Professional Documents
Culture Documents
(2017) 1-23
brill.com/skep
Abstract
Heightened awareness of the origins of our moral judgments pushes many in the
direction of moral skepticism, in the direction of thinking we are unjustified in holding
our moral judgments on a realist understanding of the moral truths. A classic debunk-
ing argument fleshes out this worry: the best explanation of our moral judgments does
not appeal to their truth, so we are unjustified in holding our moral judgments. But
it is unclear how to get from the explanatory premise to the debunking conclusion.
This paper shows how to get from here to there by way of epistemic insensitivity. First,
we reconstruct Richard Joyce’s evolutionary debunking argument from insensitivity.
Second, we raise epistemological difficulties for Joyce’s argument. Third, we develop
and defend a new debunking argument from insensitivity.
Keywords
1 Introduction
1 Discussions of this sort of moral debunking argument took off after Gilbert Harman (1977:
Ch. 1) initially developed an ambitious version of it. He argued, roughly, that since the best
The lion’s share of the discussion has fixated on the merit of the Explanatory
Premise with respect to moral realism, that is, with respect to views on which
the moral truths are objective in the sense of not being a function of human
judgments or attitudes.2 Sturgeon (1984) and others reject the Explanatory
Premise and champion “moral explanations”—explanations that appeal to ob-
jective moral truths—as part of the best complete explanation of our moral
judgments. The classic example: the truth that it is wrong to drench the cat
with gasoline and light it up for kicks partly best explains why we judge it
wrong to do so.3 Harman (1977) and others4 doubt that such realist moral ex-
planations are part of the best explanations of our moral judgments.
But what hinges on this explanatory debate? Some philosophers (Ruse 1986)
argue that the existence of moral facts is at stake—that is, that the Explana-
tory Premise implies a metaphysical conclusion concerning the nonexistence
of moral facts. Other philosophers (Joyce 2001; 2006) argue that the justifica-
tory or knowledge status of our moral judgments is at stake—that is, that the
Explanatory Premise implies a skeptical epistemic conclusion.5 The present pa-
per is interested in the latter epistemic worry. But is it a genuine worry? Walter
Sinnott-Armstrong rightly questions whether the Explanatory Premise could get
us to any kind of interesting skeptical conclusion regarding moral justification
explanation of our moral beliefs does not appeal to the facts that would make them true,
we have no reason for thinking moral truths (if construed as irreducible) exist. His bold
argument triggered an avalanche of literature on “moral explanations.” Recently, Richard
Joyce (2001: Ch. 6; 2007: ch. 6) has developed sophisticated versions of the moral debunking
argument.
2 For a useful review of the explanationist literature, see Majors (2007).
3 Harman (1977). See also Railton (1985) for the case that some moral truths best explain some
non-moral phenomena (e.g., moral truths regarding injustice partly best explain the social
tendency to revolt).
4 See Leiter (2001), Sommers and Rosenberg (2003), and Joyce (2001: Ch. 6; 2007: Ch. 6).
5 For a helpful distinction between metaphysical debunking arguments and epistemic debunk-
ing arguments, see Joyce (2013).
or moral knowledge. Yes, the premise (if true) would rule out explanationist
sources of justification for our moral judgments, that is, justification due to the
fact that the moral truths play a role in explaining our moral judgments. But
the premise would apparently do nothing to rule out coherentist, perceptual,
intuitionist, or reliabilist sources of justification (Sinnott-Armstrong 2006:
44). So how, without independently challenging these sources of justification,
could we get from the Explanatory Premise to the Debunking Conclusion? One
draws a blank.6
The present paper shows that the Explanatory Premise has skeptical impli-
cations by showing how to get from it to the Debunking Conclusion in a few
plausible steps. Good bridge premises, of course, must do the work. The hunt
for good bridge premises proceeds by investigating what appear to be the most
promising ones on offer, namely those offered by Richard Joyce (2001: Ch. 6)
in his evolutionary moral debunking argument from epistemic insensitivity.7
We reconstruct Joyce’s argument, raise epistemological difficulties for it,
and, building upon this discussion, develop and defend a new debunking ar-
gument from insensitivity which secures the inference from the Explanatory
Premise to the Debunking Conclusion.
The paper should prove interesting not only to metaethicists but also to
epistemologists because discussions of epistemic insensitivity usually cen-
ter on whether some sort of Nozickian sensitivity condition is necessary for
knowledge—that is, whether for a belief to be known it must be sensitive in the
sense of having the following modal feature: if p were false, S would not believe
that p. Nozick’s (1981) sensitivity condition for knowledge encounters recalci-
trant counterexamples and objections.8 It will not be defended here. What will
be argued, rather, is that a certain sort of process insensitivity is a sufficient de-
feater of justification. We must distinguish between sensitivity as a necessary
6 Clarke-Doane (forthcoming) considers and cogently critiques one way of bridging the gap.
For brief summary of his critique, see Section 5 of the present paper.
7 For attention to Joyce’s specific argument, see Sinnott-Armstrong (2006: 44). Michael
Ruse (1986: 254) deserves credit for offering the first notable though sketchy moral de-
bunking argument from insensitivity. For some recent attention to debunking arguments
from insensitivity, see Bedke (2014), Clarke-Doane (2012; forthcoming), and Mogensen (2014:
Ch. 3).
8 Nozick’s sensitivity condition for knowledge possesses some defenders (e.g., Becker 2008)
but is currently unpopular and has been subjected to a barrage of counterexamples—for
example, cases of inductive knowledge (Sosa 1999), higher-level knowledge (Vogel 2000), and
the charge that it unacceptably violates closure under known implication. See Luper-Foy
(1993) for a shower of such objections, also see Becker and Black (2012) for more objections
and recent developments.
9 Also see Mogensen (2014: Ch. 3), who gives recent attention to Nozickian insensitivity as
a defeater of justification.
10 Joyce (2001: ch. 6); also see Joyce (2007: Chs. 1–4).
at his specific version of the Explanatory Premise. He argues that it has a skepti-
cal upshot in virtue of implying a sort of epistemic insensitivity.11
Joyce lays out his debunking argument from insensitivity here:
11 Joyce (2001: 163–164) avoids the anticipated “companions in guilt” objection that his argu-
ment applies to our perceptual and mathematical judgments by maintaining that the
best explanation of our perceptual and mathematical judgments appeals to their truth.
12 Joyce specifically targets our categorical moral judgments, such as our judgments of what
we are morally required or obligated to do, or what would be wrong or forbidden. He does
not say anything about our judgments of moral value or virtue, that is, what we think we
have non-categorical moral reason to do or what traits we think are good to possess. Why
not? The explanation well could be that the truth of judgments of moral value or virtue
may be more likely to play a role in their explanation. To avoid cumbersome expressions,
I will drop the “categorical” qualifier and simply speak of moral judgments.
(4) So, the processes that produce our moral judgments are insensitive to the
truth of their content. (1–3)
(5) If the processes that produce our moral judgments are insensitive to
the truth of their content, then we are unjustified in holding our moral
judgments.
(6) So, we are unjustified in holding our moral judgments. (4, 5)
In what follows, we grant the Explanatory Premise, that the best explanation
(nativist, socio-cultural, or otherwise) of our moral judgments does not ap-
peal to their truth. We do so because the goal of the paper is to assess what
follows from it, in particular whether the Debunking Conclusion follows from
it by way of insensitivity. If the Explanatory Premise is true, then we can reach
insensitivity in Joyce’s sense: if the moral truths make no difference whatsoever
to why we hold our moral judgments, then were there no moral truths, the pro-
cesses producing our moral judgments would still produce those judgments.
But can Joycean insensitivity take us to the Debunking Conclusion? In the next
section, it is argued that it cannot, that is, Joycean insensitivity is not a suffi-
cient condition for the defeat of justification.
3 Epistemological Difficulties
13 The principle seems initially attractive because it could explain why judgments are de-
feated in certain cases. For a case Joyce invokes in this vein, see Joyce (2001: 159).
are trees rather than atoms arranged treewise. Nonetheless, it seems our cur-
rent beliefs about ordinary objects are justified (undefeated), even granting
our awareness of their insensitivity.
But the main objection epistemologists can be expected to raise is that
Joyce’s principle implies a nasty general skepticism, given skeptical hypoth-
eses. Processes leading to our beliefs in the denials of skeptical hypotheses—
beliefs that we are not deceived by an evil demon, that we are not brains in a
vat, and so on—are insensitive: were we deceived by an evil demon, we would
still believe we were not. And we are perfectly aware of this insensitivity. Joyce’s
principle would thus imply we are unjustified in continuing to hold such be-
liefs. If that is not implausible enough, given a plausible justification-based
closure principle and given that (say) our being systematically deceived by an
evil demon entails we do not have hands, the further implication is that we are
not justified in believing that we have hands. General skepticism.
Is there some way of avoiding the objection? Mogensen (2014: 119–120)
points out in a related context that process-based or method-based concep-
tions of sensitivity (like Joyce’s) can potentially avoid the skeptical outcome.
On these conceptions, essential reference is made to the process or method
of belief formation. A belief-forming process P is insensitive if the following
subjunctive conditional holds: if the propositional contents of P’s output judg-
ments were false, P would still produce those judgments. To evaluate the condi-
tional, we must hold the process type fixed. So whether the process leading to
the denials of skeptical scenarios is insensitive depends on whether the same
type of process would lead me to the same denial if I were in such skeptical sce-
narios. Whether it would be the same type of process in turn depends on how
processes are individuated. If processes are individuated “internally”—if what
counts is that the processes are similar with respect to their mental states (if
everything seems the same “from the inside”)—then the processes would be
of the same type and insensitivity would follow. But if processes are individu-
ated “externally,” then the processes may be of different types and insensitivity
would not follow.
Since this particular question of process or method individuation is cur-
rently a vexed matter in the literature and no epistemically neutral way of re-
solving it appears ready to hand—internalists and externalists tend to line up
on their expected sides of the issue14—it is unclear whether Joyce’s principle
ultimately avoids the skeptical objection, at least without further argument.
And since the skeptical objection continues to be a central reason why many
epistemologists reject sensitivity conditions, a better debunking argument
from insensitivity would avoid the problem. The argument we develop later in
the paper will do so.15
But another interpretation of Joyce’s argument is available, suggested by
Sinnott-Armstrong (2006: 44).16 On this interpretation, Joyce reasons that our
moral judgments are unjustified because they are “unreliable” and reasons that
they are unreliable because they are insensitive: “On the assumption that my
favored hypothesis about the ‘moral sense’ is correct, it follows that the pro-
cess by which humans form moral judgments is an unreliable one, for they are
disposed to do so regardless of the evidence to which they are exposed” (2001:
162). A key question: with what conception of epistemic reliability is Joyce op-
erating? He offers no explicit specification, so it is an open question. Theoreti-
cally, he has a number of options available.17 The textual evidence suggests we
interpret Joyce as most likely operating with a standard counterfactual con-
ception of process reliability. This is because he takes counterfactual worlds
(“were there no moral truths”) to be relevant to reliability, consistently ascribes
reliability to processes (not just beliefs), and the awareness of process unreli-
ability is widely held to be a defeater of justification (see Joyce 2001: 159–165).
On the basis of these considerations and the fact that he offers no finer-grained
specification, we may presume he has a standard conception of process reli-
ability in mind.
But what, more precisely, is the standard conception of process reliability
that is widely thought to be epistemically significant? Classically construed, a
reliable process is just one that is truth-conducive, one that produces mostly
true judgments. But after a few decades of refinement, the conception of pro-
cess reliability that is fairly widely thought to be epistemically important is a
modal or counterfactual one which maintains, roughly, that our judgments are
reliable if they proceed from processes that produce mostly true judgments
throughout nearby worlds, not just the actual world.18 And this is widely agreed
to be roughly right because distant worlds do not seem relevant to determining
15 See Section 5 for explanation of how the author’s debunking argument avoids the skepti-
cal objection.
16 Sinnott-Armstrong’s interpretation and the interpretation suggested in the present paper
are slightly different. Whereas he sees Joyce as conflating actual world process unreliabil-
ity with insensitivity, the interpretation suggested here sees Joyce as inferring standard
(counterfactual) process unreliability from insensitivity.
17 E.g., reliability as sensitivity, safety, Nozickian tracking, proper functioning, epistemically
virtuous belief-formation, reliable indication, actual world process reliability, and coun-
terfactual process reliability.
18 For a locus classicus of work on counterfactual process reliability, see Goldman (1986).
A process type P is reliable if and only if and to the degree that its in-
stantiations produce and/or sustain mostly true judgments throughout
nearby worlds.
Correlatively:
A process type P is unreliable if and only if and to the degree that its
instantiations do not produce and/or sustain mostly true judgments
throughout nearby worlds.
Importantly, distant worlds are not relevant for determining reliability and
whether a process produces mostly true judgments in the actual world is not
sufficient for determining its reliability.19
With these epistemic considerations in mind, recall that Joyce contends
that a world without moral truths is relevant to the reliability and justification
of our moral judgments. In that world we would arrive at mostly false moral
judgments, and thus, on the present interpretation of Joyce’s argument, our
moral judgments are unjustified because unreliable. But with this move from
insensitivity to unreliability Joyce faces a modal dilemma: he could maintain
that this moral factless world is a nearby world or he could concede that it is a
distant world and yet maintain that it is nevertheless relevant for determining
reliability. The latter option is not cogent because few epistemologists think
that distant worlds are relevant for determining reliability. But the former
and better option does not inspire confidence either because a world that
19 See Henderson and Horgan (2007) and Becker (2008) on the epistemic importance of
actual-plus-nearby-worlds process reliability.
20 Bedke (2014: 121) suggests such a world is not clearly a distant world either. Regardless,
Joyce needs to motivate the claim that such a world is nearby in order for his inference
from insensitivity to unreliability to succeed. If we should be agnostic about the matter,
Joyce’s argument does not go through.
The argument’s insensitivity claim is this: for each particular moral belief-
forming process P, were most of the propositional contents of its output judg-
ments false, P would still produce those (false) judgments. On the basis of this
insensitivity claim and a few other plausible bridge premises, we can reason-
ably infer the process unreliability of our moral belief-forming processes. Here
goes:
P2: If the Explanatory Premise is true, then the processes21 producing our
moral judgments are highly individually insensitive.
—More precisely: if the Explanatory Premise is true, then for each par-
ticular moral belief-forming process P, were most of the propositional
contents of P’s output judgments false, P would still produce those (false)
judgments. On the standard analysis of subjunctive conditionals,22 the
claim is this: if the Explanatory Premise is true, then the nearest worlds
where most of the propositional contents of P’s output judgments are
false are worlds where P still produces those (false) judgments.
P3: So, the processes producing our moral judgments are highly individu-
ally insensitive. (P1–P2).
P4: There are many different processes producing our moral judgments.
P5: If there are many different processes producing our moral judgments,
then, for each particular moral belief-forming process P, the nearest
worlds where most of the propositional contents of P’s output judgments
are false are worlds without a relatively small subset of the moral truths but
otherwise as similar as possible to the actual world.
P6: Worlds without a relatively small subset of the moral truths but other-
wise as similar as possible to the actual world are nearby worlds.
P7: So, for each particular moral belief-forming process P, the nearest
worlds where most of the propositional contents of P’s output judgments
are false are nearby worlds. (P4–P6).
P8: So, for each particular moral belief-forming process P, the nearest
worlds where most of the propositional contents of P’s output judgments
are false are both (i) worlds where P still produces those (false) judg-
ments and (ii) worlds that are nearby. (P3, P7).
P9: A belief-forming process is unreliable if it does not produce mostly
true judgments throughout nearby worlds.
C: So, each of our moral belief-forming processes is unreliable. (P8, P9).
A non-moral example can illustrate the main steps from the Explanatory Premise
to insensitivity, to unreliability, and in turn to the defeat of justification. Sup-
pose for a few days I have a regular experience as of people talking nearby out
of view outside my office. Regular spells of this chatter happen once in a long
while. The auditory experiences would ordinarily confer prima facie justifica-
tion upon my beliefs that there are people talking nearby. But if I subsequently
21 More precisely, processes refer to process types the instantiations of which produce our
moral judgments.
22 For a locus classicus, see Pollock (1976).
discover that a highly probable side-effect of the new medication I just started
taking a few days ago is regular auditory hallucinations—hearing voices
in particular—then this consideration (if I have no other evidence) would
suffice to defeat the justification of my auditory-based beliefs. The best ex-
planation of my beliefs that people are talking nearby (medically-induced
hallucination) does not appeal to their truth. If the truth of my beliefs makes
no difference to my forming them, then the hallucinatory process leading me
to them is highly individually insensitive: were most of the output beliefs of the
hallucinatory process false, the process would still drive me to them. Worlds
where most of the output beliefs of the hallucinatory process are false would
be nearby worlds (e.g., where there are no people talking nearby out of view),
hence the hallucinatory process is unreliable. Awareness of the unreliability
of my auditory-based beliefs would in turn constitute a sufficient defeater of
their justification.
As for P1, recall that we are granting the Explanatory Premise (to assess
what follows from it). As for P2, the inference from the Explanatory Premise to
the highly individually insensitive character of our moral belief-forming pro-
cesses is cogent. For if the best complete explanation of our moral judgments
does not appeal to their truth—if the moral truths make no difference what-
soever to why we hold our moral judgments—then the processes producing
our moral judgments are highly individually insensitive in the sense that for
each moral-belief forming process P, were we to eliminate the moral truths
corresponding to most of P’s output judgments, P would nevertheless still pro-
duce those (false) judgments. A natural why-question generated by this move
concerns the “most” operator. Why most rather than all, given that all follows
just as well? Alternatively put: given that the completely insensitive character
of our moral belief-forming processes follows from the Explanatory Premise,
why does the argument draw the weaker implication of their highly insensitive
character? This move is made for two reasons. First, the weaker implication is
all the argument needs: if in nearby worlds, most of process P’s output judg-
ments are false, that is sufficient for the unreliability of P on the standard un-
derstanding of process unreliability characterized earlier. Second, the weaker
implication preempts an anticipated objection from the alleged necessity of
moral truths, as we shall see in the next section.
The Debunking Argument from Insensitivity, given the species of insensitivity
to which it appeals, is not subject to the modal dilemma that impugned Joyce’s
inference from insensitivity to unreliability. The worlds at which each process,
respectively, do not lead us to mostly true judgments are nearby worlds where,
respectively, a relatively small subset of the moral truths does not obtain, rather
than a distant world where no moral truths whatsoever obtain (we will see
in the next section that the relevant small subset does not include any
necessary moral truths). This would not be so if there were only a few process
types (e.g., innate biases, testimony) whose instantiations produced our moral
judgments, since a world where a member of such few process types does not
lead us to mostly true judgments could very well be a distant world in virtue
of not containing a relatively large subset of the moral truths. But as P4 ob-
serves, there are plausibly many different process types behind our moral
judgments.
Support for P4 is given by reliabilists and other epistemologists in their discus-
sions of the “generality problem”—broadly, the problem of how to individuate
belief-forming process types—who almost invariably favor rather narrow
specifications of processes, even though they often do not agree on how nar-
row these specifications should be.23 This means they typically favor many
process types rather than a few. This suggests P4 receives support from the
epistemological literature.
But why think moral belief-forming processes are many rather than few?
For example, if one were to combine a certain sort of moral intuitionism (e.g.,
Huemer 2005) with skepticism about moral testimony as a source of moral
knowledge (e.g., McGrath 2009), then there would only be one belief-forming
process yielding moral knowledge, namely intuiting or adequately under-
standing moral propositions such as pain is prima facie bad.24
For the sake of argument, grant moral intuitionism and skepticism about
moral testimony. Is there then just one belief-forming process yielding moral
knowledge? Moral intuitionists invariably accept other processes, namely rea-
soning processes that come in distinct varieties, not all of which are equally
reliable. This makes sense because were intuitionists to restrict sources of mor-
al knowledge to intuiting alone, the result would be a deeply impoverished
and revisionary conception of the scope of our moral knowledge. Much of our
moral knowledge extends beyond intuited (general or particular) truths and
derives from various types of reasoning. Examples of moral reasoning include:
deduction (top-down reasoning), induction (bottom-up reasoning), analogi-
cal reasoning, inference to the best explanation, coherence-based reasoning
(e.g., the method of reflective equilibrium), and so on. Psychologists investi-
gate such reasoning processes, distinguish between them, and seek to explain
how they work, while philosophers typically focus on how they should work,
when and why they are reliable (or not), and so on. Progress has arguably been
made on both fronts in modeling and understanding the different reasoning
processes and their different normative standards, which suggests that there
are distinct causally efficacious processes in operation, not just one general
process of moral reasoning.
Rational intuition and moral reasoning, moreover, do not exhaust the pro-
cesses yielding moral knowledge. Certain emotions (e.g., sympathy, empathy)
and innate biases (e.g., reciprocity bias, fairness bias, conformist bias) contribute
to our moral judgments, and arguably can do so in a reliability-contributing
way under certain conditions. Empirical theorists in the psychological and
evolutionary sciences distinguish between various emotions and innate biases,
arguing cogently that these contributors play various roles in the origins of our
moral judgments.25 The credibility of their explanations strongly suggests a
plethora of distinct emotion-based and bias-based processes. Of course, some
emotions and innate biases (e.g., in-group biases) are reliability-detracting,
but the point is that others often seem reliability-contributing and so could
be distinct contributors to moral knowledge. Moreover, emotions and biases
often combine with each other and with various kinds of reasoning processes,
yielding many distinct efficacious processes in turn. All this suggests there are
many distinct processes behind our moral judgments.
Furthermore, if testimony is a source of moral knowledge, then the num-
ber of processes multiplies because it is plausible to individuate testimonial
processes rather narrowly, even if we have no precise way of picking out the
correct description of the process. Suppose my wiser virtuous mother tells me
that to be hospitable we should let friends of the family stay at our home over-
night, and suppose I come to believe this partly on the basis of her testimony.
What is the causally efficacious testimonial process leading to my belief? Is it
correctly described as believing whatever a human being tells me? That is not
the process because, for one thing, I would probably not have formed the be-
lief had some random stranger testified to me instead. The efficacious process
leading to my belief plausibly has to do with the agent testifying to me and
the subject matter the agent is testifying about. So it is with testimony in other
domains. The fact that there are many different types of agents and moral sub-
ject matters suggests there are many different types of testimonial processes of
moral belief-formation.26
So then, given there are many process types behind our moral judgments,
the respective worlds where each particular process produces mostly false
5 Four Objections
First Objection: Objection from Moral Necessity. The first objection to the De-
bunking Argument from Insensitivity targets P6, the claim that a world without
a relatively small subset of the moral truths is a nearby world. The objection
presses that since the moral truths are metaphysically (or logically) necessary
(e.g., it is wrong to torture someone just for kicks), worlds where any of them
do not obtain are impossible worlds and you cannot get more distant than that.
Accordingly, P6 is mistaken.
Response. The moral truths represented by the bulk of our actual moral
judgments—especially the more particular moral truths represented by our
everyday moral judgments—are contingent upon empirical facts. For exam-
ple, the moral truth that I wrongfully lied to someone depends on empirical
facts about, say, my deceptive intention and/or the consequences of my ly-
ing. Consequentialists, deontologists, and virtue ethicists highlight the depen-
dence of moral truths on various empirical facts. And since empirical facts
are contingent, the bulk of the moral truths represented by our actual moral
judgments are contingent. Moreover, most moral truths appear to be fairly
contingent rather than modally robust—depending according to normative
tastes on, say, consequences, intentions, what a virtuous person would do, or
other fairly contingent considerations. That said, there may be some necessary
moral truths corresponding to some of our most general moral judgments or
standards that do not depend on empirical facts—for example, it is wrong
to torture someone just for kicks. But granting this would not impugn P6 be-
cause the worlds in question are those where most, not all, of the moral truths
corresponding to the output judgments of a particular moral belief forming
process P do not obtain. Even if P produces some general moral judgments that
correspond to necessary (or modally robust) moral truths, these judgments
will be comparatively fewer because the bulk of our moral judgments are more
particular. So even if we grant that some necessary moral facts obtain in the
counterfactual worlds in question—since they obtain in all possible worlds—
that is perfectly compatible with the nearness of such worlds. They are nearby
worlds because they are worlds where a relatively small subset of the non-
necessary moral truths is eliminated.27
Second Objection: Objection from Moral Supervenience. The second objection
maintains that the truth of moral supervenience challenges P6, the premise
that worlds without a relatively small subset of the (non-necessary) moral
truths are nearby worlds. The objection contends that P6 is only plausible on
the assumption that the moral truths are metaphysically independent of their
27 The fact that only contingent moral truths feature in the subjunctive conditional defining
high individual insensitivity implies that the conditional is not a counterpossible, that is,
not a conditional with an impossible or necessarily false antecedent. If it were a counter-
possible, then according to the standard semantics insensitivity would be trivially true (as
would sensitivity!), which would pose a problem for P3 of the debunking argument. But it
is not a counterpossible, so no such problem from the standard semantics arises.
underlying natural facts, such that the removal of various subsets of the moral
truths would not entail the removal of some underlying natural facts as well.
That is, P6 is only plausible on the assumption that moral supervenience does
not obtain. For if moral supervenience does obtain—and it is widely held to
obtain in some form or other—this casts doubt on the claim that worlds with-
out a relatively small subset of the (non-necessary) moral truths are nearby,
since they are not just worlds where a subset of moral truths do not obtain but
rather also worlds where the natural facts which underlie these moral truths
are somewhat different.
Response. Consider the sort of moral supervenience at issue:
may not work against anti-skeptical metaethical views that accept Strong
Supervenience. Would this limitation in scope diminish the argument’s signifi-
cance? The following reasons suggest not.
First, observe that though Weak Supervenience is widely held, it is not at
all clear that Strong Supervenience is widely accepted or should be accept-
ed in the first place. There is a notable dearth of arguments for it. The most
prominent sort of argument for Strong Supervenience moves from the alleged
inconceivability of moral differences without natural differences to the impos-
sibility of such differences.28 But such arguments have a difficult time vindicat-
ing the alleged “inconceivability” and the inference from inconceivability to
impossibility.29
Second, others have pointed out that Strong Supervenience demands an ex-
planation if we are to take it seriously (it is not plausibly a brute fact), since
it amounts to the bold postulation that a metaphysical relation of necessity
holds between two types of properties. Given that plausible explanations of
Strong Supervenience are not yet clearly available,30 the commitment to Strong
Supervenience appears to be more of a theoretical cost than a metaethical
nonnegotiable.31
Finally, there is reason for thinking the argument goes through even if we
grant Strong Supervenience. For it is plausible to suppose that the bulk of the
moral truths are fairly contingent in that the elimination of small subsets of
such truths would not entail major changes in the natural facts. So though
Strong Supervenience entails there would be some natural differences if some
moral truths were eliminated, it would not clearly entail that the relevant
worlds would be beyond the epistemically relevant range of nearby worlds,
especially if the subsets of moral truths removed are relatively small and the
moral truths removed are fairly contingent.
Third Objection: General Skepticism. The worry raised earlier for Joyce’s argu-
ment was that his underlying epistemic principle implies general skepticism,
given skeptical hypotheses. Does the same sort of problem apply to the new
debunking argument? To be clear: the problem is not that the argument does
not answer skepticism—how to answer or respond to skepticism is a major
task and the argument is not intended to take it on—rather, the problem is
that the argument’s principles imply general skepticism. The specific objec-
tion in this context would be that the processes producing the beliefs that we
are not brains in a vat, not demonically deceived, and so on, are insensitive
and thus, according to the argument, unreliable. But if we are unjustified in
believing the denial of skeptical hypotheses, then we are unjustified in believ-
ing (say) that we have hands, given that justification is plausibly closed under
competent deduction.
The reason why the objection does not apply to the Debunking Argument
from Insensitivity is that what legitimates the inference to process unreli-
ability is not insensitivity by itself, but insensitivity plus the claim of nearby
worlds (see P8 and P9 of the argument). And worlds where skeptical scenari-
os obtain—worlds where we are brains in a vat—are distant from the actual
world on standard metrics of distance (for example, our ordinary explanatory
scheme would have to change radically to explain what is going on in a world
where we are envatted brains). So even if the processes leading to the denials
of skeptical hypotheses are insensitive, the argument’s epistemic principles do
not imply those denials are unjustified and hence do not imply any sort of
general skepticism.
Fourth Objection: Conflation of Debunking Challenges. Clarke-Doane (forth-
coming) suggests that some debunkers confuse or conflate two distinct de-
bunking challenges, specifically the “justificatory challenge” (the challenge to
justify our moral judgments given the Explanatory Premise) with the “reliability
challenge” (the challenge to explain the reliability of our moral judgments).
He interprets debunkers like Joyce and Sharon Street (2006) as relying on the
assumption that the Explanatory Premise implies realists cannot explain the
reliability of our moral judgments. From this assumption, they reason that our
moral judgments are debunked but not our mathematical judgments, given
that the Explanatory Premise does not apply to our mathematical judgments.
But the assumption of the debunkers is false or unsupported, Clarke-Doane
cogently argues. Why then have debunkers accepted the assumption? He sug-
gests they have done so because they have mistakenly conflated the justifica-
tory challenge with the reliability challenge. Does the debunking argument
of the present paper also rely on this dubious assumption and conflate two
distinct debunking challenges? No. The argument reasons to unreliability by
way of insensitivity, not by way of any premise concerning whether realists
can explain reliability or not. In Clarke-Doane’s terms, the argument develops
the justificatory challenge, not the reliability challenge, which shows there is
at least one debunking argument from the Explanatory Premise that does not
conflate the two.
6 Concluding Remarks
If the moral truths play no role in the best explanation of our moral
judgments—if the Explanatory Premise is true—what follows? The move
from the Explanatory Premise to the Debunking Conclusion does not inspire
confidence since it is opaque how to get from here to there. Extant appeals
to insensitivity have not helped bridge the gap. But the Debunking Argument
from Insensitivity developed in this paper shows that the Explanatory Premise
implies a certain sort of insensitivity that, in combination with a few plausible
bridge premises, leads to the Debunking Conclusion. Much then hangs on the
Explanatory Premise.32
References
32 My gratitude for comments and discussion of earlier versions of this paper is due to Kelly
Becker, Brian Besong, Tim Black, Patrick Epley, Justin Horn, Dan Korman, Steven Luper,
Walter Sinnott-Armstrong, David Wong, and to audiences of the American Philosophi-
cal Association, Society of Exact Philosophy, University of Oklahoma, and Central States
Philosophical Association. Kind thanks also to an anonymous reviewer of the present
journal.
Goldberg, S. (2012). “Sensitivity from Others.” In Becker and Black (2012), 43–65.
Goldman, A. (1986). Epistemology and Cognition. Cambridge, ma: Harvard University
Press.
Harman, G. (1977). The Nature of Morality. New York: Oxford University Press.
Henderson, D., and T. Horgan. (2007). “Some Ins and Outs of Transglobal Reliabilism.”
In Goldberg (2007), 100–130.
Hieke, A., and H. Leitbeg (eds.), (2009). Reduction, Abstraction, and Analysis. Ontos
Verlag.
Hills, A. (2009). “Supervenience and Moral Realism.” In Hieke and Leitbeg (2009),
163–177.
Huemer, M. (2005). Ethical Intuitionism. New York: Palgrave Macmillan.
Joyce, R. (2001). The Myth of Morality. Cambridge: Cambridge University Press.
Joyce, R. (2007). The Evolution of Morality, Cambridge, ma: mit Press.
Joyce, R. (2013). “The Evolutionary Debunking of Morality.” In Feinberg and Shafer-
Landau (2013), 527–534.
Leiter, B. (2001). “Moral Facts and Best Explanations,” Social Philosophy and Policy 18:
79–101.
Luper-Foy, S. (ed.). (1987). The Possibility of Knowledge: Nozick and His Critics. Totowa,
nj: Rowman & Littlefield.
Mackie, J.L. (1977). Ethics: Inventing Right and Wrong. New York: Penguin.
Majors, B. (2007). “Moral Explanation,” Philosophy Compass 2: 1–15.
McGrath, S. (2009). “The Puzzle of Pure Moral Deference,” Philosophical Perspectives
23: 321–344.
Mogensen, A. (2014). “Evolutionary Debunking Arguments in Ethics.” D. Phil Thesis.
University of Oxford.
Nichols, S. (2004). Sentimental Rules: On the Natural Foundations of Moral Judgment.
Oxford: Oxford University Press.
Nozick, R. (1981). Philosophical Explanations. Cambridge, ma: Harvard University Press.
Pollock, J. (1976). Subjunctive Reasoning. Reidel: Dordrecht.
Railton, P. (1985). “Moral Realism,” Philosophical Review 95: 163–207.
Ridge, M. (2007). “Anti-reductionism and Supervenience,” Journal of Moral Philosophy
4: 330–348.
Ridge, M. (2008). “Moral Non-Naturalism.” In E.N. Zalta (ed.), Stanford Encyclopedia of
Philosophy (Fall 2014 Edition).
Ruse, M. (1986). Taking Darwin Seriously. Oxford: Basil Blackwell.
Shafer-Landau, R. (2003). Moral Realism: A Defence. Oxford: Oxford University Press.
Shafer-Landau, R. (ed.). (2014). Oxford Studies in Metaethics: Volume 9. Clarendon:
Oxford University Press.
Sinnott-Armstrong, W. (2006). Moral Skepticisms. New York: Oxford Univ. Press.
Sommers, T., and A. Rosenberg. (2003). “Darwin’s Nihilistic Idea: Evolution and the
Meaninglessness of Life,” Biology and Philosophy 18: 653–668.
Sperber, D. (1996). Explaining Culture: A Naturalistic Approach. Cambridge, ma:
Blackwell.
Street, S. (2006). “A Darwinian Dilemma for Realist Theories of Value,” Philosophical
Studies 127: 109–166.
Sturgeon, Nicholas. (1984). “Moral Explanations.” In Copp and Zimmerman (1984),
49–78.