Excerpt from A Better Alternative: Intuitionism and Reasons Externalism

A. Moores argument against classical ethical naturalism

Adams understands classical naturalism to be the view that ethical sentences have
cognitive meaning and truth-values but are translatable into the language of science, (pp. 35-36)
where the language of science is taken to be descriptive-explanatory in that it describes what and
how things are and explains why they are as they are. At least in part, classical naturalism is a
semantic view that says that the meaning of value-language (language about what is good or bad,
right or wrong, is a virtue or a vice, what ought to-be or ought not to-be, is rational or irrational,
etc.) can be given in naturalistic terms. For instance, a classical naturalist who is a
eudaimonistic utilitarian might hold that morally obligatory means produces more happiness
on balance than any other alternative, or if all the alternatives produce unhappiness on balance,
produces the least amount of unhappiness on balance. However, Adams interprets classical
naturalism to be more than just a semantic view, for he holds that it also implies an ontological
commitment, namely, to the view that what is real is nothing beyond what the descriptiveexplanatory part of common sense and science implies is real (p. 36). For Adams, classical
naturalism is a semantic view with naturalistic ontological commitments.
G. E. Moore famously argued that what I am calling the semantic part of classical
naturalism is false, at least when it is a claim about the meaning of intrinsically good. Adams,
and others, have called this argument the naturalistic fallacy argument. He then wonders how
various interpretations of the argument involve a fallacy rather than just a mistake, and concludes
that the naturalistic fallacy argument does not refute any naturalistic position (see, p. 48). But
the naturalistic fallacy is not supposed to refute ethical naturalism; it is not an argument against

2
naturalism. Rather, it is just supposed to explain how people have mistakenly (fallaciously) been led to
that view. The Open Question Argument is supposed to refute naturalism; the naturalistic fallacy is
supposed to be a diagnosis of how people have mistakenly (but understandably) come to think that
ethical naturalism is true.
Suppose we mean by a fallacy an argument that appears good but is actually bad.1 Arguments
that commit the fallacy of equivocation can satisfy this definition of a fallacy. For instance, the
following argument commits a subtle scope fallacy: (i) what you know must be true, (ii) no proposition
about the external world must be true (if true, they are only contingently true), (iii) so we can have no
knowledge of the external world. It is not a good argument, even though it seems to to be. The problem
involves an equivocation in (i), for it can be read as either: (Q) if you know something, its necessarily
true or (R) necessarily (if you know something, its true). Read as (Q), the argument is valid, but (Q)
seems false, or at least begs the question. Why cant we know contingent truths? Why are only
necessary truths knowable? Read as (R) the argument is invalid even though (R) is true. (R) says that
what is known is true. (ii) says that no statement about the external world is a necessary truth, but that
does not imply that all such claims are not true, that is, are false. So (ii) does not deny the consequent
of (R). There is no way to read the premises so that they are all true and the argument is valid, that is,
there is no way to construct a sound argument from (i) and (ii) to (iii). Still, at first glance the argument
appears to be sound, but it is not. The argument appears to be good (sound), but it is bad (unsound).
The argument commits a fallacy.
I believe that Moore was doing two things, not one. He gives the Open Question Argument to show that

1Adams understands a fallacy to be a violation of some rule of logic (p. 70), but such a violation may
be a mistake, or blunder, but not a fallacy. If I affirm the consequent in arguing: if something is a cat,
its an animal; Ebony (a dog) is an animal; therefore, Ebony is a cat, I have violated a rule of logic but
have not committed a fallacy. I have made a logical mistake, or made a blunder, but have not
committed a fallacy.

3
good cannot be defined in naturalistic terms, and he also tries to diagnose how people went astray in
thinking that it could. His diagnosis is that they committed a fallacy. The best way to understand Moore
is to interpret him as saying that the general fallacy stems from equivocating on is. People rightly
think, say, that pleasure is good, thus employing the is of predication; pleasure has the property of
being good. They then confuse this sense of is with its meaning in, say, a brother is a male sibling,
and conclude that good means pleasure since brother means male sibling. Pleasure is a
naturalistic term, but Moore says that the same fallacy would occur if people came to think that good
meant approved by God.2 They would go from what is good is approved by God to good means
approved by God. In both cases, people would come to think that one term meant the same as another
by equivocating on is, that is, by confusing the is of predication with the the is of identity of
meaning.
People might reach a similar conclusion about what good means by equivocating on the is of
identity, or even of necessary identity, failing to distinguish it from the is of means the same as,
that is, of identity of meaning. Suppose someone accepted, a creature with a kidney is a creature with a
heart and that a brother is a male sibling, and then concluded a creature with a kidney means the
same as a creature with a heart.3 They would have equivocated on is, confusing the is of identity,
which appears in the sentence about creatures with kidneys, with the is of identity of meaning, which
appears in the sentence about brothers. Or suppose someone accepted that necessarily, water is H2O and
that necessarily, a brother is a male sibling, and then concluded that water means H2O. They
would have equivocated on is, confusing the is of metaphysically necessary identity with the is
2See Moores Principia Ethica (1903), Chpt. II, sec. 25 where he says, ...the fallacy, by reference to
which I define Metaphysical Ethics [which defines good in terms of some supersensible object], is
the same in kind [as found in Naturalistic Ethics, which defines good in terms of some natural
object]; and I give it but one name, the naturalistic fallacy. My bracketed material.
3See Principia Ethica, Chpt. I, the end of sec. 9, where Moore says that even if intelligence, and
intelligence alone, were good, it is still not true that good means intelligence.

4
of identity of meaning.
Adams considers the possibility that Moore means that ethical naturalists confuse the connotation of
two terms with their denotations when their denotations are the same. So they might confuse the
connotation of some non-value expression with the connotation of some value expression if the two
were co-extensive, in the same way that a person might come to think creature with a kidney means
the same as creature with a heart or that water means the same as H2O. While he grants that this
would be a fallacy, he thinks there is no evidence to think that ethical naturalists make this mistake (p.
43). But Adams is assuming that the person has clearly distinguished the meaning of the value term and
the meaning of the naturalistic expression, and then concludes it is implausible to think that such a
person would then confuse the two meanings (p. 44). But my diagnosis of the confusion, and I think
Moores, is that people confuse the meaning of is, not the meaning of the value and non-value terms,
and then reach the conclusion that some naturalistic expression means the same as some value term.
Adams rightly says that if people confuse the meaning of the naturalistic expression with the value
term, they would make a mistake, not commit a fallacy. But I believe that Moore thought that ethical
naturalists arrive at their semantic view by equivocating on is. That is essentially what the naturalistic
fallacy is for Moore, and nothing Adams says against calling a mistake a fallacy applies to the
naturalistic fallacy so understood.
Of course, these explanations of how people have gone wrong in coming to think that good is
definable in naturalistic terms should be accepted only if we think that good is indefinable in such
terms, that is, only if they have gone wrong. Moores Open Question Argument is meant to show that
good is indefinable in naturalistic terms.4 Suppose we define a closed question as one a person who
4The clearest statement by Moore of what has come to be called his Open Question Argument is found
in Principia Ethica, Chpt. I, sec. 13, sub-section (1), where he says, ...whatever definition be offered it
may be always asked, with significance, of the complex so defined, whether it is itself good, that is, in
my words asked, Is D, good?, where D is some proposed definition of good.

5
fully understands the question could answer simply on the basis of that understanding. So Is a brother
a male sibling? will be closed, as will, Is a brother a brother? and Is a male sibling a male sibling?
An open question will simply be a question that is not closed, that is, one that cannot be answered by
someone who fully understands the question solely on the basis of that understanding. So Is your
brother your best friend? will be open, as will Are brothers more fun than sisters?
Moores general idea is that if D means the same as F, then Is F, D?, Is D, F?, Is D, D,?, and Is
F, F? will all be closed. Surely the last two questions are closed, and the first two questions must
mean the same as the second if F means D. Just substitute D for F in the first two questions and you
get the third. Moore then considers various candidates for the definition of good and concludes that
the question Is D good? is open in each case. Even if Is D good? is open for these samples, it does
not follow that good cannot be defined in non-evaluative terms, for we might not have included the
correct definition of F in the sample. Still, it might be good inductive evidence that it cannot be defined
naturalistically.
Further, all that can justifiably be claimed in the examples is that the question seems open.
There are two possible explanations of that seeming: (1) it is open and (2) it is not open but seems that
way because we do not fully understand the question. The same alternatives are present when
philosophers try to define any term, e.g., knowledge. The question, Is non-accidentally justified
true belief knowledge? might seem open because it is, or because we do not fully understand the
concept of a non-accidentally justified true belief, or knowledge, or both.
Adams has a very good account of why a correct philosophical analysis of some concept might
produce an open question when that analysis is proposed as a replacement definition (a synonym) for
the relevant term (say, for good). According to Adams, a philosophical analysis is supposed to make
clearer and more precise the concept that some ordinary term designates (pp. 41, 54). But because of

6
that, there will be a certain looseness of fit between the ordinary meaning of a term or expression and
the proposed philosophical meaning of the term (pp. 49, 56, 58, 64). Let the philosophical analysis of
the concept the term F designates be D. Then, of course, Is F, D? and Is D, F? may seem open, but
that is no objection to the philosophical analysis of the concept F designates. We can take Adams to be
making the point that even if Moores Open Question Argument shows that no naturalistic expression
can be synonymous with any value expression, it does not show that there can be no correct
philosophical analyses of value concepts in naturalistic terms. Ethical naturalism can then be
understood to be the view that it is possible to give philosophical analyses of value expressions in
naturalistic terms (using naturalistic concepts), and Moores Open Question Argument does not refute
ethical naturalism so understood.
Now someone might try to show that even that form of ethical naturalism is false by what Ill
call an argument from failure. In epistemology, I think some people have concluded that knows is
unanalyzable because they believe that all the attempts to analyze it have failed. Every attempt has
been defeated by counterexample. Perhaps the best explanation of that failure is that the concept is
unanalyzable. But I do not think that anyone would think that knows is unanalyzable because Is
knowledge D? (or Is D knowledge?) seems open for all the candidates for D that have been tried.
Similarly, someone might conclude that value concepts are unanalyzable because all attempts to define
good or ought have been defeated by counterexample. But that is not the Open Question Argument
for the unanalyzability of good or ought, and it is an argument that can be generalized to show the
impossibility of a philosophical analysis of any concept.
Adams seems to think that it is not impossible for there to be a correct philosophical analysis of
good in itself, provided good in itself does not imply ought to-be. That is because he does not
think that there can be a correct naturalistic analysis of ought to-be. Adams thinks that what counts

7
against classical ethical naturalism is the fact that accepting a value-judgment involves having an
attitude toward that which it is about. He says that it is a linguistic absurdity to say such things as: I
ought to do x. That I acknowledge. But I am completely and unalterably opposed to doing it (p. 75).
According to Adams, classical ethical naturalism does not imply that it is a linguistic absurdity to say,
I ought to do x, but I am completely and unalterably opposed to doing it. Suppose the classical
ethical naturalist is a eudaimonistic act utilitarian. There is nothing linguistically absurd in saying, (P):
Doing A is what maximizes happiness, but I am completely and unalterably opposed to doing A. But
according to classical naturalism that means the same as, (Q): I ought to do A, but I am completely
and unalterably opposed to doing A, since I ought to do A means the same as, Doing A maximizes
happiness and so can be substituted for that expression in (P). Thus the classical ethical naturalist is
committed to denying that (Q) is linguistically absurd because he must hold that it means the same as
(P), which is not linguistically absurd. According to Adams, (Q) is linguistically absurd, so classical
ethical naturalism should be rejected because it implies that something is not linguistically absurd when
clearly it is.
But Adams is mistaken in thinking that there is some linguistic absurdity in saying I ought to A
but I am completely and unalterably opposed to doing it. To see this, assume that someone says, I
ought to treat others fairly, but because he stands to gain a great deal by not treating people fairly then
adds, But I am completely and unalterably opposed to doing it. Further, Satan can say without any
linguistic absurdity, Let evil be my good.5 He knows what he ought to do, and he is opposed to doing
it just because it is what he ought to do. Satan need not mean by evil what is thought to be, or
commonly taken to be, evil. He could be a moral realist and want to know what really is evil so he can
pursue it; thats what attracts him. He is repulsed by what is good and what he ought to do. Satan is
5See the discussion of Simon Blackburns interpretation of this prescription, as made by Satan, in the
review of his Ruling Passions in The Philosophical Quarterly (2001), p. 112.

8
evil, but, contra Adams, what he says is not absurd. Presumably, the classical ethical naturalist could
say that D is good, or what he ought to do, and at the same time meaningfully say that he is completely
and unalterably opposed to doing D-type things. If Adams were right, this would be a linguistic
absurdity. But because it is not, it cannot provide grounds for an objection to classical ethical
naturalism.
Adams summarizes what he thinks is the major flaw in classical ethical naturalism:
...it does not account for the way in which value-language is practicalthe way in which it is
tied to the springs of action. (p. 199; my italics)
Value judgments, or our acceptance of them, need not be accompanied by corresponding motivation. I
might think that treating someone in some way is unfair and wrong, and something I ought not do, and
yet not be motivated at all to treat them fairly. I might pick up money that has fallen from the back of
an armored truck whose driver is hopelessly lost, far out in the country on a lonely road in the middle
of the night, while thinking I ought to return it and yet not be motivated at all to do that. Adams is an
internalist about reasons for action, thinking they must be tied to motivation, but he would have to
defend that internalism further to make good on his claim that classical ethical naturalism does not
account for the way in which value-language is practical. Externalists about reasons for action can
counter that it does account for the way the value-language is practical because an actions being wrong
entails that there is at least some reason not to do it, where there being a reason for you to refrain from
doing something (or even your having such a reason) does not entail that you are motivated by the
consideration which is a reason.6
6Derek Parfit in On What Matters (Oxford University Press, 2011) gives numerous counterexamples to
what he calls subjectivism about reasons for action that involve scenarios where someone is not moved
by awareness of what seem to be obvious reasons for action. See, below, p. 31 ff. Mackies error
theory is founded on the following argument (from Ethics: Inventing Right and Wrong): if any
statement of moral obligation were true, there would be categorical reasons for action, that is, reasons
that do not necessarily motivate a person who is aware of those reasons. Reasons internalism in its

9
Adams thinks that what he calls the naturalistic fallacy argument and the argument about linguistic
absurdity reinforce each other in such a way that they are a serious threat to classical naturalism (p.
76). Taken on its own, the Open Question Argument does not seem to be a threat to classical naturalism,
as even Adams concedes. I have just argued that his linguistic absurdity objection is unsound. It is hard
to see how these two arguments could reinforce each other when neither has merit on its own. Classical
ethical naturalism has not been refuted by Moore or Adams. Of course, that does not mean that
classical ethical naturalism is true, or even that those failures give us reason to think its true. But it is
crucial to Adams argument for nonnaturalism that he eliminate all the rival candidates, including
classical ethical naturalism.
I now turn to Derek Parfits recent argument against what he calls Analytical Naturalism (AN),
for (AN) is equivalent to the semantic part of what Adams calls classical ethical naturalism. Perhaps
Parfit can supply the help Adams needs to eliminate this form of ethical naturalism.
B. Parfits argument against Analytical Naturalism
For Parfit Analytical Naturalism (AN) is the view that normative, or moral, expressions mean
the same as certain naturalistic expressions, where naturalistic expressions refer to natural facts and
natural facts are the sort investigated by those working in the natural or social sciences (OWM II, 265,
305). I will assume that natural facts are empirical facts, that is, facts whose presence or absence can, in
principle, be determined by empirical means, since those are the sorts of facts that the natural and
social sciences investigate. So Analytical Naturalism is identical to the semantic part of what Adams
calls classical naturalism. Parfits argument against (AN) is that if it were true, then what intuitively
seem to be substantive moral claims (or claims about what there is reason to do) would turn out to be

simplest form says that there cannot be reasons to act if awareness of them does not motivate. Because
Mackie thinks reasons internalism is true, he concludes that all statements of moral obligation are false.
If reasons internalism is false, Mackies argument fails.

10
trivial. For instance, a species of Act Utilitarianism which tells us that, all things considered, we
morally ought to maximize happiness, turns out to be the trivial claim that what maximizes happiness
maximizes happiness or what we morally ought, all things considered, do is what we morally ought, all
things considered, do! For Parfit, a claim is substantive just in case we can learn something from it or
people can disagree with it (OWM II, 343, 349, 353 and OWM II, 275 where he gives a similar account
of significant). By trivial he means the opposite of substantive. But on this account, Act
Utilitarianism can be a substantive claim, not a trivial one, because we can learn that this is what
morally obligatory means and other people can disagree with that claim. So Act Utilitarianism might
be a substantive claim even if what maximizes happiness is what maximizes happiness and what is
morally obligatory is morally obligatory are not.7
Parfits mistake seems similar to Moores. Moore thought that if a moral (or practical) term R meant the
same as some naturalistic expression, N, then if Is R, N?, or Is N, R?, is open, then so are Is R,
R? and Is N, N?. Parfit thinks that if the claim that R means N is substantive (non-trivial), then so
are R means the same as R and N means the same as N. Moore and Parfit both overlook the fact
that we can discover that two terms or expressions mean the same, which explains how Is R, N? and
Is N, R? can be open, and R means N non-trivial, even if R is R and N is N are trivial and the
corresponding questions closed.

Bruce Russell
Department of Philosophy
Wayne State University
7It is surprising that Parfit says that claims about the meanings of terms cannot be substantive, that they
must be trivial, for he says, People sometimes fail to understand, not only what other people mean, but
even what they themselves mean (OWM II, 272; cf., 292-3). Because we dont fully understand what
we mean by, say, knows and reasons, we can come to learn what we mean, and others can disagree
with us about whether that really is what we mean. For Parfit, those are the factors that make a claim
substantive.

11
Detroit, MI 48202
[email protected]