Intuitionism in Ethics (2024)

1. Intuitionist epistemology

1.1 Intuition

One of the most distinctive features of Ethical Intuitionism isits epistemology. All of the classic intuitionists maintained thatbasic moral propositions are self-evident—that is, evident inand of themselves—and so can be known without the need of anyargument. Price distinguishes intuition from two other grounds ofknowledge—namely, immediate consciousness or feeling on the onehand, and argumentation, on the other. Argumentation, or deduction, isknowledge that is ultimately derived from what is immediatelyapprehended, either by sensation or by the understanding. Immediateconsciousness, or feeling, is the mind’s awareness of its ownexistence and mental states (Price, 1758/1969, 159). It sharesimmediacy with intuition, but unlike intuition does not have as itsobject a self-evident proposition. Such immediate self-consciousnessis immediate apprehension by sensation. Intuition is immediateapprehension by the understanding. It is the way that we apprehendself-evident truths, general and abstract ideas, “and anythingelse we may discover, without making any use of any process ofreasoning” (1758/1969, 159).

The claim that intuition is immediate apprehension by theunderstanding suggests a notion of intuition in Price that is moreakin to current accounts of intuitions as intellectual seemings orpresentations (Bealer 1998; Chudnoff 2013). Intellectual seemings arethe intellectual analogue of perceptual seemings. Just as certainthings can seem perceptually to be a certain way, e.g., coloured, orstraight, so certain propositions can seem to be true, or presentthemselves to the mind as true. These seemings are not beliefs, forsomething can seem true even though one does not believe it, e.g., itmay seem true that there are more natural numbers than even numbers,but we know that is false, so do not believe it.

Similarly, on Price’s view, an intuition isn’t a beliefwith certain characteristics, such as being pre-theoretical,non-derivative, firmly held, etc. Beliefs aren’t immediateapprehensions of anything, though they may be based on suchapprehensions. One might be tempted to think that perceptual beliefs,such as the belief that there is a cat asleep in front of me, areimmediate apprehensions, but that would be to confuse the perceptualapprehension of the cat with the belief based on that apprehension.Beliefs like this, perceptual beliefs, are based on the immediateexperience (sensory intuition) I have of a cat sleeping in front ofme; they are not the sensory intuition itself. As Price understandsit, then, an intellectual intuition is something closely akin to anexperiential presentation, or seeming.

The main difference between early intuitionists’ notion of intuitionand intellectual seemings is that the latter is standardly regarded asnon-factive—that is, an intuition that p in this sense doesnot entail that p. An apprehension that p does sound factive,however. One cannot apprehend something that is not there to beapprehended. What is present to the mind in apprehension is the thingitself, not our representation of it. So whereas modernepistemologists regard intuition, understood as an intellectualseeming or presentation, as something analogous to the non-factivenotion of a perceptual experience, it looks like some classicintuitionists regarded intuition as analogous to the factive notion ofa perception. The virtue of the modern, more modest notion ofintuition is that it allows that intuition is fallible. But in doingthis it loses the appeal of the direct realist account Price seems towork with of certain propositions or facts being immediately presentto the mind.

There is scope for intuitionists to be disjunctivist about suchintuitions. They could maintain that some intuitions areapprehensions, and some are intellectual seemings. Subjectively wecannot tell one from the other, but they are, one might argue, verydifferent states. But it is not evident that anyone has defended sucha view of intuition.

It is not clear that all intuitionists understand intuitions on thisperceptual or quasi-perceptual model, as many do not use this notionat all. W. D. Ross, for example, uses the notion of apprehension, buthe tends to base his moral theory largely on our considered moralconvictions. “[T]he moral convictions of thoughtful andwell-educated people are the data of ethics, just as sense-perceptionsare the data of natural science” (1930/2002, 41). Convictionsare, however, a certain type of belief rather than an intellectualapprehension, or seeming. So it looks like we can find two notions ofintuition in intuitionist thought—one understood as anintellectual seeming or apprehension, and the other as apre-theoretical, non-inferred, firmly held belief orconviction.^[2]Which one they opt for makes a difference to their epistemology.

Like other epistemic foundationalists, Price insists that allreasoning and knowledge must ultimately rest on propositions that arenot inferred from other premises. For ethical intuitions thisnon-inferred basis of knowledge is self-evident truth grasped byintuition.^[3]It is, however, important to keep intuition and self-evidence separatefor a number of reasons. First, a conscious intuition is a certainmental state, either a non-inferential belief or an intellectualseeming. But a self-evident proposition is not a conscious mentalstate. Second, intuition is a way in which we are aware ofself-evident propositions, whereas self-evident propositions are thethings that can be known in this way. Such propositions could (pacePrice) be believed without an intuition of them. As I will explainlater, one might have some argument that leads one to believe aself-evident proposition, or one may believe it on the basis ofauthoritative testimony. Thirdly, it may turn out that things otherthan self-evident propositions can be grasped by intuition. Forinstance, we may have moral intuitions about concrete cases, such asvarious trolley cases (see below) and various anti-consequentialistcounter-examples. But it is not obvious that the contents of theseintuitions are self-evident propositions; whether they are may dependon how self-evidence is understood.

Price claims that self-evident truths are “incapable ofproof” (1758/1969,160).^[4]Most classical intuitionists endorse this view,although Ross is arguably an exception. This is easily missed, for hestates at one point that self-evident moral propositions“cannot be proved, but … just as certainly need no proof”(1930/2002, 30). But elsewhere in The Right and the Good hemakes only the more restricted claim that such propositions do notneed anyproof,^[5]anddespite the fact that he sometimes makes the further claim that theycannot receive any proof by means of argument, this furtherclaim does not seem to reflect his consideredview.^[6]Evidence that it does not can be found in anarticle written three years before the publication of The Right andthe Good where he explicitly states that “the fact thatsomething can be inferred does not prove that it cannot be seenintuitively” (1927, 121). If he thinks that some proposition canbe inferred from (justified by) other propositions and be self-evident,he clearly thinks that its being self-evident does not rule out thepossibility of a proof. In any case there is nothing in the notion of aself-evident proposition that rules out justification or argument forthat proposition. A self-evident proposition is one that we can bejustified in believing without an argument, but this does not rule outthe possibility that there may be such an argument, or justification,or that the proposition may be believed on thatbasis.^[7]Since such arguments are not needed for us to be justified inbelieving a self-evident proposition, they are what may be called’epistemically supererogatory’.

Although there can be an argument for a self-evident or an intuitive proposition, ifintuitions are understood as intellectual seemings, then intuitionscannot be justified. This is easily missed, as we tend to identifyintuitions with the beliefs based on them, rather than with theintellectual seemings on which those beliefs are based. But if we takean intuition to be an intellectual seeming, then intuitions cannot bejustified any more than a perceptual seeming can. Take a perceptualseeming, such as it’s seeming that the wall is green. Thisperceptual seeming could be explained in various ways, but itis odd to suppose that this experience could be justified by anything(which is not to rule out the possibility that the belief based on itcould be justified). This is because we are in a certain sense passivein relation to such seemings. Similarly, if some proposition presentsitself to the mind as true, then this presentation cannot be justified,although the belief based on it can be (and we might add, theproposition intuited can be), for its seeming to be true is not aconclusion we could arrive at: it’s just how certain propositionspresent themselves to the mind.

It may be that the reason that Ross switches between making thestronger claim that self-evident propositions cannot bejustified, and the weaker claim that they need no justification, isthat he had in mind a belief in some self-evident proposition when hesaid they could be justified, and our intuition (apprehension) of thatproposition when he said that they could not.

1.2 Self-evidence

The notion of a self-evident proposition is a term of art inintuitionist thought, and needs to be distinguished from certaincommon sense understandings with which it may easily be conflated. Thefirst thing to note is that a self-evident proposition is not the sameas an obvious truth. To begin with, obviousness is relative to certainindividuals or groups. What is obvious to you may not be obvious tome. But self-evidence is not relative in this way. Although aproposition may be evident to one person but not to another, it couldnot be self-evident to one person, but not to another. A propositionis just self-evident, not self-evident to someone. Secondly,there are many obvious truths that are not self-evident. Certainwell-known empirical truths, for example, that if I drop a heavyobject it will fall, or that the world is bigger than a football, areobvious but not self-evident. There are also self-evident propositionsthat may not be obvious to everyone, at least prior to reflection,e.g., that if all As are Bs and no Bs are Cs then no Cs areAs, or that one man can be the father and grandfather of the same child.

What then is it for a proposition to be self-evident? Locke says thata self-evident proposition is one that “carries its own lightand evidence with it, and needs no other proof: he that understandsthe terms, assents to it for its own sake” (1969, 139). Pricetells us that a self-evident proposition is immediate, and needs nofurther proof, and goes on to say that self-evident propositions needonly be understood to gain assent 1758/1969, 187). Ross writes, aself-evident proposition is “evident without any need of proof,or of evidence beyond itself” (1930/2002, 29), and Broaddescribes self-evident propositions as being “such that arational being of sufficient insight and intelligence could see it tobe true by merely inspecting it and reflecting on its terms and theirmode of combination” (1936, 102–3). These passages mayhave led to the standard understanding of a self-evident propositionthat one finds in Shafer-Landau (2003, 247) and Audi (2001, 603; seealso Audi 2008, 478). Audi, for instance, writes that self evidentpropositions are “truths such that (a) adequately understandingthem is sufficient justification for believing them …, and (b)believing them on the basis of adequately understanding them entailsknowing them” (2008, 478).

One should distinguish knowing a self-evident proposition from knowingthat that proposition is self-evident. The former does not imply thelatter. Someone might know some self-evident proposition, such as thatif A is better than B and B is better than C thenA is better than C, but lack the concept of self-evidence, socouldn’t know that that proposition is self-evident. One might evenknow a self-evident proposition whilst endorsing a theory according towhich no propositions are self-evident.

But given that a proposition may seem to be self-evident when it is not,it is useful to have a way of discriminating the merely apparent fromthe real ones. Sidgwick’s criteria may be regarded as helping us dothis. According to Sidgwick, to be sure that a proposition is self-evident it must:

1. be clear and distinct
2. be ascertained by careful reflection
3. be consistent with other self-evident truths
4. attract general consensus (1874/1967, 338)

If some apparent self-evident proposition does not have all of thesefeatures then we should reduce our confidence that it is a genuineself-evident proposition. It is, however, a striking feature of Sidgwick’s ownprinciples that they do not pass this test. But whether we can come toknow that some proposition is self-evident in this, or some other way,the point is that we do not need to know that some proposition isself-evident to know that it is true.

According to the standard account, a self-evident proposition is onewhich an adequate understanding of the proposition justifies us inbelieving it. But all that Locke and Price say is that we need anunderstanding of a self-evident proposition in order to believe it,and presumably to be justified in believing it. They do not say thatour understanding provides that justification, or that when we believeit, we believe it on the basis of our understanding. Indeed the ideathat it is our understanding of a self-evident proposition thatjustifies us in believing it may sound odd to many people. Certainlyif one assumes that if p justifies belief in q, then p isa reason to believe q, it is hard to see how our understandingjustifies us in believing the proposition understood, for no one canclaim that their understanding of a proposition gives them a reason tobelieve it. One explanation of this is that the sort of thing that canprovide reasons to believe that p are either evidence that por, more controversially, pragmatic considerations, such as thatbelieving that p will have certain good consequences. An adequateunderstanding of a proposition is neither of these things. The factthat I understand some proposition is not something that would makebelieving that proposition have any good consequences, and myunderstanding is not evidence for the truth of the propositionunderstood. Evidence is standardly understood as something that makesthe proposition it is evidence for more probable. But an adequateunderstanding of a proposition does not make that proposition moreprobable, so is not evidence for it. Since an understanding of aproposition is neither a pragmatic nor an evidential consideration, itseems it is not the right sort of thing to give us a reason to believethat proposition, and so not the right sort of thing to justify thatbelief. Robert Audi has, however, recently offered an account ofadequate understanding that attempts to deal with this kind ofobjection. He does not claim that adequate understanding of aproposition is evidence for that proposition, but claims that adequateunderstanding puts us in touch with the truth-makers of theself-evident proposition, and it is this that makes it the right sortof thing to justify belief (Audi, 2019, 379–380)

But the oddity of supposing that it is our understanding thatjustifies us in believing a self-evident proposition does not rest onthe assumption that justification is to be defined as reason tobelieve, or that what gives us reason to believe that p isevidence for the truth of p or some benefit of believing this. If youask someone why they believe some apparently self-evident proposition,such as that agony is bad, it would be very surprising if they repliedby saying “I believe it because I understand it”.

Given these worries about whether our understanding can justify us inbelieving the proposition understood, we should ask whether there isanything else that might justify us in believing a self-evidentproposition. If intuitions are beliefs, then our intuition that pcannot justify us in believing p. The same is true if intuitions areinclinations to believe, as Williamson (2000), Sosa (2007) and Earlenbaugh and Molyneux, 2009) claim, for the fact that I aminclined to believe some proposition is no justification for believingit. But if intuitions are intellectual seemings, then they may be ableto justify the beliefs based on them. For with this understanding ofan intuition we can say that what justifies our belief in aself-evident proposition is that it seems true, just as we might saythat what justifies us in having some experiential belief is thatperceptually, the world seems to be that way. Why do you believe thewall is green? Because it seems green. Why do you think that agony isbad? Because it seems bad.

If intuitions rather than our understanding of their content justifyus in believing that content, then intuitionists should understand aself-evident proposition as follows:

A self-evident proposition is one of which a clear intuition issufficient justification for believing it, and for believing it on thebasis of that intuition (see Stratton-Lake, 2016, 38).

An adequate understanding is necessary for one to be justified in thisway, but this is not because understanding provides justification;rather, it is because it is needed to get the proposition clearly inview, and so enables a clear intuition of it. But it is the intuitionthat justifies, not the understanding.

Whether this account helps intuitionists will depend on a more generalmetaphilosophical debate about the role of intuitions in philosophy,and whether intuitions justify. There is a difference, though. Themore general debate is about whether intuitions provide evidence forbelieving or rejecting certain theories, whereas the intuitionistneeds intuitions to justify beliefs with the same content. Secondly,there might be reason to think that although intuitions can providejustification in other areas of philosophy, they cannot do that inmorality. For instance, one might think that people’s moralintuitions vary too much to be reliable indicators of truth, or thatin morality emotions can distort our intuitions (seeSinnott-Armstrong, 2006), or that we only have the moral intuitions wedo because having those beliefs has survival value (Street, 2006).

Some recent intuitionists have shied away from the view thatcertain moral propositions are self-evident and have, instead, arguedthat all the intuitionist needs is the claim that intuitions,understood as intellectual seemings, provide non-inferentialjustification for some of our moral beliefs (Huemer 2005, 106 andBedke 2008—though Bedke rejects intuitionism). As these authorsview things, claiming that some at least of the propositions intuitedare self-evident does not gain intuitionists anything.

1.3 Disagreement

Many philosophers think that pervasive moral disagreement casts doubton the intuitionists’ claim that certain moral propositions areself-evident. If there were certain moral propositions that can beknown if adequately understood, then, it is argued, people with anadequate understanding of them would believe them, and there would beuniversal assent amongst mature, comprehending people. But there is nosuch universal assent. So there are no self-evident moralpropositions.

Sidgwick took disagreement seriously, and thought that if there wassignificant disagreement about the truth of some apparentlyself-evident moral proposition, then that casts doubt on whether thatproposition really is self-evident. Intuitionists could defendthemselves from this objection by downplaying the amount of moraldisagreement. They might claim that a lot of moral disagreement stemsfrom disagreement about non-moral facts, such as what the consequencesof a certain act will be. For instance, two people might disagreeabout whether it is permissible to boil lobsters alive just becausethey disagree about whether lobsters can feel pain. Since the basis oftheir moral disagreement is this disagreement about the relevantneurological fact, if they agreed on this non-moral fact, we couldexpect them to agree about the permissibility of boiling lobstersalive.

Furthermore, although people might disagree about the permissibilityof boiling lobsters alive, we may assume that they agree that pain isa bad thing, and the infliction of undeserved pain is prima faciewrong. If this assumption is correct, then the disputants agree aboutthe moral facts here. They disagree only about the empirical,non-moral facts.

Another factor that might explain moral disagreement isdisagreement about the strength of certain moral reasons. Manydisputants might agree about the non-moral facts, and about what ismorally relevant, but disagree about the weight that should be givento the different moral considerations. So, for example, two peoplemight disagree about whether they ought to push a large man onto therail tracks to derail a trolley that would otherwise kill five people,but this is quite consistent with them agreeing that the fact thatthis act would save five people counts in favour of pushing the manoff the bridge, and that the fact that he would die if this were donecounts against doing it. In such a case there is agreement about whatis relevant, and how it is relevant, but disagreement about the weightof the competing moral considerations—one person regards theevil of killing one person as weightier than the good of saving five,while the other regards the evil of killing one as outweighed by thegood of saving the five. There is still moral disagreement here, butit is simply a difference in judgement about the application of agreedmoral principles.

This fits with the intuitionist view held, e.g., by Ross, that causinggood outcomes is prima facie right, and that causing harm is primafacie wrong, for Ross held that both of these propositions areself-evident. He denied, however, that the stringency, or weight, ofthese different prima facie duties is self-evident (1939, 188). Aboutthis, he maintained, we could only have a probable opinion.

It is worth noting that moral disagreement does not imply that peoplehave different intuitions. Ross, for example, had the strong intuitionthat it is permissible to miss an opportunity to enjoy some innocentpleasure, but at the time he wrote The Right and the Good hedid not believe this. (Later, in the Foundations of Ethics, hechanged his mind.) He believed it would be wrong to pass up such anopportunity because pleasure is good, and we ought to maximise thegood. So although he would disagree with someone who believed that itis permissible to pass up an opportunity to enjoy some innocentpleasure, he would share their intuition that this is permissible.Similarly, it is plausible to suppose that many act consequentialistsstill have the intuition that it is wrong to harvest organs from ahealthy but non-consenting donor to save five other lives. But becausethey have persuaded themselves of the truth of act consequentialism,they would not believe this act is wrong.

One of the theoretical advantages of thinking of intuitions asintellectual seemings is that we can allow for this mismatch betweenour intuitions and our beliefs. Just as something can seemperceptually to be a certain way while we don’t believe it is thatway, as in Müller-Lyer cases, so a proposition can seemintellectually to be true even though we do not believe it. Since suchseemings are not beliefs, this does not commit intuitionists to theview that a conflict between our intuitions and beliefs entails contradictory beliefs.

Intuitionists like Ross could still allow that their non-believedintuitions provide pro-tanto justification for believing them. It’sjust that this justification is outweighed by opposing intuitions, andthe theory based on them. A consequentialist with non-consequentialistintuitions could think the same. She may regard her deontologicalintuitions as giving her some justification for believing that itwould be wrong to harvest the organs to save five, but presumablywould regard the appeal of the consequentialist theory as a whole asoutweighing this justification. She may even regard her deontologicalintuition as providing sufficient reason to believe that such acts arewrong, even though she thinks that on balance she has more reason tobelieve that such acts are permissible.

Finally, Ethical Intuitionists allow that various other factors canlead to disagreement. Clarke, for instance, allowed that stupidity,corruption, or perverseness may make one doubt self-evidentpropositions (1706/1969, 194). John Balguy also acknowledges thatself-evident moral principles, like many other plain and evidenttruths, may be, and have been, doubted, “even by philosophersand men of letters” whom, presumably, he did not regard asstupid or corrupt (1728/1969, 406). And Price maintained that allforms of knowledge, including intuitive knowledge, may be evident indifferent degrees (1758/1969, 160). Intuition may be clear andperfect, but may sometimes be faint and obscure. Such variance indegrees of clarity allows that a self-evident proposition may beimperfectly and obscurely grasped, and this may lead someone to denyits truth. Similarly, Moore claimed that “every way in which itis possible to cognise a true proposition, it is also possible tograsp a false one” (1903/1993, 36), and Ross notes thatself-evident propositions may only be evident to us once we reach acertain moral maturity (1930/2002, 29). Given all these ways inwhich the truth of a self-evident proposition may be missed, it is nosurprise that there is no universalassent.^[8]But the absence ofuniversal assent is quite consistent with self-evidence, as long asone does not regard ‘self-evidence’ to mean, or imply,obviousness.^[9]

But despite what has been said above, critics of intuitionism canclaim that the fact that there is disagreement between moralphilosophers and even between intuitionists themselves undermines the viewthat certain propositions are self-evident. These philosophers willhave thought long and hard about the relevant propositions, and (wewould hope) have a very clear understanding of them. One would,therefore, expect that if there were certain moral propositions thetruth of which could be apprehended by intuition, then moralphilosophers would converge on these truths. Persistent disagreementamongst reflective, thoughtful, and comprehending moral philosophersmay cast doubt on the view that any of these propositions areself-evident.

Furthermore, if intuitions are intellectual seemings, one might askwhy certain moral propositions seem true whereas others do not. Ifmoral facts are non-natural facts, as intuitionists maintain, andnon-natural properties lack causal powers, then moral intuitionscannot be caused by the corresponding moral facts, as, e.g., certainperceptual seemings are caused by certain natural facts. Critics wouldargue that certain things seem right and good to us, not because ofsome inherent value they have, but because we have evolved to react tocertain types of act with approval or disapproval (Singer 2005; Street2006; and Joyce 2007, ch. 6).

We have evolved to feel instant approval of acts that benefit ourgroup, such as those that exemplify reciprocal trust, and honesty, andthus enhance our chances of survival, and to feel disapproval of actssuch as deceit, and betrayal, that undermine trust and the benefitsthat brings. Intuitionists need to find some way of responding to thissort of objection without abandoning their non-naturalism.

1.4 Trolley cases and the reliability of moral intuitions

Empirical psychology has recently cast doubt on the reliability of atleast some of our moral intuitions. Since a self-evident propositionis one of which a clear intuition justifies us in believing it, thesedoubts call into question the claim that our moral intuitions justifyus in believing them, and thus whether there are any self-evidentmoral propositions. The experiments that cast doubt on our intuitionstend to focus on our intuitions about trolley cases. Consider thefollowing three cases:

Switch: there are five people on the rail track,and an out of control trolley that will kill all five of them. There isa lever that would divert the trolley onto a different track. But thereis a single person on that track who would be killed if you pull thelever and divert the trolley.

Bridge: there are five people on the rail track,and an out of control trolley that will kill all five of them. There isa large man standing on the bridge over the track. If you pushed himoff the bridge onto the track, he would be killed. But he would derailthe trolley and so the five people on the track would be saved.

Trap door: there are five people on the rail track,and an out of control trolley that will kill all five of them. There isa man standing on the bridge over the track. If he fell onto the trackhe would be killed, but would derail the trolley thus saving the fivepeople on the track. He is standing on a trap door that would open anddrop him onto the track if you pulled a lever.

People tend to say that they should pull the lever in Switch, butthat they ought not to push the man off the bridge in Bridge. Bridgelooks similar to Switch in that you would be killing one person to savefive. So why the different intuitions? One could try to explain awaythese apparently conflicting intuitions with the doctrine of doubleeffect. According to this doctrine, we may produce some good thatinvolves a bad outcome, so long as the bad outcome is not intended. If thebad outcome is a means to the good end, then it is intended (as ameans), and is not merely foreseen. So such acts are wrong according tothe doctrine of double effect.

Switch seems to be a case where the bad outcome is foreseen, but notintended. Bridge seems to be a case where the bad outcome is intendedas a means of saving the five. So one way to explain the differentintuitions in Switch and Bridge is with reference to the doctrine ofdouble effect. But this explanation is unsettled by a variant ofSwitch, according to which the large man is on the spur track, and thattrack now loops back onto the main one. Here it looks like by pullingthe lever we would be using the large man merely as a means to savingthe five, for unless he stops the trolley it will loop round and killthe five from the other direction. But it still seems permissible topull the lever, yet wrong to push the man off the bridge.

Furthermore, this explanation of people’s different intuitions iscalled into question by Trap Door. For Trap Door is like Bridge in thesense that the bystander is killed as a means of saving the five, butmany more people tend to have the intuition that it is permissible topull the lever in Trap Door (Greene et al. 2009).

The difference in people’s intuitions between Bridge and Trap Doorcasts serious doubt on the deontologist’s explanation of thedifference in their intuitions about Switch and Bridge. An alternativeexplanation of the difference is that the Bridge case is ‘upclose and personal’ in the sense that it involves physicalcontact, whereas in Switch and Trap Door the agent is remote from theperson they have to kill to save the five (Singer, 2005). But thisdifference is morally irrelevant, so if this explanation is right,then our intuitions are distorted by at least one morally irrelevantfactor.

Also it seems that our intuitions are subject to framing effects. Forinstance, our intuitions seem to be affected by whether we word ourscenario in terms of killing or saving, and by the order in which thetrolley examples are considered. If people are asked to considerSwitch first, and Bridge second, they tend to say that it ispermissible to pull the lever in Switch but not permissible to pushthe man onto the track in Bridge. If, however, they are given theBridge case first a higher percentage say that it would be wrong to pull thelever in Switch. So it looks like the order in which the cases aregiven affects people’s intuitions about the cases. But the order inwhich one considers the cases is morally irrelevant. So it looks likeour intuitions can be distorted by a secondsource.^[10]

One thing worthy of note is that these cases test intuitions aboutour overall moral judgements—that is, about what we should, ormay do in certain circ*mstances. But not all intuitionists claim thatprinciples about what we ought to do are self-evident. W. D. Ross forexample claimed that only principles of prima facie duty areself-evident, and principles of prima facie duty are, roughly,principles stating that certain facts count in favour of an act andothers count against. So these principles state, for instance, thatthe fact that one’s act would produce some good, or the fact that itwould be the keeping of a promise, or the expressionof gratitude, etc., counts in favour of it, and the fact that, forexample, it would involve the infliction of harm on someone countsagainst it. What we ought to do is determined by all of these facts,and how they weigh up against each other. Ross denied that we can everknow what we ought to do, and rejected the view that there could bestrictly universal, self-evident principles specifying what we oughtto do.

It would be interesting to hear the outcome of experiments to seewhether people’s intuitions about prima facie duty are subjectto change on morally irrelevant grounds. This is something thatrequires empirical testing, but it is hard to imagine someone thinkingthat the fact that one would have to kill an innocent person in orderto save five didn’t count against doing this act, or that thefact that their act would save five innocent people didn’t countin favour of it, regardless of their overall verdict about whetherthey should kill the one or let the five die. That their act involvedphysically pushing someone in front of the trolley, or pulling a leverthat would release a trap door dropping them onto the track wouldplausibly make no difference to such intuitions. Nor would framingeffects introduced by the order of presentation of the cases. If sucha priori expectations are correct then empirical psychology wouldraise no problems for a Rossian intuitionism that claims only thatprinciples of prima facie duty are self-evident.

James Andow (2018) has put these claims to the test, and it turns outthat a significant number of people (25%) do not accept that the factthat someone would be killed by either pulling the lever or pushingthem off a bridge to save five counts against doing this, whether theyare imagining themselves doing the act or commenting on someoneelse’s action (121). He also discovered that people’sintuitions did change between the lever case and the bridge case inrelation to whether the fact that five lives would be saved counts infavour of doing the act (121). So it seems that intuitions aboutwhether saving lives counts in favour of acting is vulnerable toframing effects.

Andow argues that we shouldn’t dismiss these results on thegrounds that the responses did not reflect the participants’intuitions about the moral relevance of killing (138ff). But it ishard to accept that 25% of respondents really thought that the factthat their act involves killing someone does not even count againstdoing it. To believe this is not just to sincerely utter the words,but involves certain counter-factual beliefs. For instance, if I didnot think that the fact that by pulling the lever I would be killingsomeone counts against pulling the lever, then if I had a second leverwhich would save the five without killing anyone, I would beindifferent about which lever I’d pull. It is difficult tobelieve that the respondents who claimed that killing did not matterin trolley case scenarios would be indifferent in this counter-factualscenario. Furthermore, thinking that killing does not count againstone’s act commits one to the thought that there is absolutelynothing to regret. But it is likely that if these people were put in areal scenario where they could only save five by killing one, theywould regret deeply the fact that they could only save the five inthis way. None of this fits with the results of Andow’sexperimental findings, and casts doubt on whether therespondent’s replies really reflect their intuitions about thecase.

But even if Ross-style intuitions were not vulnerable to suchdisagreement and framing effects, critics may object that in so far asRoss’s theory does not tell us what we ought to do, it does notgive us what we want from a moral theory. Being told that variousfeatures count for or against certain actions, and that one just hasto decide for oneself in each case what one should do, may be a verydisappointing result even if it is self-evident which features countfor or against.

A further point about the findings of such experiments is whether thesubjects’ answers to the experimenters’ questions expresstheir intuitions (Bengson 2013). When subjects have considered Bridgefirst, they are more likely to say that it would be wrong to pull thelever in Switch. This is taken to show that their intuitions arevulnerable to framing effects. But once we remember that one’sbeliefs and judgements may conflict with one’s intuitions it isnot at all clear that the subjects lack the intuition that it would bepermissible to pull the lever in the switch case when they say that itwould be wrong. A perfectly plausible alternative is that they havereasoned that, because it would be wrong to kill someone to save fivein Bridge, it must be wrong to kill someone to save five in Switch.That is quite consistent with their having the intuition that it wouldbe permissible to pull the lever in the switch case, since intuitingis not believing. This point does not however explain JamesAndow’s findings.

1.5 Non-inferential justification?

Sinnott-Armstrong claims that results from empirical psychologyshow that most of our moral beliefs are false, because they have beenformed by an unreliable process (2006, 353). The unreliable process isbasing them on intuitions that are systematically distorted by morallyirrelevant factors, such as order or wording. Sinnott-Armstrongdoesn’t deny that some moral intuitions can justify moralbeliefs. But given that the default justification provided by so manyof them is undermined by distorting factors, we need to check that somemoral intuition is not one of the undermined ones before we can take itto provide justification. But then, he argues, the intuitions that doprovide justification do so only inferentially. Since moral intuitionseither don’t provide justification at all, or do so onlyinferentially, there is no non-inferential justification for our moralbeliefs and intuitionism is false.

Nathan Ballantyne and Joshua C. Thurow (2013) maintain that thisargument does not work. They outline their point in terms ofundercutting defeaters and defeaters of those defeaters. The distortingfactors that Sinnott-Armstrong mentions are the undercutting defeatersof the justification of most of our moral beliefs. If we have evidencethat a subclass of our moral beliefs is not subject to theseundercutting defeaters, then that evidence defeats the defeater, andjustification is restored.

With this jargon in hand they argue that Sinnott-Armstrong’s argumentconflates the belief’s justification being supported by aninference with the belief itself being supported by aninference. Let U signify the undercutting defeaters for a moralbelief, D the evidence that defeats these defeaters, and B themoral belief. Sinnott-Armstrong argues that for D to defeat U andthus restore the justification for B, D must provide the agentwith a reason to believe that B was formed reliably. But,Ballantyne and Thurow maintain, D can defeat U independently ofthe agent’s ability to provide arguments for his belief. My belief isjustified only on the basis of the following argument:

D,
D defeats U,
therefore, U is defeated.

So this inference supports B’s justification. But B itself isnot supported by that inference. B is, they claim, supported by therelevant intuition alone. (414). Being able to argue that the original,non-inferential justification for B has not been undermined, does notentail being able to argue for B. All this argument does is restore theoriginal non-inferential justification for the belief.

Ballantyne and Thurow illustrate this point with the followingnon-moral example.

McCoy visits the local widget factory and sees what seems to be ared widget being carried along a conveyor belt. He believes that thewidget is red…. Soon enough, a stranger approaches McCoy andsays that the widgets are actually white but are illuminated by redlights… Upon seeing this conversation, anotherstranger—who seems to McCoy to be a factory employee—tellsMcCoy not to listen to the other stranger: he is a trickster, McCoy istold, who likes to mess around with visitors. (2013, 413)

Sinnott-Armstrong would say that in order for the undercuttingdefeater (the 1^st stranger’s comment) to be defeated by thefactory worker’s claim that the stranger is a trickster, the factoryworker’s comment must provide McCoy with a reason to think that hisbelief was formed reliably. But, Ballantyne and Thurow claim, all thathas happened is that the original, non-inferential justification hasbeen restored. “Whatever initially made B justified continuesto justify B once the defeater has been defeated. McCoy’s beliefthat the widget is red is justified by his perceptual experience orseeming” (414). The same can be said of moral intuitions. I mayregard my moral belief as justified by an intuition with the samecontent if I am justified in believing that potential undercuttingdefeaters have been defeated (or are absent). In such a case I have aninferential argument for the justification of my moral belief, butthat does not mean that I have an inferential justification for mymoral belief. All that has happened, is that the original,non-inferential justification (provided by the intuition) has beenrestored.

Nonetheless, Ballantyne and Thurow do not question the first part ofSinnott- Armstrong’s argument—that is, the claim thatbecause of partiality, bias, emotion and disagreement, we have goodreason to think that most of our moral beliefs are false. If it istrue, as Sinnott-Armstrong claims, that most of our moral beliefs arefalse, then intuitionists’ trust in our ordinary moral thoughtwill look undermined regardless of whether they can salvage a few soundmoral beliefs from the wreckage. So intuitionists would not only haveto argue that some intuitions justify non-inferentially, but that thereis a significantly large group of reliable intuitions to validate theirmethodology, and Ballantyne and Thurow’s argument does not helpwith that.

2. Intuitionist metaphysics

2.1 Indefinable, non-natural properties

Along with its moral epistemology, a distinctive feature ofintuitionist thought is its non-naturalist realism. Intuitionistsmaintain that moral judgements are cognitive states, and that some atleast of these judgements are true. They are true when the thingsreferred to have the moral property that is ascribed to them by thejudgement. The moral properties that intuitionists tended to focus on werethe thin moral properties of goodness and rightness. These propertiesare, they maintained, simple, non-natural properties. It is not alwaysclear how they understood the notion of a non-natural property (more onthis below), but for now we can say that they denied that moralproperties can be defined wholly in terms of psychological,sociological, or biological properties. Some intuitionists allowed thatgoodness can be defined in terms of rightness (Sidgwick and Ewing) orrightness in terms of goodness (early Moore). But all intuitionistsmaintained that at least one of these moral properties is simple, orindefinable.

Although their view is about the nature of moral properties, theyoften put their point in terms of moral concepts or ideas, andmaintained that these concepts are either unanalysable, or ifanalysable, not analysable wholly in terms of natural concepts. Itseems they assumed that if a concept was indefinable then itscorresponding property would be indefinable, and vice versa. Manyphilosophers today would deny this assumption.

The simplicity of moral properties, such as rightness and goodness,and our ideas of them, was important for early intuitionists such asPrice, as he accepted the empiricist doctrine that simple ideas cannotbe invented but must be acquired by immediate intuition, as all simpleideas must “be ascribed to some power of immediateperception in the human mind” 1758/1969, 141), i.e., eithersensibility or theunderstanding.^[11]Consequently, the ideas of right andwrong must be immediate perceptions of either sensibility or theunderstanding.^[12]

If right and wrong are just feelings of approval or disapproval causedin us by natural properties or objects, then the idea of right andwrong will be given by our senses, for these ideas will be merely theeffect that the perception of certain things has on sensibility. If,however, right and wrong were real properties of actions, then theycould not be apprehended by any empirical sense, for we have no suchsensation of right and wrong when we apprehend right or wrong actions.Rather, what we see is that these actions are right, orwrong.^[13]This seeing still counts asintuition, as it is an immediate apprehension, but it is intellectualrather than sensibleintuition.^[14]

It is scarcely conceivable that anyone can impartially attend to thenature of his own perceptions, and determine that, when he thinksgratitude and beneficence to be right, he perceives nothingtrue of them, and understands nothing, but onlyreceives an impression from sense. (Price, 1758/1969,144–5)

Price concedes that certain feelings may attend our apprehension ofright and wrong, but these impressions are merely the consequence ofour perception of right and wrong. They are not what is perceived. ForPrice I approve of some act because I see that it is right or good.

But even if Price is right that the ideas of right and wrong aresimple, and are grasped by the understanding, that does not imply thatthey are non-natural. For he allows that there are simple ideas ofnatural properties, and that some of these, such as causality andequality, are grasped by the understanding rather than sensibility. Soa separate argument is needed for the non-natural nature of moralproperties.

Moore is the intuitionist who laid most stress on the non-naturalnature of moral properties, though his focus was on goodness ratherthan rightness. In Principia Ethica Moore defines a naturalproperty as one that can exist by itself in time and not merely as aproperty of some natural object (1903/1993a, 93). The idea is, then,that natural properties, such as the pleasantness or squareness of anobject, can exist independently of that object, whereas the goodnessof a good thing cannot exist independently of that thing.

This definition can be understood in terms of particular instancesof some property or in terms of the universal—the propertyitself. Either way it does not distinguish natural from non-naturalproperties as Moore thought. It does not seem that the particularinstance of redness in some particular red object could exist apartfrom that object any more than the particular instance of goodness ofsome good thing could. A particular instance of any property is a waythat something is, and the way that some particular thing is cannot beseparated from the particular thing that is that way.

But things are no better if we understand Moore’s definition interms of universals, for on a Platonist theory of properties theuniversal could exist independently of any particular instance of it.If this is true, it will be true of any property. This does not,therefore, pick out what distinguishes the natural from thenon-natural. So if we are talking of property instances then noproperties can be separated from the things that instantiate them, andif we are talking of property types (properties as universals), then,at least on some views, all of them can be separated.

Moore himself later abandoned this definition of the distinctionbetween natural and non-natural properties, and described his earlieraccount of a natural property as “utterly silly andpreposterous” (1942, 582). In the Preface to the second editionof Principia Moore offers an alternative definition that issuggested in chapter two of Principia. According to thisdefinition, a natural property is one “with which it is thebusiness of the natural sciences or of Psychology to deal” (13).Since the term to be defined (‘natural’) appears in thedefinition, this definition may not seem very informative. But we canreplace the term ‘natural sciences’ with ‘empiricalsciences’ (understood to include psychology and sociology) toget a useful and workable epistemological definition of a naturalproperty. On this account, then, natural facts can be known by purelyempirical means, whereas non-natural moral facts cannot be known inthis way. Such facts involve an essentially a priori element.

Intuitively the intuitionists seem right. Empirical investigation cantell us many things about the world, but it does not seem that it cantell whether certain acts are right or wrong, good or bad. This is notto say that our moral views are not revisable in the light ofempirical findings. For instance if science told us that alobster’s neurological system is sufficiently advanced for it tofeel pain, we’d revise our view about the permissibility ofboiling them alive. But all that science would have told us is thatlobsters feel pain when boiled alive. Science does not inform us thatboiling them alive is wrong. That seems to be something that cannotbe known empirically.

Moore’s open question argument can be regarded as giving form to thisintuition. If the moral property of being good, for instance, could bedefined in wholly psychological, biological, or sociological terms,then moral truths would turn out to be either psychological,biological or sociological truths, which could then be discovered byempirical research by the appropriate science. But, Moore argues, allsuch definitions must fail, for it is always an open question whetherthe thing that has the relevant empirical property is good. Moore’sargument can be captured as follows:

1. If some property Fcould be defined in terms of some other property G, then thequestion “is something which is G, F?” would beclosed.
2. For any naturalisticdefinition of goodness, it would always be an open question whethersomething that has the relevant natural property is good.
3. Therefore, goodnesscannot be naturalistically defined.

An open question is one that is not closed, and a closed questionis one the asking of which betrays a lack of understanding of theconcepts involved. For instance, if I asked “Jones is a widow,but was she ever married?”, this would show that I don’t reallyunderstand the term “widow”. So this question is closed.Moore claims that we can test any naturalistic definition of goodnessby asking whether something that has those natural properties is good,and then seeing whether this question is open or closed. If thedefinition is true, then the question must be closed, so if it isopen, the definition must be false.

Suppose, for instance, someone proposes that goodness can be definedin terms of causality and pleasure. To be good, they claim, is just tocause pleasure. Moore’s view is that if this definition werecorrect, it would be a closed question whether something that causespleasure is good. For in effect one would be asking whether somethingthat causes pleasure causes pleasure, and that is clearly a closedquestion. But, Moore insists, the question “is somethingthat causes pleasure good?” is an open question. One could,without conceptual confusion, debate whether something that causespleasure is good. So goodness cannot be defined as causes pleasure.

Moore assumes that this will be true of every putative naturalisticdefinition of goodness, whether it be in terms of second-order desires,social approval, being more evolved, or whatever. All of thesenaturalistic definitions would fail the open question argument. If heis right, and goodness cannot be defined wholly with reference toconcepts from the empirical sciences, then goodness is a sui generisnotion, i.e., is one that can only be understood in its own, evaluativeterms.

Moore’s argument has a great deal of intuitive force, but hasbeen subject to various objections, and it is not clear that all ofthem can be answered. One of the first was that it just begs thequestion against the naturalist. Moore only considers a few very crudenaturalistic definitions of goodness, and concludes from these that allnaturalistic definitions will fail the open question argument. Frankena(1939) objected that this was premature. We cannot know in advance thatevery naturalistic definition will fail this test. We just have to waitand consider the proposals. To conclude from a few crude examples thatall naturalistic definitions will fail is just bad induction.

Another objection is that the open question argument does not tell usanything distinctive about the concept of goodness, but is simply aninstance of the paradox of analysis. According to this paradox any trueanalysis will be uninformative, because it will be reducible to atautology, and any informative analysis will be false, because itcan’t be reduced to a tautology. According to earlier versions ofwhat Moore called ‘the open question argument’, and on oneinterpretation of Moore’s argument, the reason that good cannotbe analysed in naturalistic terms is that it turns what sounds like asubstantive moral claim, e.g., that pleasure is good, into the emptytautology that pleasure is pleasure. But then this looks like aparticular instance of the paradox of analysis. To see this, considerthe following analysis of the concept ‘mammal’:

(M)

A mammal is a member of a species of which thefemales suckle their young.

This looks informative as it can tell us why whales and duck-billedplatypuses are mammals when they are so different in seeminglysignificant ways from other mammals. But if the analysis is true, and‘mammal’ just means ‘member of a species of which thefemales suckle their young’, then (M) means:

(T)

A member of a species of which the females suckle theiryoung is a member of a species of which the females suckle theiryoung

(T) is, however, just an uninformative tautology. This is a quitegeneral problem in the theory of analysis, so if it applies toseemingly informative analyses of goodness, then that would revealnothing distinctive about naturalistic analyses of moral terms.

Furthermore, some analyses are not obvious. The analysis of the conceptof a mammal is an example of a non-obvious analysis. It is for thatreason that the question “A is a member of a species the femaleof which suckle their young, but is A a mammal?” will seem open,even though it is a true analysis. Similarly, a non-obviousnaturalistic definition of good may fail the open question test eventhough it is true.

2.2 Analyses of concepts and property identity claims.

A naturalist might accept that the open question argument works inrelation to moral concepts, but deny that we can make any inferencesabout the way the world is from the fact that we think of it in certainways.^[15]For the way in whichwe think about the world is determined by our understanding of theconcepts we use to describe it, and we cannot reliably infer that theworld is a certain way from the fact that we conceive of it in thatway. To think that one can make such inferences is to confusepredicates or concepts withproperties,^[16]analytical identities with synthetic identities.We know what we mean by certain concepts by a priori reflection, butthe nature of the things to which these concepts refer can only bediscovered by empirical investigation. We did not discover that wateris H₂O or that heat is mean kinetic molecular energy by a priorireflection on what we mean by ‘water’ and‘heat’, but by empirical investigation. Furthermore, wecould not object to the view that heat is mean kinetic energy on theground that this is not what we mean when we think of something as hot.But the intuitionists seem to object to naturalistic accounts of moralproperties in precisely this way.

There is, however, reason to think that intuitionists such as Moore andRoss did not confuse concepts and properties. For they were careful todistinguish an elucidation of the meaning of words and an account ofthe nature of the world with their distinction between proper verbaldefinitions and the sort of definitions they are interested in (viz.,metaphysicalones).^[17]A proper verbal definition of ‘good’ is simply an account ofhow most people use the word, whereas a metaphysical definition is onethat tells us the nature of the thing of which the concept is aconcept.^[18]This is notquite the distinction between an analysis of a concept and an accountof the nature of a corresponding property, but it is close enough togive us reason to suppose that intuitionists such as Moore and Rosswere aware of the distinction between concepts and properties that manythink they simply conflate.

But although intuitionists may not have confused concepts andproperties, they did seem to believe that there is a certainisomorphism between the structure of our concepts and the nature of theworld, such that a proper analysis of our concepts would reveal to usthe nature of the corresponding property or thing. This belief is notobviously confused, but the examples of heat and water seem to showthat it cannot be accepted as it stands. Intuitionists need not,however, rest their view about the property of goodness on a generalthesis about the relation of concepts and properties. All they need dois identify what it is about certain concepts, like the concepts ofwater and heat, that provides us with reasons to think that thecorresponding properties are different, and then argue that thesereasons do not apply to the concept of goodness.

With concepts of natural properties and substances like heat and waterwe have two reasons for thinking that the corresponding properties maybe different. First, the concept of heat seems metaphysicallysuperficial and incomplete. It is the concept of a property that hascertain characteristic effects on us and on other things, but does notaim to tell us about the nature of the property that has those effects.The concept of water seems superficial in the same way. This conceptonly picks out certain surface features of water, such as its beingclear, odourless, tasteless, etc. It does not, however, tell usanything about the nature of the substance that has these features. Inboth cases empirical science seems well-suited to complete this pictureby investigating the property or substance that has these distinctiveeffects, or surface features. In doing this, empirical science providesus with an account of heat and water which is metaphysically deeperthan the one provided by the corresponding concepts.

Secondly, even if the concept of heat were not incomplete orsuperficial, in so far as it is a concept of a natural property we havegood reason to think that the empirical sciences are much betterequipped to discover the nature of heat than a priori reflection. Thesame is true of the concept of water. In so far as this is a concept ofa natural substance, the empirical sciences are far better suited totell us the nature of this substance than a priori reflection.

These reasons do not apply to the concept of goodness. First, thisconcept does not seem to be metaphysically superficial or incomplete inthe way that the concept of heat or water is. When we think ofsomething as good we do not think of it merely as having certaineffects on us, or as picking out certain surface properties theproperty of goodness has, but think of it as having a distinctivecharacteristic. Not all intuitionists agreed with Moore that nothingcould be said about the nature of this characteristic (though they allagreed that this is a non-natural property). A. C. Ewing, for example,maintained that the characteristic we have in mind when we think ofsomething as being good is the property it has of being the fittingobject of a pro-attitude. If this, or something like it, is correct,then the concept of goodness does not merely describe certainproperties goodness has, but aspires to tell us whatgoodnessis.^[19]It does not, therefore,call for a metaphysically deeper account from some other source in theway that the concept of heat or water does.

It is not, however, clear that this argument will persuade critics ofintuitionism and nonnaturalism. They may want more from an analysis of‘good’. They might, for example, want an analysis thathelps explain why some things rather than others are good, and whichexplains the connection between the properties that make somethinggood, and its goodness. Without these explanatory features they maywell regard the analysis of good offered by Ewing superficial and inneed of a metaphysically deeper account.

Indeed, as Robert Shaver points out (2007, 289) according to oneintuitionist account of good the analysis does call for ametaphysically deeper account of the nature of the property. C. D.Broad, for instance, analyses good as meaning“there is one andonly one characteristic or set of characteristics whose presence in anyobject that I contemplate is necessary to make me contemplate it withapproval” (1985, 283). Critics may claim that this analysis isjust as plausible as Ewing’s, and leaves room for an alternative,naturalist account of the property that explains my approval.

Shaver also points out that it is a mistake to assume that syntheticidentities can only be established by empirical means. This is amistake, because one could arrive at the conclusion that two differentnotions refer to the same property by a priori reflection. So even ifno empirical investigation can show that a moral and a non-moral termpick out the same property, this might still be shown by a priorireflection.

2.3 Queerness

Some philosophers think that there could be no moral facts asintuitionists understand these. This is because such facts would beunlike any other facts of which we know. Such philosophers think ofnon-natural facts and properties as ‘queer’ (see, Mackie1977; Joyce 2001; Olson 2014). This queerness is probably at the heart of manyphilosophers’ uneasiness about the idea of a non-naturalproperty. But we need to be clear about what it is that is supposed tobe so queer about the non-natural nature of goodness as intuitionistsunderstand it.

The intuitionist conception of goodness may be regarded asmysterious because it is alleged to be unanalysable or indefinable. Butthat wouldn’t pick up on the non-natural nature thatintuitionists claim moral properties have. This objection would applyto any notion that philosophers claim is primitive in the sense that noinformative definition in other terms can be offered, be it the notionof explanation, knowledge, or pain. It expresses rather a certainphilosophical disappointment that a definition can’t be offered.It is often said that “One person’s primitive is anotherperson’s mystery”, but this supposed sense of mysteryattaches to the unanalysability claim rather than the non-naturalnessclaim.

Furthermore, some intuitionists did not think that goodness isunanalysable. For instance, Sidgwick thought that good could beanalysed as what ought to be desired, and Ewing maintained that itcould be analysed as the fitting object of a pro-attitude. If the senseof mystery of the intuitionist notion of goodness stems from itsunanalysability, then this sense of mystery will not apply to theseintuitionists’ concept of goodness, even though it is still anon-naturalist conception. But critics might respond that thesedefinitions really only move the mysterious notion elsewhere, e.g., tothe unanalysable terms ‘ought’ and‘fittingness’.

John Mackie maintained that moral properties, understood broadlyalong intuitionist lines, are queer because they are inherentlymotivational, in the sense that when we come to see that some act isgood, we are motivated to do it. No other property we know of hassuch inherent motivational force. If such properties could beunderstood naturalistically, e.g., as being such as to elicit desire inthose who perceive them, then the inherent motivational force of moralproperties would not be queer. For if something’s goodness wereidentical with its being such as to elicit desire when perceived, itwould be no surprise that we come to desire it when we perceiveit. But intuitionists’ non-naturalism rules out thisexplanation of the intrinsic ‘to be pursuedness’ of moralproperties.

Intuitionists may respond to this objection by drawing on a recentversion of Ewing’s fitting attitude analysis of goodness. Ewingthought that to be good is to be the object of a fitting pro-attitude.T. M. Scanlon has argued that goodness is to be understood assomething’s having properties that give us reason to have apro-attitude towards it (1998, 95), and like the intuitionist viewabout goodness and rightness, he thinks that the notion of a reasoncannot be understood in other, non-normative terms (1998, 17). So hisview is something intuitionists can accept. If they do, then therewould seem to be no mystery about the magnetism of the good. If comingto see that something is good is coming to see that we have reason tohave a pro-attitude towards it, then it would be no surprise ifrational individuals come to have a pro-attitude towards perceivedgoods, any more than it would be surprising if rational beings come todo what they judge they ought to do. If one has Humean leaningsone might want an explanation of how a judgement that one has reason tohave an attitude, or that one ought to do something, can by itselfmotivate. But then the problem is not with the non-natural nature ofmoral properties, but is one within moral psychology, and involves thedebate between those who endorse a Humean theory of motivation, andthose who deny this.

But the mystery may be normative rather motivational if we assume,following Kant, that moral reasons are categorical reasons. Categoricalreasons are ones that apply to us independently of what we careabout. One might doubt that there are any such reasons. Oneargument for such a view is that normative practical reasons must bethe sort of thing from which we can act. So they must be able tomotivate us to act, and they can only do that by latching ontosomething we care about. But if all practical reasons must be able tolatch onto something we care about, then no reasons are categorical inthe Kantian sense. So the very idea of a moral reason may be quitemysterious and queer.

A final mystery is epistemic. It may be maintained that it is quiteunclear how we can know of moral facts. This mystery may stem from theidea that non-natural properties lack causal powers. So the mystery ishow we could know of something that is causally impotent. Intuitionistsmight respond by asking why a property must be causally efficacious ifwe are to be able to know which things have that property. They maypoint out that, on some views, dispositional properties such as warmth,fragility, or colour properties, lack causal powers. These powers arelocated in the non-dispositional base properties on which thedispositional properties supervene, rather than in the dispositionalproperties themselves. If this view were true, that would not implythat we could not know whether something was warm, fragile, orcoloured. And if that could be true of non-causal, dispositionalproperties, then it could be true of causally impotent moralproperties.

It may, however, be pressed that the causal impotency of moralproperties causes problems for the analogy of intellectual seemingswith perceptual seemings. Perceptually things seem to be a certain way,say red or square, because those things and their properties causallyinteract with our perceptual system. But if moral properties arecausally impotent, then causal interaction could not explain whycertain things seem to the intellect to be true.