See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/227146601

Moral Relativism

Article  in  Philosophical Studies · September 2007

DOI: 10.1007/s11098-007-9083-2

CITATIONS READS

6 332

1 author:

Torbjörn Tännsjö
Stockholm University
164 PUBLICATIONS   563 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Priority Setting in Health Care View project

All content following this page was uploaded by Torbjörn Tännsjö on 12 March 2014.

The user has requested enhancement of the downloaded file.

Philosophical Studies (2007) 135:123–143 Ó Springer 2007
DOI 10.1007/s11098-007-9083-2

TORBJÖRN TÄNNSJÖ

MORAL RELATIVISM1

ABSTRACT. Moral relativism comes in many varieties. One is a moral

doctrine, according to which we ought to respect other cultures, and allow
them to solve moral problems as they see fit. I will say nothing about this
kind of moral relativism in the present context. Another kind of moral
relativism is semantic moral relativism, according to which, when we pass
moral judgements, we make an implicit reference to some system of morality
(our own). According to this kind of moral relativism, when I say that a
certain action is right, my statement is elliptic. What I am really saying is
that, according to the system of morality in my culture, this action is right.
I will reject this kind of relativism. According to yet another kind of moral
relativism, which we may call epistemic, it is possible that, when one person
(belonging to one culture) makes a certain moral judgement, such as that
this action is right, and another person (belong to another culture) makes
the judgement that the very same action is wrong, they may have just as
good reasons for their respective judgements; it is even possible that, were
they fully informed about all the facts, equally imaginative, and so forth,
they would still hold on to their respective (conflicting) judgements. They
are each fully justified in their belief in conflicting judgements. I will com-
ment on this form of moral relativism in passing. Finally, however, there is a
kind of moral relativism we could call ontological, according to which, when
two persons pass conflicting moral verdicts on a certain action, they may
both be right. The explanation is that they make their judgements from the
perspective of different, socially constructed, moral universes. So while it is
true in the first person’s moral universe that a certain action is right, it is true
in the second person’s moral universe that the very same action is wrong.
I explain and defend this version of ontological moral relativism.

1. INTRODUCTION

Moral relativism comes in many varieties. One is a sub-

stantial moral doctrine, according to which we ought to re-
spect other cultures, and allow them to solve moral problems
as they see fit. I will say nothing about this kind of moral
relativism in the present context.
124 TORBJÖRN TÄNNSJÖ

Other kinds of relativism are metaethical doctrines. Accord-

ing to these doctrines, there are more than one way correctly
to answer a moral question. At least this is how I am going
to use the term ‘relativism’ in the present context. This means
that I do not count expressivism and emotivism and prescrip-
tivism as relativist doctrines. According to expressivism, emo-
tivism, or prescriptivism, there is no moral truth. According
to these doctrines, there exist no moral facts. I prefer to clas-
sify these doctrines as nihilist. Relativism is the doctrine that
there exists more than one truth about some moral cases.
Certainly, an expressivist or emotivist can say things like:
‘‘It is true that one ought always keep one’s promises’’.2 On
expressivism or emotivism, however, this is just another way
of expressing a certain emotion. On expressivism, emotivism,
or prescriptivism, there are no moral facts (no true moral
propositions). On these doctrines, there are no moral proposi-
tions whatever. Moral opinions are feelings, or resolutions,
and so forth.
In the present context I will focus on metaethical forms of
moral relativism and set nihilism to one side. Moral (meta-
ethical) relativism, in turn, comes in at least three forms.
One kind of moral (metaethical) relativism is semantic (or
‘indexical’) moral relativism, according to which, when we
pass moral judgements, we make an implicit reference to
some system of morality (our own). According to this kind of
moral relativism, when I say that a certain action is right, my
statement is elliptic. What I am really saying is that, accord-
ing to some (adequate) moral framework or system S, to
which I adhere, for example the one prevailing in my culture,
this action is permitted.3 I will reject this kind of relativism.
According to another kind of (metaethical) moral relativ-
ism, which we may call epistemic, it is possible that, when
one person (belonging to one culture) makes a certain moral
judgement, such as that this action is right, and another per-
son (belong to another culture) makes the judgement that the
very same action is wrong, they may have just as good rea-
sons for their respective judgements; it is even possible that,
were they fully informed about all the facts, equally
 MORAL RELATIVISM 125
imaginative, and so forth, they would still hold on to their
respective (conflicting) judgements. They are each fully justi-
fied in their belief in conflicting judgements.4 I will comment
on this form of moral relativism in passing.
Finally, however, there is a third kind of (metaethical)
moral relativism we could call ontological, according to
which, when two persons pass conflicting moral verdicts on a
certain action, they may both be right. Neither of them make
judgements with any implicit reference to any system of
norms. They both use their moral vocabulary in an absolute
sense. An objectivist non-natural moral analysis (in the style
of G.E. Moore or Henry Sidgwick) of what they say gives a
correct representation of what they are doing. And yet, for all
that, they pass conflicting judgements. The explanation why
they can both, in an absolute sense, be ‘right’ in their judge-
ments, is that they inhabit different moral (socially con-
structed) universes. So while it is true in the first person’s
moral universe that a certain action is right, it is true in the
second person’s moral universe that the very same action is
wrong. I intend to explain and defend this version of onto-
logical moral relativism.

2. MORAL UNIVERSES

2.1. How Are We to Think of Moral Universes?

A moral universe consists of a system of common sense
morality. We may compare common sense morality to gram-
mar. And we should remember that it is possible to distin-
guish between descriptive and regulative grammar. In
descriptive grammar we observe how language is actually
used. We formulate hypotheses, and we try to find general
answers to questions about language use. But we may also
discuss a language from a regulative point of view. We may
try to answer questions such as: is it correct to use certain
words in a certain order? We may want to articulate general
principles also in regulative grammar. However, these princi-
ples are not descriptive of the language in question. Rather,
126 TORBJÖRN TÄNNSJÖ

they answer the question: what makes a certain way of using

language a proper or correct one? They provide reasons for
answers to this question.
In regulative grammar we take for granted that there are
facts of the matter. There are right and wrong ways of using
language. These are facts. These facts do not exist indepen-
dently of us, of course. We may think of them as social con-
structions, i.e. as constituted by us. If we were to speak
differently, then different facts would obtain. And yet, for all
that, these facts are in a sense objective. We may be ignorant
about them. By sound argument we may be set straight with
respect to them.
I do not intend to defend any detailed account of how so-
cial constructions become possible. It should suffice here to
say that, as far as I can see, John Searle seems to be on the
right track in his characterisation and defence of social con-
structivism.5
It is worth observing that there are two different kinds of
answers to the question what ‘makes’ a certain use of lan-
guage a proper or correct one. One (regulative) kind of an-
swer does indeed stipulate a reason for a judgement about the
case: it is wrong to put the words in a certain order because,
in the language in question, the noun phrase must precede
the verb phrase. This is the kind of reason stated in a princi-
ple or rule of regulative grammar. However, the same ques-
tion, i.e. the question what ‘makes’ a certain use of language
a proper or correct one, can also be understood as an onto-
logical question about what constitutes right and wrong in a
language. And then the answer must be along the following
lines: the correctness of a certain way of using the words is
constituted by the fact that this is how the words are actually
used in the linguistic community in question. But this is not
the end of the matter. It is also crucial how experts on gram-
mar assess this way of using words. If they condone this way
of using the words, this contributes to this way’s being a
proper one. And, of course, one reason why the experts con-
done a certain way of using the words is that this is how the
words are being used. But it is also true, to some extent, at
 MORAL RELATIVISM 127
least, that the fact that this way of using the words is con-
doned by the experts contributes to the explanation of why
the words are used this way.
In a similar vein we may think of (conventional common
sense) morality (in a society). Conventional common sense
morality is learnt by children in a manner similar to how they
learn their mother tongue. They learn that it is right or wrong
to perform certain actions. Why is it right or wrong to per-
form a certain action? Once again there exist two ways of
understanding and answering the question. A certain rule can
be given, providing a reason to perform, or not perform, the
action in question. Or, the question may be understood as an
ontological question as to what constitutes right and wrong
action. Once again, the answer to the latter (ontological) ques-
tion has something to do with both how people actually be-
have, but also with how moral experts, or even people in
general, tend to judge this kind of behaviour. These facts, in
turn, may have, and certainly must have, some kind of natural
explanation, for example one in terms of evolutionary biology.
Socially constituted moral norms seem to come to us in the
form of a moral universe. This means that we think of them,
ideally, at any rate, as complete. We demand of the norms
making up our moral universe that, in principle, they answer
all moral questions (even though the answers may be hard to
come by for us) in an unambiguous way. And the set of an-
swers to these questions is the moral universe (a set of moral
facts). Now, this is an ideal. In real life it may well turn out
that there are moral questions that get no satisfactory solu-
tion. But then the typical reaction of a critical person is to try
to develop such answers and have them accepted by others.
This means that in the moral universe pursued by the moral
reformist, who is bent on moulding common sense morality,
not only answers to actual problems are sought for but also
answers to hypothetical cases. For there seems to be no better
way of testing alternative moral hypotheses, than to resort to
thought-experiments.
In general, if we want to find out about what constitutes
right and wrong action in a certain society, we should focus
128 TORBJÖRN TÄNNSJÖ

more on how people justify their actions than on what they

do. A crucial feature of morality is that we can use the
morality a person explicitly adheres to when we want to criti-
cise his or her actions, once they are at variance with this
morality. But, at the same time, if everybody tend to perform
a certain action, condemned by common sense morality, this
may mean that, at the end of the day, this action is actually
condoned by common sense morality. Common sense moral-
ity is in much way a malleable and changing entity. And
some moral reformists may well play a crucial role when
common sense morality develops.
How are we to distinguish moral (socially constituted) facts
from other kinds of (socially constituted) facts, such as facts
concerning etiquette, the existing legal situation, and so forth?
I suppose one distinguishing fact is that moral facts are
more basic. Moral considerations seem to override other
kinds of consideration. In fact, only moral considerations
provide us with, in Kant’s sense, categorical norms. Even if,
linguistically speaking, it is proper to put a comment in one
way and improper to put it in another way, this does not
mean that one ought to put it in the proper way. In a similar
vein, even if a certain action is legally prescribed, economi-
cally advantageous, and so forth, this does not as such mean
that we ought to perform it. This fact may be hidden, to
some extent, by the fact that we may believe that we have a
moral obligation to obey the law. Some may even believe that
we have a moral obligation to use language properly. How-
ever, without such obligations taken for granted, law and reg-
ulative grammar lack categorical normative force.
Moreover, to be accepted as a ‘morality’ a normative out-
look must not be too idiosyncratic, i.e., it has to shared to
some extent, it has to be supported by a kind of common
sense.
It is a moot question to what extent, within an existing
common sense morality, we can really find principles explain-
ing all actual cases of right- and wrongdoing. This problem
has a clear parallel within grammar. However, even if, in the
final analysis, experts must take up a rather ‘particularistic’
 MORAL RELATIVISM 129
view, and concede that they find no deterministic principles
capable of explaining all cases of right- or wrong-doing (or
proper or improper grammar), we all tend to believe that, in
many cases, there are correct answers to these questions.
Even if we cannot tell for certain why this is so, we know
that it is wrong to perform certain actions (or we know that
certain grammatical constructions are ill formed).
Furthermore, in some cases where we feel that we cannot
articulate deterministic principles, we tend to believe that we
can settle cases by arguing ‘casuistically’, from paradigmatic
cases to actual cases. Of course, all this is compatible with
the possible fact that there are some cases that are ultimately
undecidable. However, as was stressed above, since we con-
ceive of our common sense morality as constituting a moral
universe, critical people do not stay satisfied with this kind of
observation. They try to improve on their morality in order
to render it, in the relevant sense, ‘complete’.
It would perhaps be far-fetched to claim that, just as there
are different languages, there are completely different morali-
ties (in different societies).6 However, I think it safe to claim
at least that, just as there are somewhat different dialects or
even idiolects existing within a linguistic community, there are
somewhat different moralities existing in different societies.
Here is a possible example. In most (all) actual societies
(with their corresponding moral universes) women are in
many respects treated worse than men. However, while in
some societies such unequal treatment can be defended on
the ground that, by treating women worse than men, all
get what they deserve, such a defence is not possible in an-
other society (with another and different moral universe,
where the value of women and men is constituted as
equal).
Or, should we adopt some constraint to do with the con-
tent of a moral universe and claim that the former universe is
not moral? This is something we do when we pass moral
judgements, of course. If we feel that men and women are of
equal worth we may hesitate to speak of the view that men
are more valuable than women as a ‘moral’ one. We will
130 TORBJÖRN TÄNNSJÖ

speak of it as immoral. However, when pursuing metaethics,

it might be a good idea not to include any requirement that a
socially constituted system of norms should have any content
in particular in order to count as a ‘moral’ universe. The cru-
cial thing is that the norms composing it have an overriding
nature and are not too idiosyncratic.
All this means that while, from the perspective of one exist-
ing common sense morality (society), an action may be right,
it may yet, for all that, be wrong, when judged from the per-
spective of another existing common sense morality.
Here a comparison with John Mackie’s position may be
instructive. Ontologically speaking, Mackie is a moral con-
structivist. Semantically speaking, he is an objectivist. But he
believes that all objective positive moral judgements are false.
His error theory is a strange view. If there are socially consti-
tuted moral facts, why not think of moral language as
descriptive of them? This is what I have here suggested. If
Mackie had followed this line, instead of adopting his idea
that all positive moral statements are false, his theory had
been more coherent. After all, in the final analysis, when pur-
suing normative ethics Mackie himself seems to be a moral
constructivist of a kind.
To substantiate this claim we must find a way of disarming
Mackie’s argument from queerness, of course, while retaining
his argument from relativity. However, it is not difficult to
find faults with his argument from queerness. His idea that
moral notions are internally ‘prescriptive’ can be jettisoned,
and we then arrive at an externalist version of moral objectiv-
ism free of queerness.

3. BACK TO SEMANTIC RELATIVISM?

But does not all this mean that we are back to a semantic rel-
ativism of a naturalistic, indexical kind? Is the claim made
above not just that, when a person in one society asserts that
a certain action is right, what this person is saying is that,
from the perspective of the set of moral rules operative in his
or her society, this action is right?
 MORAL RELATIVISM 131
No, this is not the proper way of understanding the claim.
Note that we have distinguished between descriptive and reg-
ulative grammar; in a similar vein we ought to distinguish be-
tween descriptive and normative ethics. So, when a person
asserts that a certain action is right, this assertion is indeed
normative. The assertion is that the action in question is
right, period. The analysis of the assertion is non-naturalist.
No implicit reference to the system of norms is made; an
objective claim, with reference to socially constituted moral
facts, is being made. We may say that this assertion is true in
the moral universe in question if and only if the action in
question is right. The explanation of why it is right, simplicit-
er can be given, however, with reference to (the content of,
not the existence of ) principles in this system of morality – to
the extent that moral principles can be formulated within this
system of morality. The claim as such makes no reference to
the existence of these principles. It is normative and categori-
cal and objective and allows of non naturalistic analysis.
What are we to say about a situation where two persons
from different cultures make conflicting judgements about a
certain action?
Well, in order for this scenario to be possible, at least one of
the systems must have a general scope. If both systems restrict
their judgements to actions within their respective societies, no
conflict will emerge. However, while this policy of live, and let
live is comparatively common in a linguistic context (especially
between different languages but also, to some extent, between
different dialects), this kind of moral relativism (with such a
restricted scope) is rare. We tend to judge, from the point of
view of our own morality, the manners of others. This is ren-
dered possible by the fact that two different moral universes
may share (a part of) the one and only existing actual empiri-
cal universe. If a certain concrete action is part of both moral
universes, it may be right in one of the universes while wrong
in the other universe. And, as I have stressed above, if we con-
ceive of ourselves as inhabiting a moral universe, there will be
a drive towards completeness. We want answers to all moral
questions, also those arising in alien cultures.
132 TORBJÖRN TÄNNSJÖ

Suppose the two persons are actually making conflicting

moral judgements about the very same action. Let us sup-
pose, for example, that while one person asserts that a certain
action, such as the circumcision performed on a young wo-
man in the society to which this person belongs ought to take
place, another person, belonging to another society, asserts
that this very same action ought not to take place, how are
we to understand their conflict?
This is indeed a conflict. So we should avoid interpreting
the judgements as elliptical, with an implicit reference to each
person’s own system of norms (somehow corrected). They are
both to be taken as issuing categorical judgements about the
very same action. And these judgements are indeed, we may
assume, true in each moral universe. And they are ‘objective’
in the sense that they may be true, irrespective of whether
anyone knows about this. But does not this mean that they
cannot be contradictory?
I believe we should say that these people do not, strictly
speaking, contradict each other. And this is due to the fact
that their respective judgements differ in meaning. In saying
so, we assume a moderate amount if externalism about mean-
ing, of course. Each person, in each universe, refers to a sim-
ple property, but the properties differ between universes. One
judgement is true in virtue of facts obtaining in one universe;
the other is true in virtue of facts obtaining in the other uni-
verse. People inhabiting different moral universes may agree,
of course, that it is true (in their universe) that a certain ac-
tion is right, if and only if this action is right, but this, seem-
ingly identical truth condition, has different meaning in
different moral universes. We cannot analyse this meaning in
any natural terms. In each universe, the notion of obligation
is simple and not definable in any non-moral terms. However,
even if the truth-conditions of each judgement are in a way
inscrutable, they differ. And this means that these persons do
not contradict each other after all. The reason is that they
have constituted to themselves different moral universes.
They need not themselves be aware of this, of course. Each
may think that there is one true morality (theirs).7
 MORAL RELATIVISM 133
If we reject externalism about meaning we may perhaps say
that moral terms mean the same in different moral universes;
the same proposition is true in one universe, but false in an-
other universe. I don’t accept this view of meaning, however,
so I will not pursue this rather startling idea any further.
What has now been said seems to imply that we face a
kind of semantic relativism after all. And in a way we do.
However, this is a new and subtler form of semantic relativ-
ism (as compared to the traditional variety described in the
opening paragraph) since, taking norms to be categorical, it is
compatible with the fact that the judgements in question are
in conflict.
How are we to conceive of this conflict? Well, even if the
judgements are logically compatible, they, and not only those
who issue them, are in conflict in a practical sense. This prac-
tical conflict is rendered possible by the fact that these nor-
mative judgements have not only truth-conditions, but
satisfaction conditions as well. The claim that the woman in
question ought to be circumcised is satisfied if, and only if,
the women is actually circumcised. And note that this satis-
faction-condition is the same in all moral universes (since it is
not cast in moral terms).
The normative claims that the circumcision ought to take
place, and that it ought not to take place are, therefore, even
if logically consistent (because issued from different moral
universes using the word ‘ought’ in different ways), still
incompatible. There is no way to satisfy both these claims.
This practical inconsistency explains why persons belonging
to different moral universes, making moral judgements that
are, from a strict logical point of view, consistent, may feel a
need to sort out their conflicts. Even if they need not feel any
intellectual drive to do so, they may well, for practical rea-
sons, find it urgent to reach an agreement.
It is certainly true that even a semantic relativist of the
indexical kind, who interprets moral judgements as elliptical,
with an implicit reference to a system of norms (where ‘this is
wrong’ means, for example, that according to the norms in my
society, this action is considered wrong), may speak of a kind
134 TORBJÖRN TÄNNSJÖ

of tension between the judgement made by one person,

belonging to one society, to the effect that a certain action is
right, and the judgement made by another person, belonging
to another society, to the effect that it is wrong. It is probable
that these persons have attitudes and interests that conflict.
However, since, upon closer inspection, their respective judge-
ments are descriptive (not normative), their judgements as
such cannot conflict in the way the judgements issued from
different moral universes can do. The latter, in contradistinc-
tion from the former, have not only truth-conditions but sat-
isfaction-condition as well. This is why we can say that it is
sometimes not possible to satisfy conflicting moral
judgements, issued from different moral universes.8

4. MORAL EXPLANATION

Furthermore, while it is hard to see how we could offer moral

explanations, if semantic relativism of the naturalistic, indexi-
cal, variety were true, this is something we can do on the
(objectivist) version of relativism here defended. If ‘This is
wrong’ means that, according to a certain (adequate,) frame-
work or system of norms S, this is not permitted, then there
seems to be no way of explaining what it is that makes the ac-
tion in question wrong. There may be a causal explanation
why the action has come to be prohibited, but there is no pos-
sibility of finding any moral explanation of its wrongness.
However, on the version of relativism here defended, we may
easily find a moral explanation of why the action in question is
wrong, provided the moral universe is rich enough to provide
such an explanation. We may say, for example, that it is wrong
since it means that an innocent person is harmed (provided it is
wrong to harm innocent persons in the universe in question).
But this fact cannot explain why, according to S, it is not per-
mitted. The (social) fact that it is not permitted by S, if it can
be explained at all, must be explained in terms that seem to be
morally irrelevant. The explanation might be, for example,
that the action in question is condemned in the system because
the system condemning these kinds of actions have had a
 MORAL RELATIVISM 135
certain function, explaining why it has been selected, and so
forth. Such an explanation is interesting from a sociological
point of view but irrelevant from a moral point of view.
Perhaps a moral relativist of Harman’s or Wong’s kind
would like to object that a moral framework (in one of
Harman’s formulations) or an adequate moral system (in
Wong’s formulation) will or is likely to contain a norm for
harming innocent persons. Why is it a more satisfying expla-
nation to refer to facts in a socially constructed moral uni-
verse about the wrongness of harming innocent people than it
is to refer to the norm of an adequate moral system that pro-
hibit the harming of innocent people?9 But my point is not
that one explanation, qua explanation, is better than the other.
If the person who makes the explanation in question is clever
enough, there is no limit to how good each explanation can
become. However, a crucial difference still remains between
them. One is cast in moral terms. The explanans of the kind
of explanation I defend is moral in character. The ‘covering
law’ included in the explanans is a moral statement (principle)
proper. The explanation can be spelled out in the following
manner. It is wrong to harm innocent people (principle), say,
and this is an innocent person (we assume) who has been
harmed (we observe). Hence, this action was wrong. It was
wrong because it meant that an innocent person was harmed.
The kind of explanation that Harman and Wong can offer
can be just as good, but it is different in kind. It doesn’t rely
on any moral premises at all. It says something like the fol-
lowing. This action is wrong. This means that the action is
forbidden in a relevant moral framework or adequate moral
system. Why is it forbidden? Well, the most plausible hypoth-
esis is perhaps that this has to do with the functioning of
moral frameworks or adequate moral systems. They have
developed the way they have because they help people in a
society or group to correlate their actions. This moral frame-
work or moral system serves this function well and that is
why it is in place (this is the ‘covering law’ included in the
explanans). This may be a perfectly sound and informative
(causal) explanation. And yet, it is not a moral explanation.
136 TORBJÖRN TÄNNSJÖ

5. LOGICAL RELATIONS BETWEEN MORAL JUDGEMENTS

According to the kind of moral absolutism here defined, there

are indeed moral judgements, capable of being true or false.
It is true that the meaning of words such as ‘right’ and
‘wrong’ tend to vary between different moral universes, but
within each moral universe there exist true and false moral
judgements. This means that we can conceive of logical rela-
tions in terms of truth. A moral argument is valid if, and
only if, necessarily, if the premises of the argument are true,
then the conclusion of the argument is true as well. To obtain
the right moral implications we need to clarify the normative
terminology, of course. But this can easily be done, for exam-
ple in the following manner. A particular action is right if
and only if it is not wrong. And an action is obligatory if and
only if it is wrong not to perform it. Since rightness, wrong-
ness and obligatoriness are, even if socially constituted, yet
objective and genuine properties of particular actions we need
only standard first-order predicate logic to get our deontic
logic off the ground.
Objectivist relativism therefore means an advantage over
the kind of moral relativism defended by Gilbert Harman, in
that it retains normal logical relations between moral propo-
sitions. Harman’s semantic relativism does not. At least this
is how things seem to me. But Harman himself thinks other-
wise. According to Harman, his brand of moral relativism re-
tains ordinary logical relations between moral judgements. Is
he right about this?
According to Harman, in order to grant truth-conditions
to our moral judgements, we have to conceive of them as
elliptical, with a reference to a moral framework of some
sort. This means that they are descriptive judgements. When I
say that it is wrong of P to do D, I say, if we spell it out,
that, according to the moral framework with which I’m asso-
ciated, it is wrong of P to do D. This allows us to say that
moral judgements are true or false, of course. But does it also
allow us to derive moral implications from standard implica-
tions among relativised propositions? I think not. For
 MORAL RELATIVISM 137
suppose a relativist says that it is wrong of P to do D, while
maintaining that it is right of P to do A, while acknowledging
that, by doing A, P does D. Could this person be set straight
with reference to his moral framework? We cannot take for
granted that this is possible. As a matter of fact, his frame-
work may well contain all these judgements.
I suppose Harman would claim that this means that there
is an inconsistency in this moral framework, and he wants
such inconsistencies to be eliminated from the framework be-
fore we assign truth-values to moral judgements with refer-
ence to it. But then the question arises what it means when
we say that the framework is inconsistent? According to
Harman, the moral judgements that constitute the framework
lack truth-values.
We may here speak of practical inconsistencies, of course.
A practical inconsistency is detected, as we have seen, when
we realise that, in some situation, it is not possible to satisfy
two normative judgements. If the framework contains both
the idea that it is always wrong to tell a lie and the idea that
it is always mandatory, at all means, to save lives, and we
run upon a situation where, only by telling a lie, a person can
save a life, then we know that the framework is, practically
speaking, inconsistent. The mere possibility of such a situa-
tion means that the system is inconsistent.
However, Harman owes us an explanation why we should
avoid practical inconsistencies in our moral framework. If the
norms making up the framework are neither true nor false,
the purpose to be filled by the framework must certainly be
practical. We have it for a purpose. Suppose it fulfils its pur-
pose, in spite of containing some practical inconsistencies.
Why change it? And if one wants to change it, how should
one change it?
Harman could argue that, under the circumstances, there is
no need to change the system. The corrected system is to be
seen as a mere theoretical device. Our moral judgements can
be understood and given truth-conditions with reference to it.
I suppose Harman’s answer to the latter question, as to
how the system should be changed, when we construct our
138 TORBJÖRN TÄNNSJÖ

system we need to refer to, when we give truth-condition to

moral judgements, would be to make minimal changes in the
actual, inconsistent framework, in order to get rid of the
practical inconsistencies. However, there may more than one
way of doing this. Or, there may exist no way of doing this
while having the framework fulfilling its purpose.
Furthermore we want to say that there are moral inconsis-
tencies that do now show up in practice. There is a way to
satisfy both the norm that one should not work on a Sunday
and the norm that one may either work on a Sunday or not.
Both norms are satisfied if, as a matter of fact, we do not
work on Sundays. However, we feel that these norms are
inconsistent. But there is no way for Harman to account of
their inconsistency. So I conclude that moral relativism of
Harman’s variety does not retain standard logical implica-
tions among normative judgements.
Could not the same thing be said about socially con-
structed normative properties? Could not a socially con-
structed moral universe contain contradictions? I think not.
When we take our moral judgements to be true or false, we
may conceive of moral inconsistencies in the ‘ordinary’ way.
Two inconsistent norms cannot both be true. And we will
thus, in our pursuit for truth, have an incentive to get rid
also of moral inconsistencies. When we think of moral facts
as constitutive of a moral ‘universe’, it becomes imperative
that all moral truth can be combined into one single and true
(consistent) conjunction describing this universe. And our
well-known epistemic goal of believing a proposition if it is
true, and of not believing it if it is false, applies to our moral
beliefs just as well as to other aspects of our belief system.

6. SOCIAL CONSTITUTION AND THE ‘IS’ — ‘OUGHT’

QUESTION

If there exist socially constituted moralities, does this mean

that moral truths can be derived from descriptions of our ac-
tual behaviour? It does not. On the account here given, the
gap between ‘is’ and ‘ought’, between fact and value, remains.
 MORAL RELATIVISM 139
It is true that, by making certain judgements, and forming
certain expectations, we come to constitute a certain morality
within our society. However, constitution is one thing, logical
implication quite another thing. Constitution is not a logical
relation.
However, when there exists a relation of constitution, must
there not exist also a relation of explanation? And does not
such an explanation take the form of a deductive argument?
In particular, can we not explain a certain moral fact (a par-
ticular obligation) with reference to the fact that we behave in
a certain (systematic) manner? And doesn’t this mean that we
can derive an ‘ought’ from an ‘is’?
No, we cannot. This is not to take the idea of constitution
seriously. In order to explain the particular moral (normative)
fact we must have recourse to (socially constituted) moral
principles. The moral facts we constitute are of an ontological
kind in their own right. What follows from the description of
our behaviour is merely the fact that we tend to accept certain
moral propositions, not that these propositions are true.
This may sound odd, but this is the way socially consti-
tuted (institutional) facts behave with respect to brute facts.
And this is indeed why we speak of socially constituted facts
as forming an ontological category in its own right.
Note however that this ontological difference between brute
and social facts is consistent with social facts being superve-
nient upon brute facts so that, unless there is a ‘brute’ differ-
ence between two situations, a socially constituted fact cannot
exist in one of them but fail to exist in the other.

7. MORAL RELATIVISM AND MORAL REALISM

It should be noted that if moral ontological relativism, of the

kind here described, is true, then epistemic moral relativism is
trivially true. It is obvious that, if conflicting moral judge-
ments, made by people living in different moral universes,
may be true, then each advocate of each one of the conflict-
ing views may have equally good reasons to support his or
her favoured position. But what should a ‘transcendental’
140 TORBJÖRN TÄNNSJÖ

moral realist of Moore’s or Sidgwick’s brand say about the

view just outlined?
It is true that, most philosophers who have advocated
some kind of social constructivist view of ethics have been
‘nihilists’, in the sense that they have denied that there are
any moral facts, existing independent of our thinking and
acting.10 Social constructivism has been resorted to by them
as a kind of ersatz realism. However, this does not mean
that social constructivism is inconsistent with strong moral
realism. Indeed, I think it would be foolhardy of a ‘tran-
scendental’ moral realist, who believes that there are objec-
tive moral reasons and facts existing independently of us to
deny that different (conventional) moral universes of the
kind here described exist as well. Moral realists should be
prepared to accept that, when some people pass moral ver-
dicts on actions, they express their views from the perspec-
tive of the conventional morality into which they have been
socialised. So normative ontological relativism is made true
by the existence of this kind of socially constituted or con-
structed conventional morality. However, a moral realist,
who believes that there exist moral facts independently of
our conceptualisation or actions, should be expected to
want to add to this picture that, once the existing different
moral universes have been described, there is a way of
transcending them all. One more question remains to be
asked and answered: which one, if any among competing
moral claims, is the uniquely correct one? They must be
supposed to ask for reasons that are not only categorical
and objective but also such that they are in no way of our
making.
When the moral realist of Moore’s or Sidgwick’s variety
poses this question, using the standard moral terminology in
the way he or she does, it is quite possible that the question
cannot be answered with reference to any principles designed
to rationalise any actual conventional morality. And, more
importantly, the answer to the question, if such an answer
exists, cannot be taken to be constituted by us. On the con-
trary, it is assumed that the answer is ‘out there’, to be
 MORAL RELATIVISM 141
found, in the same way that answers to questions posed by
physicists are thought to be ‘out there’ to be found.
Suppose the moral realist is right about this. Let us sup-
pose that the moral realist is right in insisting that, when dif-
ferent conventional moralities have been identified, there is
one more question to be asked and to be answered: which
one (if any one) of two competing moral claims, is the correct
one?11 What does this mean?
Well, it is obvious that the moral realist is here using the
moral vocabulary in a slightly different sense from the senses
used in the two competing conventional moral universes. But
it is no coincidence that all the parties to this controversy
translate the relevant words into ‘right’, and ‘wrong’ in their
own terminology. For even if these terms, when they occur in
different moral universes, take on a somewhat different mean-
ing, they all share an important function. They are all cate-
gorical and they do all guide choices.
This is not to say that ‘internalism’ must be true in any
strong sense of any of these terminologies. I have already
indicated that it should be rejected. In particular, of the real-
ist vocabulary I think internalism is true only to the extent
that people who use this terminology tend to want to do what
the believe is right. However, this is enough to give them
practical reasons to do what they believe they have moral
reasons to do. And it is not far-fetched to assume that this is
true also of people who abide by their respective conventional
moralities. And this, then, explains why they all translate the
moral terms the way they do.
What does all this mean for how we should understand the
moral conversation between a moral realist and a person who
has never contemplated this ‘further’ question about absolute
rightness, merely passing moral judgements from his or her
conventional moral universe?
The obvious answer is that we should understand this dif-
ference of opinion in a way similar to how we understood the
conflict between people belonging to different (conventional)
moral universes. The moral realist and the person issuing
moral judgements from the perspective of a particular
142 TORBJÖRN TÄNNSJÖ

conventional moral universe pass moral judgements that

are logically consistent (because they have different truth-
conditions); however, these judgements are, since they are
categorical and normative, and therefore without implicit
reference to any moral universe, truly conflicting. In many
cases these persons will find that there is no way of satisfying
the moral demands put forward by the other party. So there
will be a drive somehow to solve the conflict.

8. CONCLUSION

Global ontological relativism is a self-defeating position, or

so I would be prepared to argue. However, the existence of
many different conventional moral systems or moral ‘uni-
verses’ prepares the way for a more restricted kind of moral
ontological relativism. I have argued that such moral onto-
logical relativism is true. Here Nelson Goodman’s daring
metaphor of ‘world-making’ is indeed appropriate.
It is noteworthy that even a moral realist who believes that
there exists one uniquely true morality ‘out there’, to be found
by us, may concede that moral ontological relativism of the
kind here described is true. The moral realist need not reject
relativism, but only add a realistic element to the picture.

NOTES
1
Work on this paper was partially supported by the Bank of Sweden
Tercentenary Foundation in connection with the project Relativism. Many
people have made valuable comments on earlier drafts of the paper, in
particular I would like to thank an anonymous reviewer for Philosophical
Studies, Ragnar Francén, Gilbert Harman, Hans Mathlein, Folke
Tersman, and David B. Wong.
2
Cf. Charles L. Stevenson, Facts and Values (New Haven: Yale Univer-
sity Press, 1963), Ch. 11, in particular p. 219, about this.
3
This version of relativism is famously defended (in several places) by
David Wong and Gilbert Harman. It should be noted that Harman
doesn’t put forward his theory as a claim about what we actually mean
by a word like ‘right’; rather, if we want to understand the use of ‘right’
in moral context as contributing to claims that have truth-values, then this
is how we should understand these uses.
 MORAL RELATIVISM 143
4
I know of no philosopher who has actually defended this (indeed
quite defensible) position.
5
See John R. Searle, The Construction of Social Reality. It goes without
saying that I do not agree about any detail in Searle’s analysis, but that
does in no way affect the argument in this paper.
6
Some moral relativists may be prepared to make this claim, but I find
it exaggerated; in general, relativists tend to exaggerate the degree to
which different moral communities hold on to different basic moral posi-
tions. It seems to me that much apparent moral disagreement can be ex-
plained away as depending on different empirical beliefs.
7
This is true of semantic, indexical kinds of relativism as well, of
course. Neither Harman nor Wong need to insist that it is obvious to
speakers referring to different frameworks or systems, that this is what
they are actually doing.
8
A way for the semantic relativist (naturalist) to account for disagree-
ment could be to incorporate an expressivist element in the meaning or
function of ethical statements, of course. I owe this observation to an
anonymous referee for this journal. However, by so doing it may be hard
for the relativist to explain why not go expressivist through and through
and give up the claim about descriptive (indexical) meaning altogether.
9
I owe this objection to an anonymous referee for this journal.
10
For example, this seems to me to be a correct diagnosis of John
McDowell. See for example his ‘Non-Cognitivism and Rule-Following’, in
Steven H. Holtzman and Christopher M. Leich (eds.), Wittgenstein: to
Follow a Rule, London: Routledge & Kegan Paul, 1981.
11
Note that even if the moral realist is right about this, there may be no
way of finding out the answer to this question. If this is not possible, a
kind of strong moral epistemic relativism seems to emerge. On this version
of epistemic relativism a unique moral truth exits, so two conflicting mor-
al opinions cannot both be right. And yet, for all that, two people may
have equally and perfectly good reasons to believe in their favoured solu-
tion to the moral problem.

Department of Philosophy
Stockholm University
Stockholm 106 91
Sweden
E-mail: torbjorn.tannsjo@philosophy.su.se
View publication stats