufcient conditions for the application of the

    term.

    Like Wright, the functionalist sees such principles as constituting our

    grasp of the concept of truth. But there are also some interesting differ-

    ences between Wright's view and an explicitly functionalist theory. First,

    it is important to see that on the functionalist account, these and other

    alethic principles aren't simply a list. They form a structure. In the phi-

    losophy of mind, the relevant principles can be divided into those that

    730 Michael P. Lynch

  • primarily concern the inputs and outputs of the system (typically, sensa-

    tions and behavior, respectively), and those that concern relations between

    the mental states themselves. Not all of the principles involved need fall

    neatly into one category or another, although most will, and not all

    mental states need be functionally dened, although most will. Similarly,

    some alethic platitudes will primarily concern the relationship between

    truth and other closely connected core notions (fact, proposition), while

    some relate these concepts to those that are intuitively farther outside

    these core concepts. As an example of the latter, consider the following:

    . To claim that p is true implies that one believes that p.

    . One knows that p only if it is true that p.

    . Honest people typically speak the truth.

    . Deliberately asserting what you know to be false is a lie.

    Relative to core platitudes concerning truth and fact, principles like the

    above serve to relate the core concepts to others in our overall cognitive

    system. Again, it need not be entirely clear how ``central'' a platitude is,

    or whether it primarily concerns relating alethic concepts to nonalethic

    concepts. Here, like everywhere else, types of concepts shade off into one

    another. And while central alethic concepts will be functionally under-

    stood, it will remain an open question whether concepts further removed

    from the center (concepts like knowledge, for example) will also be

    functionally dened.

    In the case of human psychology, most of the platitudes will be causal

    in nature (e.g., ``Pain causes worry''). But not all will be. Others, like ``A

    toothache is a type of pain,'' will be quasi-logical. With regard to alethic

    terms, this order is reversed: one would suspect that most of the alethic

    principles would be quasi-logical, although there is nothing to rule out

    the possibility that some may also be causal.10

    If we grant the existence of the alethic network as a whole, we can

    characterize each of the alethic concepts in terms of the role it plays

    within the network. This is a second way in which the functional account

    of truth differs from Wright's analytic theory. Since truth, fact, proposi-

    tion, etc., are explicitly package-deal concepts according to functional-

    ism, the same platitudes that demarcate our concept of truth will also

    demarcate our other alethic concepts. In short, we can take a core alethic

    A Functionalist Theory of Truth 731

  • concept as the member of a system of properties that together realize the

    interlinked structure of platitudes partially set forth above. Thus truth,

    for example, can be seen as whatever property plays the role demarcated

    by the associated principles and platitudesthe truth role, in other

    words. Other central members of that system will similarly be function-

    ally understood. Propositions are whatever is true or false, whatever is

    asserted or denied, that which can take the place of the unquoted vari-

    able in `` `p' says that p,'' and so on. Similarly, we can understand a fact

    as whatever makes a proposition true, as conforming to the schema ``It is

    a fact that p if and only if p'' etc.

    The functionalist can spell this out in greater detail, and with much

    more rigor, by employing F. P. Ramsey's and David Lewis's method for

    dening theoretical terms (see Lewis 1970). This method treats the term

    (or concept) being dened as if it were a new term being introduced into

    an ongoing scientic theory. To employ the method here, we would rst

    form a conjunction of all, or some weighted subset, of the commonsense

    or ``folk'' platitudes about truth and related concepts in our network, A.

    This conjunction will include both T terms and O terms. T terms are the

    names for properties with unspecied natures that the theory is being

    used to introduce or dene, in this case, terms like ``true'' or ``fact.'' O

    terms are the terms in the ``old,'' introducing vocabulary, such as

    ``cause,'' ``object,'' ``person,'' ``snow,'' and so on. To make things easier,

    we'll stipulate that each T predicate explicitly represents a property (so

    that, e.g., ``x is true'' becomes ``x has the property of being true''). We

    can then imagine A being written down in one long sentence roughly like

    this:

    AT1; . . . ;Tn;O1; . . . ;OnThe next step is to replace the T terms (but not the O terms) with vari-

    ables, and then to prex an existential quantier for each variable. The

    result is the modied ``Ramsey sentence'' of A. If ``t1'' is a variable

    standing in for ``true,'' then the modied Ramsey sentence of A could be

    used to dene ``true'' roughly as follows:

    (FT) x is true,df bt1; . . . ; btnAt1; . . . ; tn;O1; . . . ;On & x has t1This says that x (a proposition, say) is true just when there are certain

    alethic properties t1; . . . ; tn that are related among themselves as well to

    732 Michael P. Lynch

  • nonalethic properties as specied in A and ``x has t1.'' By functionally

    dening the word ``true,'' (FT) gives us a way of understanding the con-

    cept of truth according to which a proposition is true just in case it has a

    property that plays the truth role marked out in A. Further, (FT) reveals

    that the identity of every alethic concept depends on its relation to every

    other concept; in a sense, we dene each concept not one by one but all

    at once, or en masse. That is, we dene every alethic concept in terms of

    having a property that uniquely bears certain relations to the other

    properties expressed by our alethic concepts and to the referents of the O

    terms.

    So, to be true is to play the truth role. But functionalism allows that

    this role might be realized or occupied by different properties. We can say

    that a property realizes the truth role for a discourse just when it is the

    unique realizer (or near perfect realizer; see below) of that role for the

    propositions that compose the discourse. Thus propositions from two

    different discourses may have distinct properties that realize the truth

    role. Naturally enough, how one understands the nature of a particular

    discourse will determine which property, if any, one takes as uniquely

    playing the truth role for that discourse. For example, if one understands

    moral propositions to be about the subjective feelings of the speaker,

    then this metaphysical position will rule out moral propositions' being

    true in virtue of their relation to mind-independent facts. As we will

    shortly see with the case of legal discourse, one's overall metaphysical

    views provide a collection of a priori constraints or conditions on the

    realization of truth for that discourse. But functionalism itself remains

    neutral on just what these constraints will be.

    We should not expect there to be a sharp and clear line between dis-

    courses or forms of thought. In ordinary life, we know the difference

    between talking about physics and talking about ethics. But sometimes

    we may say things that don't clearly fall into either category, as we might

    if we were involved in a discussion about the ethical consequences of cer-

    tain physical experiments for instance. It follows that we should expect

    some vagueness as to what discourse a particular proposition belongs.

    But I don't need to know what discourse I am engaging in to know what

    I mean when I say that something is true. According to the functionalist

    view, it is a fact about the concept of truth that no matter what discourse

    A Functionalist Theory of Truth 733

  • I may happen to be engaged in, what I say will be true just when it has

    a property that plays the truth role for that discourse. If what I say,

    the proposition I express, is not clearly a member of one discourse or

    another, how its truth is realized will also be unclear. But that is a ques-

    tion for deep metaphysicsnot something I need to know in order to

    understand what I mean by the word ``true.''

    A functional characterization of the alethic concepts ts well with

    our holist intuitions about concepts like truth and fact. Because of their

    obviously interconnected nature, we learn such concepts not one by one

    but more or less all at once, as the functionalist theory implies. This is

    really no more mysterious than learning a complex skill like riding a

    bicycle (Heil 1998, 102). Acquiring this skill requires the coordination of

    a whole set of tinier skillsbalancing, pedaling, steering, etc.none of

    which are learned in isolation. You learn to master one as you come to

    master another, and vice versa. The alethic concepts are similar in this

    respect. Here as elsewhere, ``light dawns gradually over the whole''

    (Wittgenstein 1969, sec. 141).

    So far we've seen how (FT) explicates the concept of truth. But we also

    want to know what truth is, or about the property of truth. Those

    familiar with the debates over psychological functionalism know that we

    face a choice at this point. Functionalist theories in the philosophy of

    mind differ over whether to take mental properties as identical to the role

    properties or the realizer properties. The rst alternative sees any given

    mental property as a ``higher-order'' property, or the property of having

    a property that plays a certain causal role; the second identies properties

    with the ``rst-order'' properties, or properties that realize that causal

    role in a system. In the same vein, we must choose between saying that

    the property of truth is the higher-order property of having a property

    that plays the truth role and saying that it is identical to the lower-order

    property that realizes that role in a particular discourse.

    There are two reasons to prefer the former alternative when it comes

    to truth. First, to identify truth with its realizer property in some context

    (coherence, correspondence or whatever it might turn out to be) would

    immediately raise the ugly problem of ambiguity. Should it turn out that

    the truth role is realized by different properties in different contexts, we

    would no longer be able to talk of truth simpliciter but only of moral

    734 Michael P. Lynch

  • truth, mathematical truth, physical truth, and so on. Second, truth is our

    chief cognitive goal. We want our beliefs to be coherent, or to correspond

    to fact, because we want them to be true, not the other way around. But

    this means that in cognition we are aiming at truth itself, not the prop-

    erties that exemplify or realize it. Therefore, we should take truth itself as

    the higher-order or role property.

    If we do so, then we can still be pluralists about the realizer properties

    of truth but without any threat of ambiguity. This means that by adopt-

    ing functionalism, the alethic pluralist can explain universally applicable

    concepts like validity as well as can the alethic monist. Valid inference

    preserves truth. According to functionalism, truth is a higher-order

    functional property. Therefore, on the functionalist account, valid infer-

    ence preserves that higher-order property. Of course, what makes this or

    that premise true (what realizes its truth) may be quite different from

    what constitutes or realizes the truth of the conclusion. But that is a

    question of deep metaphysics, not a formal, conceptual concern.

    By acknowledging that truth is always and everywhere the property of

    playing the truth role, functionalism is not abandoning pluralism. Con-

    sider the case of a mental state like pain. According to functionalist

    accounts that take pain as a (higher-order) role property, there is indeed

    a sense in which the nature of pain is uniform across species. This is

    because, by and large, the pain role is uniform across species. But

    explaining the pain role does not explain what pain is in a more funda-

    mental sense. It does not explain how that function is performed in a par-

    ticular organism. For that, we must look to the details of the organism's

    neuronal structure: we must look for the lower-level property that real-

    izes the pain role. In the same way, the functional role of truth does not

    explain how that role is lled in a particular discourse. For the underly-

    ing nature of truth, we must look to the details of the type of thought in

    question.

    3 A (Brief) Case in Point: Juridical Truth

    Let us consider a type of thought where the underlying nature of truth is

    plausibly other than correspondence with mind-independent facts. The

    domain in question is the law, or juridical truth. I emphasize that the

    A Functionalist Theory of Truth 735

  • discussion will be extremely rough. My overall concern isn't with the

    nature of juridical truth itself. In this paper, juridical propositions are an

    example to aid in spelling out my main concern: the abstract structure of

    a functionalist theory.

    When I write ``juridical propositions,'' I have in mind what Ronald

    Dworkin calls ``propositions of law'' or ``all the various statements and

    claims people make about what the law allows or prohibits or entitles

    them to have'' (1986, 4). This includes both general claims, such as the

    proposition that segregation is illegal or that the law protects ag burn-

    ing, and specic propositions, e.g., that Exxon must compensate Alaska

    for an oil spill. Propositions like this are surely capable of being true. We

    assume that in most cases, ag burning is or isn't protected under the law

    and that Exxon either is or is not required to pay compensation.

    In a platitudinous sense, juridical propositions like these will be true

    when they correspond to the facts, tell it like it is, and so on. But in

    this case, it seems unwise to read the platitude as picking out a substan-

    tive metaphysical relation between propositions and mind-independent

    objects. Laws and legal entities are in some sense conventional; they are

    human constructs. But this hardly means that propositions of law are

    incapable of being true or false. It means only that it is unlikely that they

    are true in virtue of referential relations with mind-independent objects

    and properties. It is innitely more plausible that, as Dworkin notes,

    ``propositions of law are true or false (or neither) in virtue of other, more

    familiar kinds of propositions of which these propositions of law are (as

    we might put it) parasitic. These more familiar propositions furnish . . .

    the `grounds' of law'' (1986, 4).

    Dworkin's point is that legal propositions are naturally thought of

    as true because of their relation to other ``grounding'' propositions, not

    because they correspond to mind-independent objects called ``laws'' (or

    worse, ``The Law''). Of course, just what is included within the grounds

    what types of propositions, in other words, make propositions about

    the law trueis a matter of serious dispute. But at the very least, they

    include those propositions expressed in the Constitution, previous stat-

    utes, and past judicial decisions. It is these sorts of propositions that

    we think matter for whether it is true that a particular corporation is

    736 Michael P. Lynch

  • required to pay compensation. Collectively, we might refer to them as the

    body of law.

    This suggests a particular way of thinking about purely legal truth.

    Roughly, we think that a proposition of law is true when it coheres with

    its immediate grounds and with the grounds of propositions inferentially

    connected to it. In short, legal truth consists in coherence with the body

    of law.

    A virtue of the functionalist theory is that it helps to explain this intu-

    itive thought. Without the functionalist theory, the idea that legal truth

    consists in coherence with the body of law would seem to raise the ugly

    ambiguity problem. But with a functionalist story of truth in place, we

    can say that when it comes to propositions of law, the truth role is real-

    ized by a coherence relation of some sort. The fact that purely legal

    propositions concern conventional, mind-dependent matters does not

    mean that they are not really true or that they have a second-class status,

    but only that their truth is realized by a property different than what

    realizes truth about the physical world.

    Of course, what it means for a proposition to cohere with the body of

    law would be (were this proposal worked out) a matter for substantive

    metaphysical inquiry. One question concerns the strength of the coher-

    ence relation involved. For example, for it to be true that ag burning is

    protected under the law, is it enough that this proposition cohere with

    the body of the law now? Or, weaker still, would it be enough that it was

    presently believed to cohere with the body of law? The latter certainly

    seems too weak. We need to allow room for mistakes about the lawfor

    objectivity, in other words. One suggestion would be to understand

    juridical truth along the lines of what Wright calls ``superassertibility.'' A

    proposition is superassertible when it is ``justied by some accessible

    state of information and will continue to be so justied no matter how

    that state of information is improved'' (1999, 236). In other words, to be

    superassertible is to be durably justied without defeaters. Roughly,

    Wright's view (1999, 228; see also this volume, chap. 32) is that super-

    assertibility is a legitimate realizer of the ``truth concept'' for a class of

    propositions when those propositions are knowable and when evidence

    for or against such propositions is (in principle, at least) always accessi-

    ble (1999, 236). These conditions seem to be met by legal truth. It is a

    A Functionalist Theory of Truth 737

  • priori that legal truths are all knowable. It would be absurd to think that

    certain actions could be illegal without anyone even being capable of

    determining that they are. Second, the states of information that bear on

    the truth of juridical propositions, such as the empirical facts and the

    written laws themselves, are accessible to human investigationat least

    in principle. Thus perhaps what makes a proposition of law true is that it

    durably or continually coheres with the body of the law. If so, then what

    is believed to be legal or illegaleven by a judge or the entire legal

    communitymay not be. In short, juridical truth might turn out to be

    realized by ``supercoherence'' with the body of law, where a proposition

    can fail to have this property even if it coheres with the law in the short

    run, or coheres with judicial decisions that are later overturned.11

    Much more would have to be said to make the case that juridical truth

    is realized by either coherence or supercoherence with a body of law. But

    I won't be saying it here. I mention it simply as an illustration of what it

    would be for truth to be realized by something other than correspon-

    dence with mind-independent fact.

    4 A Note on Realization

    Denitions like (FT) allow for multiple realization: in different contexts,

    different properties may occupy the functional role it describes. Yet (FT)

    describes that role in terms of the conjunction of all the alethic platitudes.

    If so, then one might wonder how any sort of epistemic property like

    supercoherence could play that role. For one may well think that one of

    our alethic platitudes is that every proposition is either true or false.12

    But it is a matter of serious dispute (see, e.g., Dworkin, 1986, 37 ff.)

    whether this platitude applies in the legal case. Further, one might think

    that if legal truth is a matter of coherence or supercoherence, then there

    may be juridical propositions that are neither completely true nor com-

    pletely false. Perhaps there are cases where a proposition is neither super-

    coherent nor not supercoherent with a given body of law, because all

    possible evidence would fail to determine the matter. In any event, if such

    cases are possible and if juridical truth is realized by supercoherence,

    then not every juridical proposition need be determinately true or false.

    But how could this be if (FT) implies that whatever plays the truth role

    738 Michael P. Lynch

  • must satisfy all alethic platitudes? The worry, in short, concerns how

    (FT) could be realized by such different types of propositions.

    The solution is to see that a putative realizer property needn't satisfy

    every last one of the platitudes in A in order to count as realizing the

    truth role. After all, sophisticated functionalists in the philosophy of

    mind (those of the commonsense or analytical sort that we've been dis-

    cussing) have long acknowledged that there might not be any single

    neural property that perfectly satises every folk-psychological platitude

    connected with pain (e.g., Lewis 1972). Indeed, it would be surprising if

    the world was so accommodating of our folk intuitions. It seems much

    more sensible, both in the alethic and psychological cases, to count prop-

    erties that are nearly perfect in realizing the role in question as realizers

    of that role.

    Recall that a functional role is a job, and (FT) is a job description.

    Suppose that we were wondering who was doing a certain job in a big

    company. It could turn out that no one person could fulll management's

    written job description more than slightly. No one would then be doing

    that job. Or it could turn out that two or more people were each doing

    various aspects of the job, in which case, again, no one person would

    have that job. Or it could turn out that while no one could do every aspect

    of the job in that company, one person was in fact doing 90 percent of

    the job. In that case, no one was doing the job perfectly, but the job was

    getting done.

    In the same fashion, we can allow that while there may be no perfect

    realizer in a discourse of the role marked out by (FT), there may be a

    property that is a near perfect realizer of that role in that discourse. In

    such cases, we can take whatever property uniquely fullls the truth role

    near enough to be the property of truth in that discourse. If there is no

    property that comes close to realizing the role in the context, or if there

    are two or more properties that equally fulll the role in that context, the

    discourse in question will remain unalethic.

    Of course, what counts as ``near enough'' may be rather difcult to

    pin down precisely. Clearly, a property's fullling all but one of the

    platitudes in (FT) would seem to be near enough. Yet it seems reasonable

    to hold that the principles that comprise the alethic network may have

    different weight or importance within that network. Relatively central

    A Functionalist Theory of Truth 739

  • principles like the equivalence schema or the idea that a proposition can

    be justied but not true will likely receive more weight than, e.g., the

    intuition that it is propositions that are true or false. Thus it may be that

    satisfying many of the heavily weighted principles will count more than

    satisfying all of the less heavily weighted ones. If this amendment is

    allowed, and there seems no reason why it shouldn't be, one can give

    truth a functional denition in terms of the total set of alethic platitudes

    but still allow that looser forms of thought can be true or false.

    5 Objections

    In discussing functionalism with other philosophers, I sometimes en-

    counter puzzlement over how this theory differs from minimalism or

    deationism about truth. I suspect that the main cause of this puzzlement

    is that the typical deationist, like the functionalist, rejects the traditional

    correspondence and coherence accounts of truth. Further, the reasons

    given for this rejection are often similar, e.g., that the traditional theories

    fail to explain what truth is across the board (see, e.g., Horwich 1998,

    12). Nonetheless, the lessons that minimalists draw from the failure of

    the traditional theories are quite different from those drawn by the func-

    tionalist. From the fact that the traditional theories can't explain how

    truth can have the same nature in every discourse, minimalists conclude

    that truth has no nature. This is often put by saying that the word ``true''

    fails to express a genuine or substantive property of propositions. In

    contrast, the functionalist holds that truth does have a nature, and that

    ``true'' does express a substantive property.

    This metaphysical difference between minimalists and functionalists

    about truth is like the difference between eliminativists and functionalists

    in the philosophy of mind. Eliminative materialism is the view that folk-

    psychological concepts like pain, belief, desire, etc., fail to pick out any

    actual physical properties or states. According to the eliminativist, there

    is, strictly speaking, no such thing as the nature of pain. The function-

    alist, on the other hand, holds that pain is a functional property, one that

    can be realized by distinct brain states in different species. The function-

    alist does not deny that pain has a nature; she holds that it can have more

    740 Michael P. Lynch

  • than one. In the same way, an alethic functionalist holds that truth is a

    higher-order property that can be realized differentlyor have a different

    naturein different discourses.

    Besides the metaphysical differences between deationary theories and

    alethic functionalism, there is also a signicant epistemological differ-

    ence. Deationists hold that truth is a philosophically unimportant con-

    cept. This is sometimes put by saying that we do not need to appeal to facts

    about truth in order to explain, e.g., knowledge, meaning, or any other

    philosophically troubling notion. Not so for functionalism. Just as under-

    standing the underlying facts about how pain is realized is important for

    understanding a host of other psychological phenomena, so understand-

    ing the underlying facts about how truth is realized is essential (on the

    functionalist view, at least) for understanding a host of related philo-

    sophical concepts. Unlike the deationist, the functionalist theory sees

    truth as having a job that impacts the entire philosophical economy.

    Of course, there are differences between psychological and alethic

    functionalism. One might argue that these differences are greater than

    I've allowed, and as such, they undermine the case for calling alethic

    functionalism a type of ``functionalism'' at all. One objection of this sort

    is the following. In the case of mental concepts, analytic functionalists

    typically hold that it is a contingent a posteriori matter as to what occu-

    pies, e.g., the pain role. But surely it is an a priori matter whether some

    property plays the truth role, and if so, then perhaps this property must

    play that role in every discourse.13 And this might be thought to under-

    mine the claim of alethic functionalism that truth is multiply realizable.

    In fact, there is not much difference between alethic and psychological

    functionalism on this matter. For according to alethic functionalism,

    truth in any discourse is the higher-order property of having a property

    that plays the truth role for that discourse. So realizers, as we noted

    above, are always realizers for a discourse. And as we saw in the case of

    legal discourse, what plays the truth role in any given discourse is deter-

    mined by the nature of the discourse in question. Analogously, what

    plays the pain role for a given organism depends on that organism's

    material constitution. Given the way the organism is constituted, there

    will be one and only one property that best plays the pain role. And given

    A Functionalist Theory of Truth 741

  • the way a discourse is constituted, there will be one and only one prop-

    erty that best plays the truth role. So what plays either sort of role is a

    contingent matter in that it is a contingent matter as to how any organ-

    ism or any discourse is constituted. Given that constitution, what plays

    the role in question will do so necessarily. The difference between the two

    cases rests only in the differences in the subject matter. Figuring out what

    realizes the pain role in a particular organism is an a posteriori matter;

    guring out what realizes the truth role in a particular discourse is

    (mostly) an a priori task. But these are epistemic facts; they do not impact

    the metaphysical question of multiple realizability at all.

    A different type of worry is that functional denitions of alethic con-

    cepts like (FT) will be subject to what Michael Smith (1994) has called

    the ``permutation'' problem. According to Smith, the permutation prob-

    lem arises for functional analyses of a type of concept when the plati-

    tudes used in the denitions are so tight-knit that they don't connect the

    things to which the concept apply to things outside the network. When

    that happens, Smith suggests, we won't have enough relational informa-

    tion left (after we've stripped the platitudes of the type of concepts we

    wish to dene) to x on any unique properties that we can identify as the

    realizers of the concepts. This means that there may be more than one

    property realizing every concept in the network. As a result, we lose the

    ability to distinguish the members of the network from each other.

    As Smith acknowledges (1994, 54) one can never be sure that a given

    functional analysis is subject to the permutation problem unless one

    actually goes out and completes the analysis for each of the concepts in

    question. Fortunately, we can be condent that alethic functionalism is

    immune from the permutation problem without engaging in this task.

    Two points are relevant. First, there are plenty of platitudes connecting

    alethic concepts to other sorts of concepts. It is these platitudes that dis-

    tinguish the various alethic concepts (and their realizers) from each other.

    Consider, for example, the differences elicited by the following simple

    principles: the fact that the butler's prints are on the murder weapon can

    cause us to suspect that he is guilty of the crime, but no proposition, true

    or false, can cause anything; facts make propositions true, not vice versa;

    propositions can be doubted, facts cannot be; and so on.

    742 Michael P. Lynch

  • Second, what determines the best realizer for the truth role isn't the

    alethic network alone. As I've repeatedly emphasized, part of what

    determines which property best realizes being true (or being a fact or

    being a proposition) in a discourse depends on a priori facts about that

    discourse. If, being committed physicalists, we were to take it as a priori

    that there could be no properties such as rightness or goodness, then the

    range of possible realizers for truth in the moral realm is signicantly

    limited. Again, just as an organism's neural structure will determine what

    unique property best plays the pain role, so the constitution of a partic-

    ular form of thought determines what unique property best plays the

    truth role.

    The nal objection I'll discuss also concerns functionalism's analysis of

    the concept of truth. In the philosophy of mind, Ramsey-Lewis denitions

    are frequently taken as reductions of our mental vocabulary to physical

    vocabulary (see Lewis 1972). The idea is that since functional denitions

    of mentality contain no mental terms (only variables), we can be said to

    have explained our mental concepts without circularity. One objection to

    alethic functionalism is that a moment's reection indicates that the pool

    of O terms (terms that don't presuppose alethic concepts) in our alethic

    platitudes is too small to do the job required. There are bound to be

    concepts in (FT) that indirectly presuppose the concept of truth. As a

    result, (FT) cannot be an informative, noncircular analysis of truth.

    The underlying assumption behind this objection is that functional

    denitions like (FT) can be informative only if they are completely reduc-

    tive. But this is surely a fallacy. The functional method of denition sup-

    plies a way of giving the job descriptions for every property of the type in

    question all at once. But the fact that we can do so does not guarantee,

    either here or in the philosophy of mind, that we have entirely reduced

    the concepts in question to another set of concepts without remainder.

    And in the present case, our inability to reduce truth to more simple

    concepts is neither surprising nor avoidable. Arguably, truth is the most

    basic concept we possess; one cannot even get logic off the ground without

    some concept of truth. Prima facie, there is little reason to hope for any

    account of alethic concepts in terms that do not already presuppose them

    at some level of analysis or other. But to claim on this basis alone that a

    A Functionalist Theory of Truth 743

  • functional specication of truth is uninformative is to assume without

    argument that the only informative analyses are those that are completely

    reductive. Yet clearly one can informatively and usefully explicate a

    concept without completely reducing it to more basic concepts. If we did

    have to live up to such high standards, we would be able to understand

    very few of our concepts indeed.

    A more concessive answer to this objection is also possible for the

    functionalist. I have described alethic functionalism as (in part) a concep-

    tual project, one that, roughly speaking, attempts to specify the meaning

    of alethic terms. Yet as Jackson has recently emphasized (1998, 143; see

    also Chalmers 1996, 6164), there is more than one way to understand

    how a description of truth conditions like (FT) can ``specify meaning.''

    The more traditional way, which I have thus far implicitly assumed,

    takes (FT) as a priori and necessary. On this understanding, (FT) says

    that p's being true means that p has the higher-order property of having a

    property that plays the truth role for that discourse.14 On the other hand,

    we can take ``true'' to ``rigidly predicate'' a particular property in every

    possible world in which it predicates anything at all.15 If we do, we will

    take the right-hand side of (FT) as giving us the meaning of ``true'' in a

    very different sense, that is, as xing the reference of ``true.'' But since

    ``true'' is a rigid predicator and the description on the right-hand side of

    (FT) is not, this would be similar to treating ``is the man who taught

    Alexander'' as what xes the reference for ``Aristotle.'' And of course,

    Aristotle is only contingently the man who taught Alexander. On this

    alternative, (FT) is a priori but not necessary (Jackson 1998, 144). On

    this latter method, there is no barrier to employing other alethic concepts

    in our explanation of what ``true'' happens to ascribe to propositions.

    We take (FT) as xing the reference of ``true,'' not as ``dening'' it in the

    traditional sense. Nonetheless, there is a sense of ``meaning'' in which

    (FT) is telling us the meaning of our concept of truth.

    I see functionalism as engaged in the traditional project of meaning

    giving or conceptual explication without, as I've emphasized above, that

    project being necessarily reductive. But ofcially, the functionalist can

    remain neutral on the status of (FT). Taken either as a nonreductive

    conceptual explication or as xing the reference, (FT) reveals important,

    a priori facts about the concept of truth.

    744 Michael P. Lynch

  • 6 Conclusion

    A complete functionalist theory of truth has to meet at least three major

    demands. It must explain what it means to say that truth is a functional

    property; it must give a detailed account of how truth might be realized

    differently in distinct contexts; and it must explain the theory's impact on

    other important philosophical concepts, such as meaning, knowledge, or

    (what has come to be called) truth-aptness. In this paper I have been

    concerned with the rst demand, having touched only briey on the

    second. I have left the third aside altogether. Nonetheless, I think it is

    already clear that a complete functionalist theory would have signicant

    advantages.

    First, the functionalist theory of truth manages to satisfy both pluralist

    and monist intuitions. At one level, the functionalist theory is consistent

    with monism. In every discourse, the concept of truth is the concept of a

    particular higher-order propertythe property of having the property

    that plays the truth role for that discourse. But at the level of deep

    metaphysics, alethic functionalism allows that this role may be realized

    by distinct properties that depend on the discourse in question. So to

    have the property of truth is to have a property that can, by its very

    nature, be realized in multiple ways.

    Second, the functionalist theory explains the interrelatedness of the

    alethic concepts that has often made more traditional theorists about

    truth uncomfortable. A common reason for dismissing, e.g., ``fact talk''

    among many deationary-minded philosophers is that they are the mere

    shadows of propositions; that is, one cannot say what a fact is without

    invoking the notion of truth. But as just noted, this is what the func-

    tionalist would expect. In the philosophy of mind, it is common to think

    of belief and desire as so interrelated that the one cannot be explained

    without reference to the other. Yet this fact does nothing to show that

    such concepts, and the properties they are concepts of, are metaphysi-

    cally suspect.

    Third, the theory is ecumenical. Functionalism does not dictate in

    advance how truth will be realized in various discourses. To specify how

    truth is realized, we must look to the particulars of our thought. But once

    we do so, the functionalist theory may act as a neutral frame for a less

    A Functionalist Theory of Truth 745

  • reductive picture of how our forms of thought relate to each other and to

    the world around us.

    Notes

    Many people have helped me in writing this paper. The essential ideas wereoriginally expressed in papers read at the 1999 Bled Epistemology Conferenceand at the Catholic University of Lublin, Poland. For conversations and corre-spondence I also thank William Alston, Terry Berthelot, Paul Bloomeld, RobertBarnard, Marian David, Hartry Field, Charles Fletcher, Terry Horgan, FrankJackson, Peter Klein, Philip Pettit, Matjaz Potrc, Tadeusz Szubka, Robert West-moreland, Michael Williams, Cory Wright, Crispin Wright, and audiences at theabove places and at Connecticut College and the University of Southern Missis-sippi. Thanks also to the Master of St. Edmund's College, Cambridge, where,with the help of a grant from the University of Mississippi, I was able to completea rst draft of this paper.

    1. See Quine (chap. 20), Horwich (chap. 24), Field (chap. 21), and Grover (chap.22).

    2. I refer here to the causal-relation theory of correspondence, espoused by Field(chap. 16) and Devitt (1984, see also chap. 25).

    3. For objections of this sort to epistemic theories, see Lynch 1998, 107 ff., andAlston 1996. For a defense of the claim that epistemic theories work in somecontexts, see the appendix to Wright 1999.

    4. For a more detailed discussion of the problem of scope faced by correspon-dence, epistemic, and deationist theories of truth, see Lynch 2000.

    5. For further remarks on how my view differs from Horgan's and Wright's, seeLynch 1998, 129 ff.

    6. Wright has emphasized (e.g., 1995, 215) that he too thinks that the concept oftruth is univocal. In his contribution to this volume (chap. 32) Wright secures thisunivocality by distinguishing, much as I do below, the concept of truth and theunderlying properties that realize truth. Nonetheless, as I note, Wright stops shortof endorsing a functionalist account of alethic concepts.

    7. I rst began to suggest this idea in 1998, pp. 125 ff. Subsequently I learnedthat Philip Pettit had made the suggestion that truth could be understood func-tionally in his 1996 comments on Wright. Correspondence with Pettit conrmsthat our views on this matter are quite similar; I have also benetted from Jack-son's 1998 account of moral functionalism (although Jackson does not endorsethe functionalist theory of truth).

    8. For instance, in the philosophy of mind, one can understand functions com-putationally (Putnam), causally (Lewis, Armstrong), or biologically (Millikan,Dretske).

    9. The analogy is Bob Barnard's (correspondence).

    746 Michael P. Lynch

  • 10. As I see it, there is no reason to settle which is which in advance. That isexactly the sort of information we would hope our theory would reveal. Thus,the platitude that a proposition is true because it corresponds to some fact may ormay not mean that this correspondence with fact is the literal cause of the prop-osition's being true. The ordinary word ``because'' can also be used to mean ``invirtue of,'' and x can be the case in virtue of y without x being causally related toy. Which relation ``because'' ends up picking out may depend, for all we know inadvance, on the type of propositions and facts being considered.

    11. Of course, lawyers and judges often disagree about what the law is; that is,they disagree about what is or should be included within the body of law. Oftenour decisions about what lies within or without the body of law depends in parton what we might call metalegal propositions, or propositions concerning a law'sjust or unjust nature, its accordance with ``natural'' law, its usefulness for society,and so on. Although the distinction cannot be an absolutely precise one, metal-egal propositions are surely distinct from propositions of the law proper (or whatwe might call purely legal propositions). As such, we should expect that ouropinions about metalegal matters (our opinions about what to include within thebody of law) will be true or false in a different way than purely legal propositions,which the present proposal takes to be true in virtue of coherence with the bodyof law. See Walker's essay (chap. 6) for further discussion of what he calls alimited coherence theory of truth.

    12. One reason that it is reasonable to suppose this is because an endorsement ofthe equivalence schema, together with a similar schema for falsity and the law ofexcluded middle, entails that a proposition is either true or false.

    13. A related worry is this. According to recent ``two-dimensional'' semantics,terms can be understood as having both a primary and secondary intension(Chalmers 1995, 56 ff.; Jackson, 1998). A primary intension is a function fromworlds considered as actual to a property, while a secondary intension is a func-tion from worlds considered as counterfactual to a property. Now the primaryand secondary intensions of ``water'' pick out different properties at a givenworld. When Putnam's Twin Earth world is considered as actual, ``water'' picksout XYZ, but when it is considered as counterfactual, ``water'' picks out H2O.The worry is that any true functional term or concept must be similar to ``water''in this respect. That is, it must have a primary intension that picks out differentproperties from what its secondary intension picks out in some worlds. But ofcourse, ``true'' does not work like this. When I imagine that some other world isthe actual world, I don't think that another property is being ascribed by ``true'';indeed, the primary and secondary intension of the word ``true'' seem to be thesame.

    Like the problem above, this worry is unfounded. The key is to remember thaton the functionalist theory I've presented here, truth is identical to the role prop-erty. For any discourse, truth is the higher-order property of having a propertythat plays the truth role in that discourse. Thus, like the higher-order property ``iscolored like the sky at this world'' (where ``this'' refers to the world in question,

    A Functionalist Theory of Truth 747

  • whether or not it is the actual world), the truth property will remain constantacross worlds (discourses). But in both cases, it is obvious that different lower-level properties can realize the higher-order property. (I thank Philip Pettit forhelp with this point.)

    14. If we were to take (FT) as giving the primary intension of the concept of truth(see last note) we would have to say that (FT) is necessarily true in all possibleworlds considered as actual (Chalmers 1996, 63; also see Jackson 1997, chap. 2).

    15. I take this term from Sydney Shoemaker 1975, 400.

    References

    Alston, W. P. 1996. A Realist Conception of Truth. Ithaca: Cornell UniversityPress.

    Blackburn, S. 1998a. ``Wittgenstein, Wright, Rorty, and Minimalism.'' Mind107: 158182.

    Blackburn, S. 1998b. Ruling Passions. Oxford: Oxford University Press.

    Blackburn, S., and Simmons, K. 1999. Truth. Oxford: Oxford University Press.

    Chalmers, D. 1996. The Conscious Mind. Oxford: Oxford University Press.

    Davidson, D. 1990. ``The Structure and Content of Truth.'' Journal of Philoso-phy 87: 279328.

    Davidson, D. 1967. ``Truth and Meaning.'' Synthese 17: 304323.

    Devitt, M. 1984. Realism and Truth. Princeton: Princeton University Press.

    Dworkin, R. 1986. Law's Empire. Cambridge: Harvard University Press.

    Field, H. 1972. ``Tarski's Theory of Truth.'' Journal of Philosophy 69: 347375.

    Field, H. 1995. ``Disquotational Truth and Factually Defective Discourse.'' Phil-osophical Review 103, no. 3: 405452.

    Gupta, A. 1993. ``A Critique of Deationism.'' Chap. 23 of this volume.

    Horgan, T. 1986. ``Psychologism, Semantics, and Ontology.'' Nou s 20: 2131.

    Horgan, T. 1991. ``Metaphysical Realism and Psychologistic Semantics.''Erkenntnis 34: 297322.

    Horgan, T. 1996. ``The Perils of Epistemic Reductionism.'' Philosophy andPhenomenological Research 56, no. 4: 891897.

    Horwich, P. 1990. Truth. Cambridge: Basil Blackwell.

    Horwich, P. 1998. Meaning. Oxford: Oxford University Press.

    Jackson, F. 1998. From Metaphysics to Ethics: A Defense of Conceptual Analysis.Oxford: Oxford University Press.

    Kirkham, R. 1992. Theories of Truth: A Critical Introduction. Cambridge: MITPress.

    Lewis, D. 1970. ``How to Dene Theoretical Terms.'' Journal of Philosophy 67:427446.

    748 Michael P. Lynch

  • Lewis, D. 1972. ``Psychophysical and Theoretical Identications.'' AustralasianJournal of Philosophy 50, no. 3: 249258.

    Lynch, M. 1997a. ``Minimal Realism or Realistic Minimalism?'' PhilosophicalQuarterly 47, no. 189: 512518.

    Lynch, M. 1998. Truth in Context. Cambridge: MIT Press.

    Lynch, M. 2000. ``Alethic Pluralism and the Functionalist Theory of Truth.'' ActaAnalytica 15, no. 24: 195214.

    Pettit, Philip. 1996. ``Realism and Truth: A Comment on Crispin Wright's Truthand Objectivity.'' Philosophy and Phenomenological Research 56, no. 4: 883889.

    Putnam, H. 1981. Reason, Truth, and History. Cambridge: Cambridge Univer-sity Press.

    Putnam, H. 1994. ``Sense, Nonsense, and the Senses: An Inquiry Into the Powersof the Human Mind.'' Journal of Philosophy 91, no. 9: 488517.

    Sainsbury, M. 1996. ``Crispin Wright: Truth and Objectivity.'' Philosophy andPhenomenological Research 56, no. 4: 899904.

    Shoemaker, S. 1975. ``Functionalism and Qualia.'' Philosophical Studies 27:291315.

    Smith, Michael. 1994. The Moral Problem. Cambridge: Blackwell.

    Wittgenstein, L. 1969. On Certainty. Cambridge: Blackwell.

    Wright, C. 1992. Truth and Objectivity. Cambridge: Harvard University Press.

    Wright, C. 1995. ``Truth in Ethics.'' Ratio (new series) 8, no. 3: 210226.

    Wright, C. 1996. ``Response to Commentators.'' Philosophy and Phenomeno-logical Research 56, no. 4: 911941.

    Wright, C. 1998. ``Comrades against Quietism.'' Mind 107: 184203.

    Wright, C. 1999. ``Truth: A Traditional Debate Reviewed.'' In Blackburn andSimmons 1999.

    A Functionalist Theory of Truth 749

  • This excerpt from

    The Nature of Truth.Michael P. Lynch, editor. 2001 The MIT Press.

    is provided in screen-viewable form for personal use only by membersof MIT CogNet.

    Unauthorized use or dissemination of this information is expresslyforbidden.

    If you have any questions about this material, please [email protected].

    The Nature of Truth: Chap31 - A Functionalist Theory of Truthpage 1page 2page 3page 4page 5page 6page 7page 8page 9page 10page 11page 12page 13page 14page 15page 16page 17page 18page 19page 20page 21page 22page 23page 24page 25page 26page 27

    Copyright notice