Is Truth a Norm ?

Pascal Engel

To cite this version:

Pascal Engel. Is Truth a Norm ?. Peter Kotatko, Peter Pagin. Interpreting Davidson, CSLI, Stanford,
pp.37-51, 2001, 3. �ijn_00000152�

HAL Id: ijn_00000152

https://jeannicod.ccsd.cnrs.fr/ijn_00000152
Submitted on 23 Sep 2002

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
3
Is Truth a Norm?
PASCAL ENGEL

1 Introduction
A familiar theme—indeed a sort of slogan—in contemporary philosophy is
that meaning and mental content are ‘normative’ or have a normative di-
mension. One of its origins is in Wittgenstein’s rule-following considera-
tions: to understand the meaning of a word, or to possess a certain concept,
is to be able to appraise correct or incorrect uses of it. This feature is often
held to be the main obstacle to a naturalistic analysis of meaning and inten-
tional concepts. Davidson is equally skeptical about such an analysis, since
he has long held that the ‘constitutive ideal of rationality’ which governs
our concepts of propositional attitudes have no ‘echo’ in physical theory
and in naturalistic concepts in general. One of his main arguments for this
view is that ‘the concepts we use to explain and describe thought, speech
and action are irreducibly normative’ (Davidson 1999:460). In a previous
article (Engel 1999 1), I have tried to elucidate in what sense we can say

1 That paper was a descendent of the one I read at the Karlowy Vary conference on Inter-
preting Davidson in 1996. The present paper is an attempt to answer Davidson’s reply (Da-
vidson 1999) and to further the discussion that I have initiated in Engel 2000 and Engel forth-
coming. I have been much influenced, while writing this paper (especially the second sec-
tion), by Paul Boghossian’s article ‘The Normativity of Content’, read at the Summer School
on Normativity and reason in Parma (July 2000). Boghossian is, of course, in no way respon-
sible for the mistakes and misunderstandings of his views that might be present in this paper.

Interpreting Davidson
Petr Kotatko, Peter Pagin, and Gabriel Segal (eds.).
Copyright © 2001, CSLI Publications

37
38 / PASCAL ENGEL

that, for Davidson, mental concepts have a normative dimension, and I have
suggested a reading of this claim, which, I thought, he could at least partly
agree with: that the normative dimension in mental content resides in the
specific norms attached to concepts, along the lines of conceptual role ac-
counts and theories of concept possession, and in a general norm of truth
attached to the concept of belief, which is central among the propositional
attitudes. But Davidson disagrees. He sees little promise in conceptual role
accounts of concepts, and bluntly rejects the second suggestion: ‘Truth is
not, in my opinion, a norm’ (Davidson 1999:461). Since he has taught us
that what matters in interpretation (including the interpretation of philoso-
phers) is more understanding than agreement, I shall leave aside here the
first suggestion, and shall try to articulate better the second, in the hope of
furthering this dialogue with him. In the first part of this paper, I try to
spell out what is at stake in the claim that truth is a norm and why David-
son could have grounds to oppose this claim. In the second part, I try to
argue that there is a reasonable sense in which we can, and must, say that
truth is a norm of belief, and that most of our epistemic norms are
grounded in this one.

2
What does it mean to say that truth is a norm, and what is at stake in such
a claim? There are, basically, two strands in this debate, although they are
intimately connected. One strand is with the analysis of the concept of truth
and of the meaning of the word ‘true’, which opposes, in contemporary
philosophy, two camps. A number of philosophers hold that truth is a ‘ro-
bust’ property, to be analyzed in terms of such ‘substantive’ notions such as
correspondence, coherence, or perhaps along pragmatist lines. A number of
other philosophers hold that truth is a ‘thin’ property, which does not con-
tain more than what is expressed by the disquotational feature of the truth
predicate or by the trivial equivalence ‘It is true that P iff P’, and propose
‘deflationary’ or ‘minimalist’ conceptions.2 For a philosopher of the second
persuasion, any attempt to read into the concept of truth more than these
trivial or formal features, would be an attempt to ‘pump more content in the
concept of truth’ than a minimalist theory should allow (David-
son 1996:310). Thus, when Crispin Wright (1992), following to a large
extent Dummett (1959), claims that a deflationary theory of truth should be
‘inflated’ in order to accommodate the fact that truth is ‘a norm of our asser-
toric practice’ and of belief, this claim is rejected by one of the main con-

2 There are, of course, a number of different versions of ‘minimalism’ about truth. But for
the purposes of the present discussion, I shall largely ignore these differences.
 IS TRUTH A NORM? / 39

temporary proponents of the deflationary theory, Paul Horwich (1998). 3 The

second strand concerns the question whether truth itself is a norm, or a
value, which should be pursued for its own sake: is truth a goal or an ideal
of our inquiries? Is there any sort of ethical command to search after the
truth? Some pragmatist philosophers, such as Peirce, believe that there is;
other pragmatists, such as Rorty (1986, 1995) or Stich (1990), disagree.
They do not see why we would be aiming at truth, more than utility or
pragmatic value. And the connection between the two strands seems to be
this: in claiming that truth is a norm or an ideal, we reintroduce the idea
that truth is a grand or robust notion, which, according to minimalist theo-
ries, it is not.
Where does Davidson stand in these debates? With respect to the issue
of robustness vs. minimalism about truth, it seems quite obvious that he
sides with the second camp, although he does not accept Horwich’s version
of deflationism, and does not hope to give any general definition of truth
along this or other lines (Davidson 1996). So it seems that his denial that
truth could be a norm is based on his suspicion that any attempt to read
more into the concept of truth than the familiar formal features would be an
attempt to define or to explain further this fundamental and essentially ‘in-
definable’ concept. With respect to the second issue, he is equally clear: ‘I
do not think it adds anything to say that truth is a goal, of science or any-
thing else. We do not aim at truth, but at honest justification’ (David-
son 1999:461). His reason for this claim is this: ‘There must be cases of
fully justified beliefs where the fact that they are false will never come to be
known. I cannot see that cases of the last kind are more desirable than cases
of the first kind’ (ibid.). The point, presumably, is this: if truth were a goal
of inquiry or of science, then in some sense truth should be essentially
knowable and justifiable, the concept of truth would coincide with the con-
cept of justification and truth would have to be an epistemic concept. But
this implies a form of anti-realism about truth, and an attempt at ‘humaniz-
ing truth by making it basically epistemic’ (Davidson 1990:298). On the
other hand, the goal of belief is justification, but there is no point in link-
ing justification intrinsically to truth, as the claim that truth can be a norm
of belief and knowledge seems to imply. So theories of truth, which like

3 Other writers who have advocated the idea that truth is a norm include Hornsby (1997 and
1998). She defends this claim in the context of her own version of the ‘identity theory of
truth’: ‘The conception of truth which the identity theory brings with it allows truth to be a sui
generis norm, in play where there are rational beings who may go right or wrong in their
thought and speech’ (Hornsby 1997:22). I have myself defended the view that truth is a norm
in Engel 1991, in the context of the philosophy of logic, and along more general lines in
Engel 1998.
40 / PASCAL ENGEL

Peirce’s, or Putnam’s, tie truth to rational acceptability at the ‘limit’ of

scientific inquiry and which make truth a goal of inquiry, are equally guilty
of epistemologizing truth.
Here, Davidson’s position seems to be quite close to Rorty’s, who, in a
paper targeted at Wright’s claim that truth is a norm of our assertoric prac-
tice (Rorty 1995), has also denied that truth could be a goal of inquiry, on
behalf of a ‘Davidsonian’ conception of truth. In spite of Rorty’s tendency
to try to read more (and more Rortian theses) into Davidson’s views than
Davidson himself would allow, it is useful to briefly go over Rorty’s ac-
count of these issues.
Crispin Wright (1992) basically agrees with minimalism about truth
that truth is a ‘metaphysically lightweight’ notion, which contains not
much more that the usual platitudes attached to it (namely that ‘P’ is true
iff P, that our statements are true if things are such as they say they are, or
that every true statement has a negation).4 But he denies, contrary to what
is claimed by pure deflationism, that the whole meaning of the word ‘true’
is exhausted by the ‘Disquotation Schema’:
(DS) ‘P’ is true if an only if P
Wright claims that deflationism is subject to a characteristic tension, since
the acceptance that (DS) says all there is to say about truth, associated to
the acceptance of the platitude about negation:
(Neg) Every statement P, has a negation, not P
entails inconsistent claims about the relations between truth and assertibil-
ity. Both are, according to Wright, norms of assertoric discourse, in the
sense that in making an assertion, we aim at truth, and aim at making asser-
tions that are warranted or justified. So (DS), according to Wright, says
more than simply the fact that ‘P’ and ‘it is true that P’ are equivalent. It
also says that if we have a reason to say that P is true we have thereby a
reason to assert or to accept it. DS entails that ‘true’ and ‘is warranted as-
sertible’ or ‘justified’ coincide in ‘normative force’: to think that P is true is
to think it warrantedly assertible, and conversely. In this sense deflationism
is committed to granting the ‘normativity’ of the truth predicate. Now if
deflationism holds that DS exhausts what there is to say about truth, it
should also deny that truth and warranted assertibility are distinct norms of
assertion. But these are distinct norms, since the predicate ‘is true’ and ‘is
warrantedly assertible’ diverge in their extension. For it follows from (DS)
and (Neg) that :

4 Wright calls his own view ‘minimalism’, but he intends to dissociate it from other varieties
of minimalism, in particular from Horwich’s (1990) deflationary conception.
 IS TRUTH A NORM? / 41

(1) ‘It is not the case that P’ is true iff it is not the case that P
and contraposing on DS :
(2) It is not the case that P iff it is not the case that ‘P’ is true
and by the transitivity of the biconditional:
(3) ‘It is not the case that P’ is true iff it is not the case that ‘P’ is true
But replacing ‘true’ by ‘is warrantedly assertible’ in (3) is incorrect: for if P
nor the negation of P are warrantedly assertible, it may be that neither ‘P’
nor ‘It is not the case that P’ are warrantedly assertible.
It should be noted that Wright’s argument, to the effect that truth is a
distinctive norm of assertion in no way implies the coincidence of truth and
warranted assertibility or justification. On the contrary, the argument trades
on the difference between a statement being true and our being warranted in
asserting it. So in Wright’s hands, the thesis that truth is a distinctive norm
of assertoric practice does not amount to making truth an essentially epis-
temic concept.
Now, of course, Wright’s argument rests upon the claim that truth is a
norm of our assertoric practice, or at least a norm of belief, and someone
who denies the existence of such a norm would immediately reject it as
question begging. But is this claim so controversial? It seems to amount to
the familiar idea that the concepts of belief, of assertion, and of truth are
intrinsically interconnected: to assert that P implies (or at least implicitates)
that one believes that P, and to believe that P is to believe that P is true
(hence the oddity of utterances such as those which give rise to Moore’s
paradox: ‘P, but I believe that not P’). Now, why should Davidson deny
this obvious point? Has he not repeatedly claimed that the concept of belief
and the concept of truth are intimately connected, that the basic attitude
which lies behind belief and which is evidence for it is the ‘holding-true’ of
sentences? And has he not emphasized that the indispensability of the Prin-
ciple of Charity amounts to the claim that ‘belief is by nature veridical’?
But Rorty urges us not to interpret this basic conceptual link between
belief and truth as implying that there is a distinctive norm of truth:
To say, as Davidson does, that ‘belief is in its nature veridical’ is not
to celebrate the happy congruence of subject and object but rather to
say that the pattern truth makes is the pattern which justification to us
makes (Rorty 1995:286).
This, according to Rorty, is not alien to Davidson’s emphasis on the ‘nor-
mative character’ of meaning and intentionality. But this character does not
amount to the recognition of the distinctive normative nature of truth itself:
The pattern that truth makes is, in fact, indistinguishable from the pat-
tern that justification to us makes—so it might be best to say simply
42 / PASCAL ENGEL
that ‘most beliefs held by anybody are justifiable to us’ rather than
‘most beliefs held by anybody are true’. […] The former expression
seems to me the clearest way to exhibit the force of Davidson’s claim
… that the guiding principles used in detecting this pattern ‘derive
from normative considerations’ and to bring out the importance of his
reference… to ‘the norms that govern our theories of intentional attri-
bution’. The need to justify our beliefs and desires to ourselves and to
our fellow-agents subjects us to norms, and obedience to these norms
produces a behavioural pattern which we must detect in others before
confidently attributing any beliefs to them. But there seems no occa-
sion to look for obedience to an additional norm, the commendment
to seek the truth. (Rorty 1995:287).
In a sense, Rorty does not deny that there is, for our beliefs, a norm, in the
sense that any interpretation of a being as having beliefs derives from cer-
tain ‘guiding principles’: precisely those that Davidson calls ‘norms of ra-
tionality’ governing our attributions of beliefs and other propositional atti-
tudes to agents. Truth, in the form of the principle of charity as a principle
of veridicality (the other half of the principle being a principle of coherence)
may well be such a ‘norm’. But there is no point in taking this ‘norm’ to
be different from the norm according to which our beliefs have to be justi-
fied. But ‘justified’ here does not mean objectively justified, for there are as
many ways of justifying our beliefs and our actions as there are human in-
terests, and there is no way in which these interests could be subsumed
under a single interest or goal. This would come up to reintroducing into
truth the very metaphysical weight that deflationism or ‘quietism’ about
this notion have gotten rid of. This is why, according to Rorty, talking of a
‘norm of truth’ as distinctive, and attempting to read it into our very con-
cept of belief, is highly misleading. On his view, pragmatism, in the form
advocated by Dewey and by himself, does recognize this fact, as does (ac-
cording to Rorty) Davidson’s own version of ‘pragmatism’:
If Dewey and Davidson are asked ‘What is the goal of inquiry?’ the
best either could do would be to say that it has many different goals,
none of which has any metaphysical presupposition: for example, get-
ting what we want, the improvement of man’s estate, convincing as
many audiences as possible, solving as many problems as possible,
and so on. Only if we concede to Wright that ‘truth’ is a name of a dis-
tinct norm will metaphysical activism seem desirable. For Dewey and
Davidson, that is an excellent reason not to view it as a
norm (Rorty 1995:298-299).
It seems to me that Rorty gets Davidson right on the negative side, and that
his reasoning is quite close to the one which underlies the latter’s refusal to
consider truth as a distinctive norm of belief. On the positive side, how-
ever, it is less obvious to me that Davidson would subscribe to the so-
called ‘pragmatist’ idea that inquiry has as many different goals as one
 IS TRUTH A NORM? / 43

might want to have, and that there is some essential relativity in these
goals. For he says, as we have seen, that ‘we aim at honest justification’,
and I do not see how this could imply that there are as many justifications
that it might please to us, as ‘Dewey-and-Davidson’’s view seem to imply.
I other words I do not see why Davidson would agree with what ‘pragma-
tism’, according to Rorty, implies, i.e. that justification cannot be objec-
tive, but is purely subjective or context–relative.
Be it as it may, is the negative part of this reasoning correct, and are
Rorty and, for that matter, Davidson, right in denying that there is any in-
teresting sense in which we can call truth a norm? In order to see whether it
is the case, we should try to spell out a little more what this talk of ‘norm’
and ‘normativity’ amounts to.

3
The above description of the debate about the normativity of truth suffers,
just as do the debates about the normativity of meaning and mental content,
from serious ambiguities. Indeed both claims are far from clear.5 When we
talk about the ‘normative’ character of truth are we saying that its normative
property or (properties) exhaust the concept, or are essential to it, or are we
merely saying that truth has, among other (non-normative) properties, some
normative properties as well? And if there are any such properties, do we
want to say that they enter this concept directly or that they enter merely
indirectly? If truth is ‘normative’ what kind of norm is it, and in what sense
is this norm attached, or implicated, by the very concept of truth? There are
plenty of norms: ethical, legal, social, culinary, architectural, and so forth,
relative to a whole range of human activities. Truth is presumably a norm
of the cognitive kind, and not a norm pertaining to certain actions. But then
in what sense is it normative?
We have encountered above two senses of ‘norm’ which may be relevant
for answering these questions: (a) truth is a norm of belief and of assertion,
in the sense that it is constitutive of belief and assertion that ‘belief aim at
truth’ and that asserting something is asserting something that one takes to
be true; (b) truth is a norm of belief, in the sense that it is the goal of in-
quiry or of our epistemic enterprises. The celebrated phrase ‘belief aims at
truth’ seems to imply both senses, and in fact to conflate them, and since it
has been used by Dummett (1959) and by Williams (1971), it has been the
source of many confusions. On the one hand, saying that belief ‘aims at
truth’ seems to say something about what belief is, namely that the concept

5 I have tried to spell out further what these claims mean in the case of meaning and mental
content in Engel 2000 and forthcoming.
44 / PASCAL ENGEL

of belief has to be defined, or at least understood, through the concept of

truth, and this points in the direction of the sense (a) of the notion of norm.
On the other hand, saying that belief ‘aims at truth’ seems to indicate what
belief is for, and to suggest that aiming at belief is a conscious, or uncon-
scious goal of all believers, or that it is something desirable or a value.
Rorty’s discussion of Wright on this point explicitly moves from the first
to the second, when he claims that taking truth to be constitutive of asser-
toric practice implies that there is some ‘duty’ to attain the truth
(Rorty 1995:288). But certainly there is no such direct implication, and the
two senses are distinct. For instance the fact that ‘being an unmarried per-
son’ is constitutive of the concept of bachelor does not imply that there is
any special value in being an unmarried person, or some duty attached to
having this status.
Still, it is often claimed that there is some element of normative ap-
praisal, or some action-guiding implication, in the concept of truth itself.
For it is often said that one of the ordinary uses of the word ‘true’ is an
endorsing use: saying that something is true, or adding to a given asser-
tion: ‘And that’s true’, seems to imply that the speaker praises his asser-
tion, and enjoins his hearer to believe it or to take it as worth asserting.
This idea often underlies what is called the ‘performative’ theory of truth:
saying ‘It is true that …’ is the performance of a sort of illocutionay or per-
locutionary act. This, indeed, is but one version of the deflationary concep-
tion. 6 But it is not clear that this meaning of the word ‘true’ exhausts all
the possible uses of it, and it is even less clear that this kind of use implies
that the speaker invokes some sort of duty, on the part of his hearer to be-
lieve or to assert what he claims is true. Certainly there are uses of ‘true’
which are, at least prima facie, purely descriptive, and which carry no sort
of commitment. And there are truths that we do not even dream of asserting
or of believing, and hence for which we do not feel any sort of obligation to
believe or to assert them. Hence saying that ‘truth is normative’ needn’t
commit us to some performative conception of truth.
In order to get a clearer grip on the normative implications of the con-
cept of truth, let us try first to spell out different senses of the word ‘norm’
in ordinary parlance. A number of theorists who have dealt with the mean-
ing of ‘norm’ in moral philosophy and in law have long remarked that there

6 The view is often attributed to Strawson (1950), but it lies also behind Rorty’s own version
of deflationism: ‘Truth is but a little tap that we do on the shoulders of the beliefs that we
like’. Strawson himself has later rejected his previous view. See his note at the end of the
reprinting of his 1950 paper in Blackburn and Simmons eds. 1999.
 IS TRUTH A NORM? / 45

are, basically two main kinds of normative vocabulary or of normative con-

cepts:7
(a) concepts of the ‘deontic’ kind or normative concepts proper such as:
ought, obligation, norm, requirement, permission, regulate, correct,
rectified, etc. Such concepts are action-guiding, in the sense that
they are such that they imply that an appropriate response to what
they prescribe, permit, or prohibit is a voluntary or intentional ac-
tion: ought implies can;
(b) concepts of the evaluative kind such as : good, valuable, desirable,
worth, etc. The appropriate responses to judgments involving such
concepts are not actions performed by the agents, but feelings or
psychological attitudes.
This is a very rough characterization, but it is enough for our purposes here.
We can ask the question: if truth is a norm of belief or of assertion, and if
saying or believing something true involves something ‘normative’, does it
involve a norm of the first kind or of the second kind, and if so in what
sense does it involve it?
Let us suppose first that it involves a norm in the first, deontic sense.
The idea would then be that in ascribing truth to certain statements or be-
liefs, we are in some sense speaking ‘oughts’, or implicitly relying on cer-
tain imperatives about what one ought to believe or assert. The plausible
sense in which we could cash out imperatives, prescriptions, or permissions
from the application of the predicate ‘true’ to a sentence or a belief would
presumably be by issuing conditionals of the following form:
(1) If it is true that P, then one ought to believe (assert) that P.
For instance:
(1') If it is true that snow is white, then one ought to believe (assert) that
snow is white.
And if truth is a norm of belief, or of assertion, such conditionals must be
issued not simply for some particular truths, but also for any particular
truth, including those that we do not even dream of believing or of assert-
ing, either because they are so trivial that they do not merit consideration
(for instance the truth that there are presently 10 264 blades of grass on the
piece of lawn before me), or because we could never figure out any one of
them. But this is absurd, for nobody is in the position to believe every
truth whatsoever. So general conditionals of the form:

7 I am indebted here to a talk given in Paris in March 2000 by Peter Railton. Since then, I
have heard from Kevin Mulligan that Brentano had a similar kind of classification.
46 / PASCAL ENGEL

(2) For any P, if it is true that P, then one ought to believe (assert)
that P
are absurd. Certainly we are under no such obligation. Given that ‘ought’
implies ‘can’, and that there is no possible way in which we can perform
this obligation (or even the corresponding permission), the claim that truth
could be a norm in the deontic sense (a) is absurd. There is a further reason,
if our characterization of norms of type (a) is correct, why truth could not be
a norm in this sense. It is that we are supposed to conform ourselves to
such norms through actions. It may be that believing that P, or coming to
believe it, in the sense of accepting it, or judging it, is a form of action, as
is asserting it. But belief per se, the state of believing that P, is not volun-
tary, and it cannot be an action (in the sense of Williams 1971, but more on
this below). Hence belief could not be subject to an ‘ought’ of the deontic
kind.
Could it be, then, that the norm involved in truth ascriptions is a norm
of the second kind? The idea would then be that the truth of a particular
belief or assertion would imply that it is in some sense valuable, or that it
merits appraisal, or that it provokes in us some positive feeling. This time,
the response to the norm would not be a voluntary action, and it would be
plausible to say that it could be an involuntary state such as a belief. But
the idea is equally hopeless. For consider:
(3) If it is true that snow is white, then it is valuable to believe that
snow is white
or possibly:
(4) If it is true that snow is white, then one appraises, or feels well
about believing that snow is white
and by generalizing:
(5) For any P, if P is true, then it is valuable to believe that P.
This is absurd for the same reason as above, since even on the reading of
‘norm’ as ‘value’, it makes no sense to say that we value all true beliefs,
including the most trivial. So it makes no sense to say that the norm of
truth consists in a form of intrinsic value of truth.
So, on the supposition that our two senses of ‘norm’ exhaust the possi-
ble senses of this notion, the claim that truth is a norm of belief is obvi-
ously false. It may be what Davidson and Rorty are aiming at when they
declare that truth is not a norm, and not a goal of belief or of inquiry. For
what the remarks above suggest is that it is not simply truth in general
which interests us, but, so to say, interesting or relevant truth. And, to
pursue this line, what kinds of truth are interesting depends upon our ex-
 IS TRUTH A NORM? / 47

planatory interests, which may vary from context to context. In some con-
texts, certain truths are interesting or valuable, and should be attended to,
and others are not, and are not worthy of consideration (for instance a
climber will be interested in the truth that the rock on which he is leaning
is friable, but a tourist who contemplates the cliff from below has no such
interest). We are not interested in truth as such, but in knowledge, in so far
as knowledge is relative to our human interests (just as we are not interested
in believing any truth whatsoever, we are not interested in knowing any-
thing whatsoever). And in so far as knowledge is justified true belief (what-
ever that means), what we are interested in is, as Davidson says, ‘honest
justification’.
On this much, then, and in so far as the notion of ‘norm’ is cashed out
in such conditionals as (2) or (5), I quite agree with Davidson and Rorty
that truth is not a norm, in any interesting or important sense. But is this
the sense which we want to put forward when we say that truth is a consti-
tutive norm of belief, or a norm of assertoric practice? Certainly it is false
to say that ‘belief aims at truth’ if truth is conceived as the goal, conscious
or not, of belief, or as a form of obligation to believe everything that is
true. But this is not what one ordinarily means when one says that truth is
the norm of belief. What one means, rather, is that the concept of truth is
constitutive of the concept of belief, in the following sense: if one has good
reasons (is justified) to believe that P, then one has thereby good reasons (is
justified) to believe that P is true. The concept of truth goes with the con-
cept of good reason, or of justification for belief. Reason or justification for
belief is reason or justification for truth. Certainly saying that a belief is
justified does not entail that it is true, and justification is not truth, but
saying that a belief is justified is a prima facie reason for thinking that it is
true.
Now the concepts of ‘reason’, of ‘good reason’, or of ‘justification’ are
certainly normative concepts, if there are any. They are relevant to whether
it is correct to believe something. So we can try to cash it out in terms of
conditionals of the preceding form. But (to use only deontic concepts of the
first type) the correct form of the conditional should not be (2) or (5) but
the following :
(6) For any P, one ought to believe that P only if P.
And this makes perfectly respectable sense. The claim is not, like (5), that
if something is true then one ought to believe it, but that one ought to be-
lieve only what is true. This does not have the implication that we should
believe everything that is true. This imperative, in a sense, is quite obvi-
ous, for it amounts to saying that claims to belief are claims to true beliefs.
It is in this sense that truth is the fundamental norm of belief, since some-
48 / PASCAL ENGEL

one who would not recognize the truth of this imperative would not under-
stand the very concept of belief.
But here again, one may object that it is not right. May it not be the
case that someone can find it valuable, desirable, or even pressing or obliga-
tory, to believe something that is false and that he recognizes as such? After
all, cases of self-deception, as they are analyzed by Davidson (1985) are
precisely cases where an agent believes that P is false but nevertheless be-
lieves that P, as a result of his desiring or wanting to believe that P. And
the self-deceived person seems to reach the latter belief on the basis of some
action that he performs. We could add, in the vein of James’ Will to Believe
that there are cases where it may be desirable to believe things that one be-
lieves to be false, if it is otherwise desirable, or useful to believe such
things, in spite of all the evidence that one has against their truth, or in
spite of having insufficient evidence for believing them. And a ‘pragmatist’
of the Rortyan stripe would certainly press this point.
Such cases, or such prima facie exceptions, might well be exceptions to
the claim that truth is the goal of belief, or that, in this (intentional) sense
belief ‘aims at truth’. If we agree with James that truth in itself may some-
times not be valuable, or that the epistemic value of truth may be overrid-
den by other personal or subjective values that an agent happens to have, or
that epistemic reasons alone may not be sufficient for belief, or that there
may be advantages to willful belief, then certainly one will not agree that
truth is a norm of belief in the sense of an exceptionless goal. So I grant to
Rorty, and to Davidson – if such considerations underlie his claim – that in
this sense truth is not uncontroversially a norm of belief or of assertion.
But once again, this is not the sense in which I want to claim that truth
is a norm. The fact that truth is a constitutive norm of belief, in the sense
that I intend to promote, is in no way refuted by the fact that we may some-
times (and possibly justifiably so) want to believe what is not true, and
succeed in doing so. The point is familiar from discussions about the pos-
sibility of believing ‘at will’.8 Suppose that someone wants to believe that,
say, the Dalai Lama is a Living God, because he has a special interest in
believing this (he praises the value of religious belief, or he is paid a large
sum of money for believing this), and moreover that he succeeds in acquir-
ing this belief (admittedly not immediately, ‘just like that’, but by some
indirect means, such as a drug or self-indoctrination). But he could not at
the same time want to believe that the Dalai Lama is a Living God and
succeed in performing an action which would result in his believing this,
for the very fact that he wants to believe this shows that he does not believe

8 I have reviewed these arguments, which stem from Williams 1971, in Engel 1999a.
 IS TRUTH A NORM? / 49

it in the first place. Hence, if he succeeds, he would be in a state both of

believing that the Dalai Lama is not a living God and (as a result of his
success in realizing what he wants) of believing that the Dalai Lama is a
Living God. The situation is reminiscent of Moore’s paradox and of the
case of self-deception (although I am not saying that wishful thinking and
self deception are the same). Whether or not someone can succeed in being
in such states, and whatever is their proper etiology, the reason why they
are paradoxical and the reason why we hesitate to attribute to the agent both
the belief that P and the belief that not P is that when someone has a belief
that P, he thereby has the belief that P is true. If he comes to believe (con-
sciously, at the same time) that his belief that P is false, then either he does
not have either one belief, or he is not really, with respect to these contents,
in a state of belief. So even someone who, for any reason, is not moved by
an interest for truth, or who rejects the idea that it can be a goal for his be-
liefs, has to recognize that truth is what his beliefs are aiming at, in virtue
of their being beliefs. Otherwise he does not understand the very concept of
belief, or us, the ascribers, are at a loss in attributing to him such a state.
Deciding to believe what is false is to undermine the very possibility of
believing.
This is the sense in which we are entitled to say that truth is a norm of
belief, and that ‘beliefs aim at truth’. Although it can be formulated in
terms of ‘oughts’ like in (6) this norm is not a norm in the deontic or in the
evaluative sense: rather it is definitional of the state of belief and of our
concept of this state. The same point is often put by saying that beliefs
have a certain ‘direction of fit’ (‘mind to world’) and that desires have an
opposite direction of fit (‘world to mind’). This is why there is no corre-
sponding norm for desires: having a desire carries no implication that the
desire should be satisfied. I am claiming that it is what grounds the truth of
conditionals such as (6). And the important point is that this norm is in
place even when one does not take truth to be a goal of our inquiries, or of
science, or whatever. It is essential to what we understand by the notion of
belief. It is a conceptual norm.
Does the recognition of this norm entail, as Rorty thinks (1995:298),
that truth itself is the ‘fixed’ goal of inquiry, and that we should subscribe
to the metaphysical (realist) picture which underlies this view? No, for it is
one thing to say that our concept of belief is such that they purport to be
true, and quite another thing to say that there really is something in which
their truth consists, a sort of real essence underlying all our belief claims.9

9 Cf. Wright 1996:914: ‘It is one thing for an expression to be used in the making of a dis-
tinctive kind of normative judgment; quite another matter for there to be such a thing as a
bearer really deserving a judgment of that kind’.
50 / PASCAL ENGEL

In this respect I do not see why the notion of a norm of truth, in the sense
defended here, would be inconsistent with a minimalist conception of truth.
For a minimalist conception says that we cannot define truth through some
explanatory or essential feature over and above the platitudes which charac-
terize this concept. And the idea that truth is a norm of belief is such a
platitude, although it is an important one.
The upshot of this is that there are, with respect to truth, two quite dif-
ferent notions of norm: a conceptual norm, which goes with the notion of
belief, and perhaps with the concept of assertion,10 and a cognitive norm,
which has to do not with what belief is, but with the value or with the re-
quirements of belief. I have claimed that there is no reason to think that
admission of the first norm automatically entails admission of the second,
in the sense in which truth would be the general goal of inquiry, and vice
versa. But the fact that these notions are different should not lead us to
think that they are unconnected, and that what I have called the conceptual
norm of truth does not underlie the cognitive norm of truth. For suppose
that one says, like Davidson, that what we aim at is not truth, but justifica-
tion, whatever way we are to analyze justification (as evidential support,
probabilistic nor not, or as coherence, or reliability, and so forth). If this is
so, we ought to believe that P only if P is justified, or is believable to a
high degree, etc. But I do not see how our grounds for taking P to be justi-
fied could not be also grounds for taking P to be true, even if it turns out
that P is not true. Justification and truth are not the same thing, but it is
difficult to see how aiming at the first would not be aiming at the second.
More, of course, is needed in order to establish that truth is the main, or the
only, cognitive value, and that a form of pragmatism which, like Rorty’s,
flatly denies this, is false. But this form of pragmatism seems to me to be
self-defeating when it denies that truth is a norm of belief.

10 I say ‘perhaps’ because I do not want to commit myself fully to Wright’s view that truth is
a norm of ‘assertoric practice’, and to the linguistic expression of belief, but only to the cor-
responding claim about belief and thought. It is not clear that there is a norm of asserting
what is true in the same sense as the one in which there is a norm of believing what is true.
Lying does not undermine the possibility of assertion as deciding to believe what is false
undermines the possibility of belief. (I am here again indebted to Paul Boghossian).
 IS TRUTH A NORM? / 51

References
Blackburn, S. and Simmons, K. eds. 1999. Truth. Oxford: Oxford University
Press.
Boghossian, P. 2000. The Normativity of Content. Paper read at the Summer
School on Normativity and reason, Parma, July 2000, unpublished.
Davidson, D. 1985. Deception and Division. Actions and events: Perspectives
on the Philosophy of Donald Davidson, eds. E. Lepore and B. McLaughlin,
138-48. Oxford: Blackwell.
Davidson, D. 1990. The Structure and Content of Truth. Journal of Philosophy
87:279-328.
Davidson, D. 1996. The Folly of Trying to Define Truth. Dialogue and Univer-
salism 6:39-53. Reprinted in Journal of Philosophy 94:263-78. Reprinted
in Blackburn, S. and Simmons, K. eds. 1999, 308-22. Page references to the
second reprint.
Davidson, D. 1999. Reply to Pascal Engel. The Philosophy of Donald Davidson
ed. L. Hahn. La Salle: Open Court 1999:460-61.
Dummett, M. 1959. Truth. Proceedings of the Aristotelian Society, LIX, 141-62.
Reprinted in Truth and Other Enigmas, 1978m 1-24. London: Duckworth.
Engel, P. 1991. The Norm of Truth. Hemel Hemstead: Harvester Wheatsheaf and
University of Toronto Press.
Engel, P. 1999a. Volitionism and Voluntarism about Belief. Belief, Cognition
and the Will, ed. A. Meijers, 9-25. Tilburg: Tilburg University Press.
Engel, P. 1999b. The Norms of the Mental. The Philosophy of Donald Davidson.
L. Hahn, ed, 447-59. La Salle: Open Court.
Engel, P. 2000. Wherein lies the Normative Dimension in Mental Content? Phi-
losophical Studies 100:305-21.
Engels, P. Forthcoming. Intentionality, Normativity and Community.
Hornsby, J. 1997. Truth: the identity theory. Proceedings of the Aristotelian
Society XCVII:1-24.
Horwich, P. 1990. Truth. Oxford: Oxford University Press.
Horwich, P. 1998. Meaning. Oxford: Oxford University Press.
Rorty, R. 1986. Pragmatism, Davidson and Truth. Truth and Interpretation, ed.
E. LePore, 333-55. Oxford: Blackwell.
Rorty, R. 1995. Is Truth a Goal of Inquiry? Davidson vs. Wright. Philosophical
Quarterly 45:281-300.
Stich, S. 1990. The Fragmentation of Reason. Cambridge, Mass.: MIT Press.
Williams, B. 1973. Deciding to Believe. Problems of the Self, 136-51. Cam-
bridge: Cambridge University Press.
Wright, C. 1992. Truth and Objectivity. Cambridge, Mass.: Harvard University
Press.