“Moral Skepticism” names a diverse collection of viewsthat deny or raise doubts about common beliefs in various roles ofreason in morality. Different versions of moral skepticism deny ordoubt moral knowledge, justified moral belief, moral truth, moralfacts or properties, and reasons to be moral.

Despite this diversity among the views that get labeled “moralskepticism”, many people have very strong feelings about moralskepticism in general. One large group finds moral skepticism obvious,because they do not see how anyone could have real knowledge of themoral status of anything or how moral facts could fit into a physicalworld. Others see moral skepticism as so absurd that any moral theorycan be refuted merely by showing that it leads to moral skepticism.Don’t you know, they ask, that slavery is morally wrong? Orterrorism? Or child abuse? Skeptics who deny that we have reason tobelieve or obey these moral judgments are seen as misguided anddangerous. The stridency and ease of these charges suggests mutualmisunderstanding, so we need to be more charitable and moreprecise.

• 1. Varieties of Moral Skepticism
• 2. A Presumption against Moral Skepticism?
• 3. Arguments for Moral Skepticism
  • 3.1 Moral Disagreements
  • 3.2 Moral Explanations
  • 3.3 A Regress
  • 3.4 Skeptical Hypotheses
  • 3.5 Relations Among the Arguments
• 4. Pyrrhonian Moral Skepticism
• Bibliography
• Academic Tools
• Other Internet Resources
• Related Entries

1. Varieties of Moral Skepticism

Moral skeptics differ in many ways (cf. Machuca 2017), but they sharea common core that makes them all moral skeptics. What makes moralskepticismmoral is that it concerns morality rather thanother topics. Moral skeptics might go on to be skeptics about theexternal world or about other minds or about induction or about allbeliefs or about all norms or normative beliefs, but these otherskepticisms are not entailed by moral skepticism alone.

What makes moral skepticsskeptics is that they raise doubtsabout common beliefs. Moral skeptics then differ in the kinds ofcommon beliefs that they doubt. They might, for example, doubtcategorical or absolute moral beliefs without doubting weaker kinds ofmoral beliefs. Moral skeptics also differ in the kinds of doubts thatthey raise. Since general skepticism is an epistemological view aboutthe limits of knowledge or justified belief, the most central versionof moral skepticism is the one that raises doubts about moralknowledge or justified moral belief.

There are two main traditions in epistemological skepticism. Onetradition makes the claim that nobody ever knows or can know anything.This claim is sometimes namedCartesian skepticism (althoughDescartes argued against it) orAcademic skepticism (despiteother interpretations of skeptics in the ancient Academy). For lack ofa better description, we can call itdogmatic skepticism,because such skeptics dogmatically assert a universal claim. Incontrast, no such claim is made byPyrrhonian skeptics. Theyalso don’t deny any claim like this. They have so much doubtthat they refrain from taking any position one way or the other onwhether anyone does or does not or can or cannot know anything.

Moral skepticism comes in two corresponding varieties.Pyrrhonianskeptics about moral knowledge refuse to assert or endorse theclaim that some people sometimes know that some substantive moralbelief is true. They doubt that moral knowledge is possible. Still,they do not go on to make the opposite claim that moral knowledge isimpossible. They doubt that, too. Their doubts are so extreme thatthey do not make any claim one way or the other about the actuality orpossibility of moral knowledge. Parallel views can be adoptedregarding justified moral belief.Pyrrhonian skeptics aboutjustified moral belief suspend or withhold belief about theactuality or possibility of any justified moral belief.

In contrast, dogmatic moral skeptics make definite claims about theepistemic status of moral beliefs:

Dogmatic skepticism about moral knowledge is the claim thatnobody ever knows that any substantive moral belief is true (cf.Butchvarov 1989, 2).

Some moral skeptics add this related claim:

Dogmatic skepticism about justified moral belief is the claimthat nobody is ever justified in holding any substantive moralbelief.

(The relevant way of being justified is specified in Sinnott-Armstrong2006, chap. 4.) These two claims and Pyrrhonian moral skepticism allfall under the general heading ofepistemological moralskepticism.

The relation between these two claims depends on the nature ofknowledge. If knowledge implies justified belief, as is traditionallysupposed, then skepticism about justified moral belief impliesskepticism about moral knowledge. However, even if knowledge doesrequire justified belief, it does not requireonly justifiedbelief, so skepticism about moral knowledge does not imply skepticismabout justified moral belief.

One reason is that knowledge implies truth, but justified belief doesnot. Thus, if moral beliefs cannot be true, they can never be known tobe true, but they still might be justified in some way that isindependent of truth. As a result, skepticism about moral knowledge isimplied, but skepticism about justified moral belief is not implied,by yet another form of moral skepticism:

Skepticism about moral truth is the claim that no substantivemoral belief is true.

This claim is usually based on one of three more specific claims:

These last three kinds of moral skepticism are not epistemological,for they are not directly about knowledge or justification. Instead,they are directly about truth, so they are usually based on views ofmoral language or metaphysics.

Some philosophers of language argue that sentences like“Cheating is morally wrong” are neither true nor false,because they resemble pure expressions of emotion (such as “BooKnicks”) or prescriptions for action (such as “GoCeltics”). Such expressions and prescriptions are kinds of thingthat cannot be either true or false. Thus, if these analogies hold inall relevant respects, then substantive moral beliefs are also not theright kind of thing to be either true or false. They are not apt forevaluation in terms of truth. For this reason, such linguistictheories are often taken to imply skepticism about moraltruth-aptness. Views of this general sort are defended by Ayer (1952),Stevenson (1944), Hare (1981), Gibbard (1990; cf. 2003), and Blackburn(1993), although recent versions often allow some minimal kind ofmoral truth while denying that moral beliefs can be true or false inthe same robust way as factual beliefs. For more detailed discussionsof these views, see the entry onmoral cognitivism vs. non-cognitivism.

Such views are often described asnon-cognitivism. That labelis misleading, because etymology suggests that cognitivism is aboutcognition, which is knowledge. Since knowledge implies truth,skepticism about moral truth-aptness has implications for moralknowledge, but it is about truth-aptness and not directly about moralknowledge.

Whatever you call it, skepticism about moral truth-aptness runs intoseveral problems. If moral assertions have no truth-value, then it ishard to see how they can fit into truth-functional contexts, such asnegation, disjunction, and conditionals (Sinnott-Armstrong 2000). Suchcontexts are also unassertive, so they do not express the sameemotions or prescriptions as when moral claims are asserted. Indeed,no particular emotion or prescription seems to be expressed whensomeone says, “Eating meat is not morally wrong” (cf.Schroeder 2010), since a person who asserts this sincerely might becompletely indifferent to whether anyone eats meat and need not getangry at vegetarians who critize meat eaters. Expressivists andprescriptivists respond to such objections, but their responses remaincontroversial (cf. Sinnott-Armstrong 2006, chap. 2).

Many moral theorists conclude that moral assertions express not onlyemotions or prescriptions but alsobeliefs. In particular,they express beliefs that certain acts, institutions, or people havecertain moral properties (such as moral rightness or wrongness) orbeliefs in moral facts (such as the fact that a certain act is morallyright or wrong). This non-skeptical linguistic analysis still does notshow that such moral claims can be true, since assertions can expressbeliefs that are false or neither true nor false. Indeed, allsubstantive moral assertions and beliefs are false (or neither truenor false) if they claim (or semantically presuppose) that moral factsor properties exist, but no moral facts or properties really do exist,as claimed by this metaphysical thesis:

Skepticism about moral reality is the claim that no moralfacts or properties exist.

Skepticism about moral reality is, thus, a reason for skepticism withmoral falsehood, as developed by Mackie (1977), or skepticism aboutmoral truth-value, as developed by Joyce (2001). Opponents of sucherror theories often object that some moral beliefs must betrue because some moral beliefs deny the truth of other moral beliefs.However, error theorists can allow a negative moral belief to be true,but only if it merely denies the truth of the corresponding positivemoral belief and does not make any other positive moral claim. Forexample, error theorists can endorse “Eating meat isnot morally wrong” if they mean only that “Eatingmeatis morally wrong” is not true, and they do not goon to claim anything positive like “Eating meat ought to bepermitted” or “Vegetarians should not criticize orinterfere with meat eaters.” If such denials of moral beliefsare not substantive moral beliefs (as denials of astrological beliefsare not astrology), then error theorists can maintain that allsubstantive moral beliefs are false or neither true nor false. Formore detailed discussions of these views, see the entry onmoral anti-realism.

Error theorists and skeptics about moral truth-aptness disagree aboutthe content of moral assertions, but they still agree that nosubstantive moral claim or belief is true, so they are both skepticsabout moral truth. None of these skeptical theses is implied by eitherskepticism about moral knowledge or skepticism about justified moralbelief. Some moral claims might be true, even if we cannot know orhave justified beliefs about which ones are true. However, a converseimplication seems to hold: If knowledge implies truth, and if moralclaims are never true, then there is no knowledge of what is moral orimmoral (assuming that skeptics deny the same kind of truth thatknowledge requires). Nonetheless, since the implication holds in onlyone direction, skepticism about moral truth is still distinct from allkinds of epistemological moral skepticism.

Yet another non-epistemological form of moral skepticism answers thequestion “Why be moral?” This question is used to raisemany different issues. Almost everyone admits that there issometimes some kind of reason to be moral. However, manyphilosophers deny various universal claims, including the claims thatthere isalways some reason to be moral, that there is alwaysa distinctivelymoral reason to be moral, and/or that thereis alwaysenough reason to make it irrational not to be moralor at least not irrational to be moral. These claims and their denialsdepend on specific views about which facts provide reasons (are allreasons about desires or self-interest?), which reasons are moral innature (are all moral reasons impartial?), and which reasons overridewhich other reasons (are moral reasons ever overridden?). Thesedistinct denials can be seen as separate forms ofpractical moralskepticism, which are discussed in more detail in the followingsupplementary document:

Supplement on Practical Moral Skepticism

Practical moral skepticism resembles epistemological moral skepticismin that both kinds of skepticism deny a role to reasons in morality.However, epistemological moral skepticism is about reasons forbelief, whereas practical moral skepticism is about reasonsforaction. Moreover, practical moral skeptics usually denythat there isalways enough reason for moral action, whereasepistemological moral skeptics usually deny that there isever an adequate reason for moral belief. Consequently,practical moral skepticism does not imply epistemological moralskepticism. Some moral theorists (such as Kant on someinterpretations) do assume that a reason to believe that an act isimmoral cannot be adequate unless it also provides a reason not to dothat act. However, even if the two kinds of reasons are related inthis way, they are still distinct, so practical moral skepticism mustnot be confused with epistemological moral skepticism.

Overall, then, we need to distinguish the following kinds ofepistemological moral skepticism:

We also need to distinguish these epistemological moral skepticismsfrom several non-epistemological kinds of moral skepticism:

These kinds of moral skepticism can be diagrammed as follows:

Figure 1.

Skepticism about justified moral belief will be the primary topic forthe rest of this entry, and I will refer to it henceforth simply asmoral skepticism.

2. A Presumption Against Moral Skepticism?

Opponents often accuse moral skepticism of leading to immorality.However, skeptics about justified moral belief can act well and benice people. They need not be any less motivated to be moral, nor needthey have (or believe in) any less reason to be moral thannon-skeptics have (or believe in). Moral skeptics can hold substantivemoral beliefs just as strongly as non-skeptics. Their substantivemoral beliefs can be common and plausible ones. Moral skeptics caneven believe that their moral beliefs are true by virtue ofcorresponding to an independent moral reality. All that moral skepticsneed to deny or doubt in order to be moral skeptics is that their (oranyone’s) moral beliefs are justified. This meta-ethicalposition about the epistemic status of moral beliefs need not trickledown and infect anyone’s substantive moral beliefs oractions.

Critics still argue that moral skepticism conflicts with common sense.Most people think that they are justified in holding many moralbeliefs, such as that it is morally wrong to beat your opponentsenseless with a baseball bat just because she beat you in a baseballgame. People also claim moral knowledge, such as when a neighbor says,“I know that it is wrong for him to spank his daughter so hard,but I don’t know what I should do about it.” Moralskepticism conflicts with these common ways of talking and thinking,so moral skeptics seem to owe us some argument for their controversialclaim.

Dogmatic moral skepticism is, moreover, a universal and abstruseclaim. It is the claim that all moral beliefs have a certain epistemicstatus. Normally one should not make such a strong claim without somereason. One should not, for example, claim that all biological beliefsare unjustified unless one has some reason for this claim.Analogously, it seems that one should not claim that all moral beliefsare unjustified unless one has some positive argument. Thus, its form,like its conflict with common sense, seems to create a presumptionagainst moral skepticism.

Moral skeptics, in response, sometimes try to shift the burden ofproof to their opponents. Anyone who makes the positive moral claimthat sodomy is morally wrong seems to need some reason for that claim,just as someone who claims that there is life on Mars seems to needevidence for that claim. If the presumption is always against thosewho make positive moral claims, then it is opponents of moralskepticism who must carry the burden of proof. Or, at least, moralskeptics can deny that the burden of proof is on moral skeptics. Thenmoral skeptics may criticize any moral belief or theory withoutneeding to offer any positive argument for moral skepticism, and theiropponents need to take moral skepticism seriously enough to argueagainst it (cf. Copp 1991).

This controversy about burden of proof might be resolved bydistinguishing dogmatic moral skepticism from Pyrrhonian moralskepticism. Dogmatic skeptics about justified moral belief make auniversal claim that conflicts with common sense, so they seem to havethe burden of arguing for their claim. In contrast, Pyrrhonian moralskeptics neither make nor deny any claim about the epistemic status ofany moral belief. They simply raise doubts about whether moral beliefsare ever justified. This difference suggests that Pyrrhonian moralskeptics do not take on any or as much burden of proof as do dogmaticskeptics about justified moral belief.

3. Arguments for Moral Skepticism

Whether or not they need to, moral skeptics do offer a variety ofarguments for their position. Here I will focus on arguments fordogmatic skepticism about justified moral belief, but essentially thesame arguments could be formulated to support dogmatic skepticismabout moral knowledge. I will return later to Pyrrhonian moralskepticism in section 4. Also, although here I will sometimesformulate these arguments in terms of moral truth for simplicity, theycould be restated in ways more congenial to skeptics about moraltruth-aptness.

3.1 Moral Disagreements

The simplest and most common argument for moral skepticism is based onobserved facts: Smart and well-meaning people disagree about the moralpermissibility of abortion, affirmative action, capital punishment,active euthanasia, nuclear deterrence, welfare reform, civil rights,and so on. Many observers generalize to the conclusion that no moralclaim is or would be accepted by everyone.

However, all of these disagreements together still do not exclude thepossibility of agreement on other moral beliefs. Maybe nobody deniesthat it is morally wrong to torture babies just to get sexualpleasure. Moreover, even if no moral belief is immune to disagreement,the fact that some people disagree with me does not prove that I amunjustified in holding my moral belief. I might be able to show themthat I am right, or they might agree with me under idealcircumstances, where they are better informed, more thoughtful, lesspartial, and so on. Moral disagreements that are resolvable do notsupport moral skepticism, so any argument for moral skepticism frommoral disagreement must show that moral disagreements are unresolvableon every issue. That will require a separate argument. (For furtherdiscussions, see the entry onmoral disagreement, as well as Bergmann & Kain 2014, Besong 2014, Enoch2009, Huemer 2007, Kagan 2023, and Vavova 2014.)

3.2 Moral Explanations

Another way to argue for moral skepticism is to cite a requirement onjustified belief. On one view, we cannot be justified in believing anyclaim unless the truth of that claim is necessary for the bestexplanation of some independent fact. Some philosophers then arguethat moral truths are never necessary for the best explanation of anynon-moral fact (cf. Harman 1977). It follows that we cannot bejustified in believing any moral claim (cf. Hill 2016).

This argument can be countered in at least three main ways. First, onecould deny that justified belief must always involve inference to thebest explanation. It is not clear, for example, that beliefs aboutmathematics or colors are or must be grounded in this way, althoughsuch beliefs still seem justified. (Compare Harman 1977 on mathematicsand color.) If these beliefs are justified without any inference tothe best explanation, then it is not clear why moral beliefs need thatparticular kind of justification.

A related reply is that, even if moral facts are not necessary forexplanation, moral facts, truths, or beliefs still might be needed fordeliberation (Enoch 2011, McPherson & Plunkett 2015). If we needto be able to deliberate about what to do, and if we cannot deliberatewithout believing in moral facts or truths, then our beliefs aboutmoral facts or truths are supposed to be justified in some sense,according to this position.

A third common response is that sometimes a moral truth is necessaryfor the best explanation of a non-moral fact (cf. Sturgeon 1985).Hitler’s vices are sometimes cited to explain his atrocities.Slavery’s injustice has been said to explain its demise. And thefact that everyone agrees that it is morally wrong to torture babiesjust to get sexual pleasure might be best explained by the fact thatthis common belief is true. Moral skeptics usually reply that suchexplanations can be replaced by non-moral descriptions of Hitler,slavery, and torture. If such replacements are always available, thenmoral truths are not necessary for the best explanation of anything.However, it is not clear whether or not non-moral explanations reallydo work as well as moral explanations in all cases.

A special case of this argument, called an evolutionary debunkingargument, has led to vigorous debate recently (cf. Kahane 2011, May2018). Some moral skeptics (or at least skeptics about moral realism)argue that moral beliefs can be explained by evolutionary biologyperhaps with help from psychology, sociology, or culture withoutappeal to any moral fact or truth. Such truth-independent explanationsare then supposed to show that moral beliefs are either not true, notreally (mind independently) true, or not justified (cf. Joyce 2006,Street 2006, Braddock 2017, Bhogal 2023). As with other arguments fromthe explanatory impotence of moral facts, critics can reply either byarguing that moral facts do some explanatory work (e.g. Copp 2008 withreply by Street 2008) or by arguing that moral beliefs can bejustified even if moral facts do not do any explanatory work (cf.Bergmann & Kain 2014).

3.3 A Regress

The next argument develops a skeptical regress. This form of argument,which derives from Sextus Empiricus (Outlines of Scepticism),is sometimes used to support the more general skeptical claim that nobelief about any topic is justified. Nonetheless, it might seem tohave special force within morality if supposedly foundational moralbeliefs are especially problematic in some way.

The argument’s goal is to rule out one-by-one all of the ways inwhich a person might be justified in believing something. It startswith a definition:

A personS isinferentially justified in believing aclaim thatp if and only if what makesS justifiedis (at least in part)S’s ability to inferpfrom some belief ofS.

Notice that what matters for this way of being justified is notwhether the personactually bases the moral belief on anoccurrent inference but only whether the person isable toinfer that moral belief. This ability is usually understood to bepresent when the person has other beliefs that fit into a structurethat the person could use to infer the moral belief.

There are, then, only two ways to be justified:

(1) If any personS is justified in believing any moral claimthatp, thenS must be justified eitherinferentially or non-inferentially.

The moral skeptic denies both possibilities in turn. First:

(2) No personS is ever non-inferentially justified inbelieving any moral claim thatp.

Moral intuitionists and reliabilists as well as some moralcontextualists deny premise (2). Moral intuitionists claim that somemoral beliefs do not need to be justified by any inference becausethey are self-evident or we can directly perceive their truth (see theentry onintuitionism in ethics,as well as Audi 2015, Huemer 2007, Tropman 2017). Moral reliabilistsclaim that some moral beliefs do not need to be justified by anyinference because they are produced by some reliable non-inferentialprocess (cf. Shafer-Landau 2005). Moral contextualists claim that somemoral beliefs do not need to be justified by any inference because thecontext does not call for any argument (cf. Wellman 1971, Timmons1999). In response, moral skeptics argue that too many beliefs wouldbe justified if people did not need to be able to provide any reasonor inference to support their moral beliefs. If Thelma knows thatLouise believes that eating meat is morally wrong, and Louise knowsthat Thelma believes that eating meat isnot morally wrong,then they cannot both be non-inferentially justified in holding on totheir disputed beliefs. Conflicting beliefs can sometimes both bejustified, but it seems less plausible to hold that such conflictingmoral beliefs are all justified without any ability to support thebelief with any inference when each believer knows that other peopledisagree. If such conflicting beliefs are not justified in the absenceof a reason, and if such conflicts are widespread enough to undermineall non-inferential justification, then premise (2) is true.

Another way to argue for premise (2) invokes science. Psychologistshave found that many moral judgments are subject to a variety ofdistorting influences, including framing effects and certainmisleading emotions (e.g. McDonald et al. 2022. Biologists thensuggest that moral judgments evolved in ways that seem independent oftheir truth. Such indications of unreliability are supposed to showthat moral judgments are not justified without inference(Sinnott-Armstrong 2006, Chapter 9, pp. 184–219; but see repliesby Beaulieu 2009, van Roojen 2013, and May 2018). That would supportpremise (2).

Premises (1) and (2) together imply an intermediate conclusion:

(3) If any personS is justified in believing any moral claimthatp, thenS must be justified inferentially.

This means that, to be justified,S must be able to inferp from some other beliefs held byS. Of course,S does not have to draw any actual inference, butSstill needs to hold some other beliefs that could be used to justifythat belief.

But which other beliefs? There are three main possibilities:

(4) If any personS is inferentially justified in believingany moral claim thatp, then S must be justified by aninference with either (a) no normative premises or (b) some normativepremises but no moral premises or (c) some moral premises.

To the first possibility, moral skeptics respond with a variation onthe maxim that you can’t get “ought” from“is”:

(5) No personS is ever justified in believing any moralclaim thatp by an inference with no normative premises.

Naturalists in moral epistemology deny (5) when they try to derive aconclusion that an act is morally wrong from purely non-normativefeatures of the act. (See the entry onmoral naturalism.) However, moral skeptics retort that suchderivations always depend on a suppressed premise that all acts withthose features are morally wrong. Such a suppressed premise seemsmoral and, hence, normative. If so, the naturalist’s inferencedoes not really work without any normative premises. (On deriving anormative “ought” from a descriptive “is”, seeMaguire 2015, Pigden 2015, and Sparks 2018.) Naturalists still mightinvoke inferences to the best moral explanation, but then moralskeptics can deny that any moral hypothesis provides the bestexplanation independently of prior moral assumptions (see above).

The next possibility is to justify a moral conclusion with aninference whose premises are not moral but are normative in anotherway. This approach, which is adopted by some contractarians and idealobserver theorists among others, can be callednormativism.Normativists usually start with premises about rationality andimpartiality that are each supposed to be normative but morallyneutral. If rational impartial people under relevant circumstanceswould agree to certain moral standards, this is supposed to show thatthe corresponding moral beliefs are true or justified.

One problem for this general approach is that different theories ofrationality, impartiality, and relevant circumstances are allquestionable and lead to contrary moral beliefs. This suggests thatsuch theories are not morally neutral, so these derivations do notavoid moral premises. Other arguments from non-moral norms to moralconclusions run into similar problems. Moral skeptics concludethat:

(6) No personS is ever justified in believing any moralclaim thatp by an inference with some normative premises butno moral premises.

Premises (4)–(6) imply another intermediate conclusion:

(7) If any personS is justified in believing any moral claimthatp, thenS must be justified by an inferencewith some moral premise.

In short, moral beliefs must be justified by moral beliefs.

This creates a problem. Although the justifying beliefs must includesome moral beliefs, not just any moral beliefs will do:

(8) No personS is ever justified in believing a moral claimthatp by an inference with a moral premise unlessSis also justified in believing that moral premise itself.

Premise (8) is denied by some contextualists, who claim that, even ifa moral belief is not justified, if it is shared within a certainsocial context, then it may be used to justify other moral beliefs(cf. Wellman 1971, Timmons 1999). However, moral skeptics reply thatsocial contexts are often corrupt, and no social context by itself canshow that a moral belief is true, reliable, or, hence, justified inthe relevant way.

But then how can moral premises be justified? Given (7)–(8), themoral premises must be justified by inferring them from still othermoral beliefs which must also be justified by inferring them fromstill other moral beliefs, and so on. To justify a moral belief thusrequires a chain (or branching tree) of justifying beliefs orpremises, which must have one of two forms:

(9) If any personS is justified in believing any moral claimthatp, thenS must be justified by a chain ofinferences that either goes on infinitely or circles back to includep itself as an essential premise.

The first of these two alternatives is almost never defended, sincemost accept:

(10) No personS is ever justified in believing any moralclaim thatp by a chain of inferences that goes oninfinitely.

Moral skeptics also deny the other possibility:

(11) No personS is ever justified in believing any moralclaim thatp by a chain of inferences that includesp as an essential premise.

Any argument that includes its conclusion as a premise will be valid.However, anyone who doubts the conclusion will have just as muchreason to doubt the premise. So, according to skeptics, nothing isgained when a premise just restates the belief to be justified.

Premise (11) is opposed by moral coherentists (e.g. Brink 1989,Sayre-McCord 1996, Sinnott-Armstrong 2006, Chapter 10, pp.220–251). Recent coherentists emphasize that they do not infer abelief from itself in a linear way. Instead, a moral belief issupposed to be justified because it coheres in some way with a body ofbeliefs that is coherent in some way. Still, moral skeptics deny thatcoherence is enough to make a moral belief justified. One reason isthat the internal coherence of a set of beliefs is not evidence of anyrelation to anything outside the beliefs. Another reason is that everybelief — no matter how ridiculous — can cohere with somebody of beliefs that is internally coherent. Because so manyincompatible systems seem coherent, moral skeptics deny that coherencealone is sufficient to make beliefs justified.

Now the moral skeptic can draw a final conclusion. (9)–(11)imply:

(12) No person is ever justified in believing any moral claim.

This is dogmatic skepticism about justified moral beliefs.

Many opponents find this conclusion implausible, but the regressargument is valid. Hence, its conclusion cannot be avoided withoutdenying one of its premises. Different opponents of moral skepticismdeny different premises, as indicated. However, it remains to be seenwhether any of these responses to the regress argument is defensiblein the end.

3.4 Skeptical Hypotheses

The final kind of argument derives from René Descartes (1641).I do not seem justified in believing that what I see is a lake if Icannot rule out the possibility that it is a bay or a bayou.Generalizing, if there is any contrary hypothesis that I cannot ruleout, then I am not justified in believing that what I see is a lake.This common standard is supposed to be required for justified belief.When this principle is applied thoroughly, it leads to skepticism. Alla skeptic needs to show is that, for each belief, there is somecontrary hypothesis that the believer cannot rule out. It need not bethe same hypothesis for every belief, but skeptics usually buywholesale instead of retail, so they seek a single hypothesis that iscontrary to all (or many common) beliefs and which cannot be ruled outin any way.

The famous Cartesian hypothesis is of a demon who deceives me in allof my beliefs about the external world, while also ensuring that mybeliefs are completely coherent. This possibility cannot be ruled outby any experiences or beliefs, because of how the deceiving demon isdefined. This hypothesis is also contrary to my beliefs about thelake. So my beliefs about the lake are not justified, according to theabove principle. And there is nothing special about my beliefs aboutthe lake. Everything I believe about the external world isincompatible with the deceiving demon hypothesis. Skeptics concludethat no such belief is justified.

Notice that the deceiving demon hypothesis plays two separate roles.First, it is incompatible with common beliefs about the world, sinceit implies that there is no lake that I see. Second, it explains whywe would hold our common beliefs even if the hypothesis weretrue—namely, because the demon would deceive us into believingthat I see a lake in the hypothesized world. Some skeptics might holdthat an irrefutable contrary hypothesis is enough for a skepticalargument to show that the believer is not justified. For example, thehypothesis that our common beliefs are all wrong is contrary to thosebeliefs, but it does not explain why we hold them. However, anunexplanatory hypothesis like this strikes many as having less forcein skeptical arguments than an explanatory hypothesis like thedeceiving demon. For this reason, most skeptics try to construct theirskeptical hypotheses to include some explanation of why people wouldhold the same common beliefs as they do, even if they were wrong andthe skeptical hypothesis were true.

This kind of skeptical argument is often dismissed on the grounds thatthere is no reason to believe in a deceiving demon or that nobodyreally doubts whether there is an external world. In contrast, thisform of argument is not subject to such objections when it is appliedto morality, because some people really do adopt and even argue for aparallel skeptical hypothesis in morality:

Moral Nihilism = Nothing is morally wrong.

Moral nihilism here is not about what is semantically ormetaphysically possible. It is just a substantive, negative,existential claim that there does not exist anything that is morallywrong. Since smart and knowledgable people do take moral nihilismseriously and even argue for it (Mackie 1977, Joyce 2001), moralnihilism cannot be dismissed as readily as Descartes’s deceivingdemon.

Moral nihilism by itself is enough to be contrary to common moralbeliefs, but it is not explanatory. To make their argument moreforceful, therefore, moral skeptics usually add some explanation ofwhy people hold moral beliefs that are false (just as the story ofDescartes’s deceiving demon is supposed to explain why we holdfalse perceptual beliefs). The most popular explanations of our moralbeliefs refer to evolutionary and social advantages of holding certainmoral beliefs, especially when these moral beliefs are sharedthroughout society. Shared moral beliefs can facilitate socialcoordination and cooperation even when they are false (much as ashared non-moral belief that my neighbors will in fact catch andpunish me if I ever steal from them will reduce theft and increasetrust, even if this belief is false and I actually could get away withstealing).

When bare moral nihilism is supplemented with an explanation of whypeople would hold common moral beliefs even if they were false, thisconjunctive skeptical hypothesis can be calledexplanatory moralnihilism. Moral skeptics can then argue that explanatory moralnihilism forestalls any refutation. Since moral nihilists question allof our beliefs in moral wrongness, they leave us with no startingpoints on which to base arguments against them without begging thequestion at issue. Moreover, their explanations of our moral beliefspredict that we would hold exactly these moral beliefs, so the truthof its predictions can hardly refute explanatory moral nihilism. Ifthis trick works, then it fits right into a skeptical hypothesisargument.

This argument is clearest when it is applied to an example. If nothingis morally wrong, as moral nihilists claim, then it is not morallywrong to torture babies just for fun. So, according to the generalstandard above, one must be able to rule out moral nihilism in orderto be justified in believing that torturing babies just for fun ismorally wrong. But one cannot rule out moral nihilism if one wouldhold the same moral beliefs even if they were false, as claimed by theexplanation embedded in explanatory moral nihilism. Moral skepticsconclude that even this apparently obvious moral belief is notjustified. More precisely:

This moral belief about torturing babies is not especially problematicin any way. It seems as obvious as any moral belief. Thus, theargument can be generalized to cover any moral belief. Moral skepticsconclude that no moral belief is justified.

There are two main responses to such skeptical hypothesis arguments(cf. May 2013). First, some anti-skeptics deny (1) and claim thatskeptical hypotheses can be ruled out somehow. They might argue thatexplanatory moral nihilism is internally inconsistent or meaningless.If so, it can be ruled out by logic and semantics alone. However,moral nihilism does seem consistent and meaningful, according to allplausible theories of moral language, including expressivism, realism,and constructivism (Sinnott-Armstrong 2006, chap. 3). So is theexplanation that is part of explanatory moral nihilism. Explanatorymoral nihilism is also not subject to the kind of argument that Putnam(1981) deploys against more general skeptical scenarios. Anti-skepticsstill might argue that explanatory moral nihilism is incompatible withsome non-moral facts or observations or their best explanations. Ifso, it can be ruled out by arguments with only non-moral premises.However, all such attempts to cross the dreaded is-ought gap arequestionable (Sinnott-Armstrong 2006, chaps. 7–8; cf. Pigden2015, Sparks 2018). A third way to rule out explanatory moral nihilismwould be based on common moral beliefs that are incompatible with it.However, just as it would beg the question to use common beliefs aboutthe external world to rule out a deceiving demon hypothesis, so itwould also beg the question to argue against explanatory moralnihilism on the basis of common moral beliefs — no matter howobvious those beliefs might seem to us, and no matter how well thesecommon beliefs cohere together (Sinnott-Armstrong 2006, chaps.9–10). Moral skeptics conclude that there is no way to rule outexplanatory moral nihilism, just as premise (1) claims.

Another recent response is to deny premise (3). This is a principle ofclosure. Since a belief entails the denial of every contraryhypothesis, this closure principle in effect says that I cannot bejustified in believingp unless I am justified in denyingevery hypothesis contrary top — that is, unless I canrule outall contrary hypotheses. This principle has beendenied by relevant alternative theorists, who claim instead that onlyrelevant hypotheses need to be ruled out. On this theory, if skepticalhypotheses are not relevant, then a belief that it is morally wrong totorture babies just for fun can be justified, even if the believercannot rule out explanatory moral nihilism. For more detail on closurein moral contexts, see Lutz 2021.

For this response to have force, however, opponents of moralskepticism need to say why explanatory moral nihilism is irrelevant.It seems relevant, for the simple reason that it is directly contraryto the moral belief that is supposed to be justified. Moreover, smartpeople believe and give reasons to believe in explanatory moralnihilism (e.g. Mackie 1977, Joyce 2001). Some people are led to moralnihilism by the absence of any defensible theory of morality. Ifconsequentialism is absurd or incoherent, as some critics argue, andif deontological restrictions and permissions are mysterious andunfounded, as their opponents argue, then some people might believemoral nihilism for reasons similar to those that led scientists toreject phlogiston. Another basis for explanatory moral nihilism citesscience. If all of our moral beliefs can be explained by sociobiologyand/or other sciences without assuming that any moral belief is true,then some might accept explanatory moral nihilism for reasons similarto those that lead many people to reject witches or elves. The pointis not that such reasons for explanatory moral nihilism are adequate.The point here is only that there is enough prima facie reason tobelieve explanatory moral nihilism that it cannot be dismissed asirrelevant on this basis. If explanatory moral nihilism is relevant,and if closure holds for all or at least relevant alternatives, and ifexplanatory moral nihilism cannot be ruled out in any way, then moralskepticism seems to follow.

3.5 Relations Among the Arguments

These arguments for moral skepticism differ in many ways, but theyseem mutually supportive. One crucial premise in the skepticalhypothesis argument claims that nothing can rule out explanatory moralnihilism. The best way to support that premise is to criticize eachmethod for ruling out explanatory moral nihilism. That is just oneinstance of what the regress argument does more generally. Theargument from moral explanations excludes yet another way to rule outexplanatory moral nihilism. So, if these other arguments work, theysupport a crucial premise in the skeptical hypothesis argument.

Conversely, one crucial premise in the regress argument claims that nomoral belief can be justified non-inferentially. Another crucialpremise, (8), claims that an inference cannot justify its conclusionunless its premises are justified. These premises claim, in effect,that a moral belief needs a certain kind of justification. One way toestablish this need is to point to a contrary possibility that is notyet ruled out. That is what the skeptical hypothesis argument does.Another way to confirm this need is to show that the moral belief iscontroversial. That is what the argument from moral disagreement does.Thus, if these other arguments work, they support a crucial premise inthe regress argument.

To skeptics, this mutual support might seem desirable. Anti-skeptics,however, might object that this mutual support makes the argumentsjointly circular. In the end, the force of the arguments depends onthe defensibility of non-skeptical views in moral epistemology. Ifmoral intuitionism, coherentism, naturalism, or normativism works tojustify some moral beliefs and/or to rule out explanatory moralnihilism, then this will undermine the crucial premises in thearguments for moral skepticism. But that remains to be seen.

4. Pyrrhonian Moral Skepticism

Although the arguments for moral skepticism are hard to refute, mostpeople reject their conclusion. This makes it natural to seek somecompromise. Various compromises have been proposed, but here I willfocus on one that extends the Pyrrhonian tradition (Sinnott-Armstrong2006, chap. 6; cf. DePaul 2009).

This neo-Pyrrhonian position can be explained in terms of contrastclasses, which are also important for non-moral epistemology. Supposea father sees an animal in a zoo and believes it to be a zebra. If thefather has adequate evidence that the animal is not a lion or a horse,then the father can be justified in believing that it is a zebra outof the contrast class {lion, horse, zebra}. Nonetheless, the fatherstill might not have any evidence that the animal is not a mulepainted to look just like a zebra, especially if there are suchpainted mules in nearby cages. Then the father is not justified inbelieving that the animal is a zebra out of the contrast class {lion,horse, zebra, painted mule}.

A similar situation arises with moral beliefs. A father might bejustified in believing that he should tell his children the truthrather than lying to them, even if the father is not justified inbelieving that he should tell his children the truth as opposed tokeeping quiet or deceiving them withoout lying.

More generally, we can distinguish two contrast classes:

Most people do not take moral nihilism seriously in ordinarydiscussions, so the modest contrast class does not include moralnihilism. Thus, anyone who can rule out all other members of themodest contrast class but cannot rule out moral nihilism is justifiedin believing the moral claim out of the modest contrast class but notout of the extreme contrast class.

Critics will ask, “If someone is justified out of the modestcontrast class but not out of the extreme contrast class, is thisbeliever just plain justified (period or withoutqualification)?” That, of course, depends on what it means tosay that a believer is just plain justified (without qualification).On one plausible account, to say that a believer is just plainjustified (without qualification) is to say that the believer isjustified out of therelevant contrast class. But whichcontrast class is relevant when?

Contextualists say that the modest contrast class is the relevant onein everyday contexts, such as hospital ethics committees, where itwould be seen as a distraction to discuss moral nihilism. Nonetheless,the extreme contrast class is said to be the relevant one inphilosophical contexts, such as philosophy classes where moralnihilism is taken seriously. This allows contextualists to hold that adoctor in a hospital ethics committee is justified in believing amoral claim that a philosophy student with the same evidence would notbe justified in believing. However, problems arise when contextscross. Consider a philosophy student who says that the doctor on theethics committee is not justified in believing the moral claim. Is thestudent’s contrast class (with moral nihilism) or thedoctor’s contrast class (without moral nihilism) really relevantto the student’s judgment about the doctor’s belief? Whenepistemic assessments cross contexts in such ways, it is hard to seeany basis for claiming that either context or either contrast classreally is the relevant one for assessing whether the believer reallyis just plain justified (without qualification).

Such paradoxes lead some moral skeptics to deny that any contrastclass is ever really relevant. This denial implies that it is nevereither true or false that a believer is justified (withoutqualification), if such claims presuppose that some contrast class isreally relevant. Alternatively, Pyrrhonian moral skeptics mightsuspend belief about whether any contrast class is ever reallyrelevant or not. Such Pyrrhonian moral skeptics refuse to take anyposition on whether any believer is or is not just plain justified(without qualification), although they can still talk about whethersomeone is justified in believing a moral claim out of a specifiedcontrast class. Pyrrhonian moral skeptics can then (i) acceptskepticism about extremely justified moral belief but (ii) denyskepticism about modestly justified moral belief and (iii) refuse toeither assert or deny (dogmatic) skepticism about any moral beliefbeing just plain justified (without qualification).

Whether or not this view is finally defensible, the point here is justthat such a Pyrrhonian compromise is available and attractive to thosewho want to avoid dogmatic moral skepticism but see no way to refuteit. There are also other possible compromises that combine differentstrands in and kinds of moral skepticism. That is what makes it sofascinating to study this important group of views.

Bibliography

• Audi, Robert, 2015,Moral Perception, Princeton:Princeton University Press.
• Ayer, A. J., 1952,Language, Truth, and Logic, New York:Dover. (First edition originally published in 1935.)
• Bambrough, Renford, 1979,Moral Skepticism and MoralKnowledge, London: Routledge.
• Beaulieu, Gerard, 2009, “Sinnott-Armstrong’s MoralSkepticism: A Murdochian Response”,Dialogue, 48(3):673–678.
• Bergmann, Michael, and Kain, Patrick (eds.), 2014,Challengesto Moral and Religious Belief: Disagreement and Evolution, NewYork: Oxford University Press.
• Besong, Brian, 2014, “Moral Intuition andDisagreement”,Synthese, 191(12): 2767–2789.
• Bhogal, Harjit, 2023, “Explanationism vs Modalism inDebunking (and Theory Choice)”,Mind, 132(528):1005–1027.
• Blackburn, Simon, 1993,Essays in Quasi-Realism, NewYork: Oxford University Press.
• Braddock, Matthew, 2017, “Debunking Arguments fromInsensitivity”,International Journal for the Study ofSkepticism, 7(2): 91–113.
• Brink, David, 1989,Moral Realism and the Foundations ofEthics, Cambridge: Cambridge University Press.
• Butchvarov, Panayot, 1989,Skepticism in Ethics,Bloomington and Indianapolis; Indiana University Press.
• Copp, David, 1991, “Moral Skepticism”,Philosophical Studies, 62: 203–233.
• –––, 2008, “Darwinian Skepticism aboutMoral Realism”,Philosophical Issues, 18:186–206.
• DePaul, Michael, 2009, “Pyrrhonian Moral Skepticism and theProblem of the Criterion”,Philosophical Issues, 19(1):38–56.
• Descartes, René, 1641,Meditations on FirstPhilosophy, translated by D. C. Cress, Indianapolis: Hackett,1979.
• Enoch, David, 2009, “How is Moral Disagreement a Problem forRealism?”,The Journal of Ethics, 13(1): 15–50.
• –––, 2011,Taking Morality Seriously: ADefense of Robust Realism, New York: Oxford UniversityPress.
• Gibbard Allan, 1990,Wise Choices, Apt Feeling,Cambridge: Harvard University Press.
• –––, 2003,Thinking How to Live,Cambridge: Harvard University Press.
• Hare, R. M., 1981,Moral Thinking, Oxford: ClarendonPress.
• Harman, Gilbert, 1977,The Nature of Morality, New York:Oxford University Press.
• Hill, Scott, 2016, “From Isolation to Skepticism”,Erkenntnis, 81:649–668.
• Huemer, Michael, 2007,Ethical Intuitionism, London:Palgrave Macmillan.
• Joyce, Richard, 2001,The Myth of Morality, Cambridge:Cambridge University Press.
• –––, 2006,The Evolution of Morality,Cambridge: MIT Press.
• Kagan, Shelly, 2023,Answering Moral Skepticism, NewYork: Oxford Unievrsity Press.
• Kahane, Guy, 2011, “Evolutionary Debunking Arguments”,Nous, 45:103–125.
• Lutz, Matt, 2021, “The Moral Closure Argument”,Journal of Ethics and Social Philosophy, 19(1): 80–110.
• Machuca, Diego E. (ed.), 2017,Moral Skepticism: NewEssays, New York: Routledge.
• Mackie, J. L., 1977,Ethics: Inventing Right and Wrong,New York: Penguin.
• Maguire, Barry, 2015, “Grounding the Autonomy ofEthics”, inOxford Studies in Metaethics (Volume 10),Russ Shafer-Landau (ed.), Oxford: Oxford University Press
• May, Joshua, 2013, “Skeptical Hypotheses and MoralSkepticism”,Canadian Journal of Philosophy, 43(3):341–359.
• –––, 2018,Regard for Reason in the MoralMind, New York: Oxford University Press.
• McDonald, Kelsey, Graves, Rose, Yin, Siyuan, Weese, Tara, &Sinnott-Armstrong, Walter, 2021, “Valence framing effects onmoral judgments: A meta-analysis”,Cognition, 212:104703.
• McPherson, Tristram, and Plunkett, David, 2015,“Deliberative Indispensability and EpistemicJustification”,Oxford Studies in Metaethics (Volume10), Russ Shafer-Landau (ed.), Oxford: Oxford University Press.
• Pigden, Charles, 2015, “Hume on Is and Ought: Logic,Promises, and the Duke of Wellington”,The Oxford Handbookof David Hume, Paul Russell (ed.), Oxford: Oxford UniversityPress.
• Putnam, Hilary, 1981,Reason, Truth, and History,Cambridge: Cambridge University Press.
• Russell, Bruce, 1988, “Two Forms of EthicalSkepticism”, in L. Pojman (ed.),Ethical Theory,Belmont, Cal.: Wadsworth.
• Sayre-McCord, Geoffrey, 1996, “Coherentist Epistemology andMoral Theory”, inMoral Knowledge? New Readings in MoralEpistemology, W. Sinnott-Armstrong and M. Timmons, (eds.), NewYork: Oxford University Press.
• Schroeder, Mark, 2010,Being For: Evaluating the SemanticProgram of Expressivism, New York: Oxford University Press.
• Sextus Empiricus,Outlines of Scepticism, translated byJulia Annas and Jonathan Barnes, Cambridge: Cambridge UniversityPress, 2000.
• Shafer-Landau, Russ, 2005,Moral Realism: A Defense, NewYork: Oxford University Press.
• Sidgwick, Henry, 1874/1966,Methods of Ethics, New York:Dover. (First edition originally published in 1874.)
• Singer, Marcus, 1973, “Moral Skepticism”, inSkepticism and Moral Principles, C. Carter (ed.), Evanston,Ill.: New University Press.
• Sinnott-Armstrong, Walter, 2006,Moral Skepticisms, NewYork: Oxford University Press.
• –––, 2010, “Expressivism andEmbedding”,Philosophy and Phenomenological Research,61: 677–693.
• Sparks, Jacob, 2018, “Is, Ought, and the RegressArgument”,Australasian Journal of Philosophy, 97(3):528–543.
• Stevenson, Charles, 1944,Ethics and Language, New Haven:Yale University Press.
• Street, Sharon, 2006, “A Darwinian Dilemma for RealistTheories of Value”,Philosophical Studies, 127:109–166.
• –––, 2008, “Reply to Copp: Naturalism,Normativity, and the Varieties of Realism Worth Worrying About”,Philosophical Issues, 18: 207–228.
• Sturgeon, Nicholas, 1985, “Moral Explanations”, inMorality, Reason, and Truth, D. Copp and D. Zimmerman (eds.),Totowa, N.J.: Rowman and Allanheld.
• Timmons, Mark, 1999,Morality Without Foundations: A Defenseof Ethical Contextualism, New York: Oxford University Press.
• Tropman, Elizabeth, 2017, “Intuitionism in MoralEpistemology”, inThe Routledge Handbook of Metaethics,T. McPherson & D. Plunkett (eds.), London: Routledge,pp. 472–483 .
• Van Roojen, Mark, 2013, “Moral Intuitionism, Experiments,and Skeptical Arguments”, inIntuitions, Booth andRowbottom (eds.), Oxford: Oxford University Press.
• Vavova, Katia, 2014, “Moral Disagreement and MoralSkepticism”,Philosophical Perspectives, 28(1):302–333.
• Wellman, Carl, 1971,Challenge and Response: Justification inEthics, Carbondale: Southern Illinois University Press.
• Williams, Bernard, 1985,Ethics and the Limits ofPhilosophy, Cambridge, MA: Harvard University Press.

Academic Tools

	How to cite this entry.
	Preview the PDF version of this entry at theFriends of the SEP Society.
	Look up topics and thinkers related to this entry at the Internet Philosophy Ontology Project (InPhO).
	Enhanced bibliography for this entryatPhilPapers, with links to its database.

Other Internet Resources

[Please contact the author with suggestions.]

Related Entries

cognitivism vs. non-cognitivism, moral |Descartes, René: epistemology |knowledge: analysis of |moral epistemology |moral non-naturalism |moral realism |moral relativism |skepticism: ancient