iii-how we know what ought to be

iii*—how we know what

ought to be

by Ralph Wedgwood

abstract This paper outlines a new approach to the epistemology ofnormative beliefs, based on a version of the claim that ‘the intentional isnormative.’ This approach incorporates an account of where our ‘normativeintuitions’ come from, and of why it is essential to these intuitions that theyhave a certain weak connection to the truth. This account allows that theseintuitions may be fallible, but it also seeks to explain why it is rationalfor us to rely on these intuitions in forming normative beliefs—althoughit is also rational for us to try to correct for these intuitions’ fallibility byrevising our normative beliefs in such a way as to approach what Rawlscalled ‘reflective equilibrium’.

I ntroduction. How can we know, or have rational or justifiedbeliefs in, normative propositions—that is, propositions

about what ought to be the case, about what people ought to door choose, or about what they ought to believe, or ought to feel?Answering this question is the central task for the epistemologyof normative belief. In this paper, I shall outline a new answer tothis epistemological question.

This new epistemology of normative belief is designed to becompatible with a robustly realist conception of the normative.According to such a realist conception, normative statementsexpress perfectly straightforward beliefs or ‘cognitive states’,just like paradigmatic examples of ‘factual statements’; andwhen these statements are true, they are true in virtue of acorresponding normative fact; and there may be no way ofreducing such normative facts to facts that could be stated, evenin principle, in wholly non-normative terms.

Anti-realists often argue that realist conceptions of thenormative cannot give any satisfactory epistemology fornormative beliefs. Many realists say little more than that wehave some cognitive faculty—sometimes called ‘intuition’ or‘reason’—which enables us to come to know and have justified

*Meeting of the Aristotelian Society, held in Senate House, University of London,on Monday, 7th November, 2005 at 4.15 p.m.

64 ralph wedgwood

beliefs in normative propositions; but they rarely give anyaccount of how exactly this alleged faculty operates, or howit could serve as a reliable source of knowledge, or whatcould justify us in relying on it.1 For this reason, anti-realistsoften accuse realists of failing to meet legitimate demands forexplanation.2 The account that I shall outline here is designedto help the cause of a realistic conception of the normative, byshowing how realists could meet this explanatory demand.

This account will be based on a certain version of the claimthat is often made in the philosophy of mind, that ‘the intentionalis normative’—that is, that there is no way to give an accountof the nature or essence of mental states that have intentionalcontent without using normative terms, or at least withoutmentioning normative properties and relations. I shall not giveany defence of this claim here.3 So the overall conclusion of theargument that I shall give here can be taken as a conditional: ifthe intentional is normative in this way, then it would be possibleto acquire knowledge of, and justified belief in, normativepropositions in the way that I shall outline.

As I shall argue then, the claim that the intentional isessentially normative can help to provide an illuminatingepistemology of normative beliefs. In a way, this is not surprising.

1. One notable recent exception is Peacocke (2004, Ch. 7). According to Peacocke,our possession of moral concepts may consist in an ‘implicit conception’ of the moralproperties and relations that those concepts stand for (so that under favourablecircumstances, our moral intuitions are causally explained by this implicit conception,in which case they may count as ‘entitling’ us to the corresponding moral beliefs).Unfortunately, I cannot discuss Peacocke’s view here. For a first attempt atexplaining why I disagree with his view, see Wedgwood (2001, 27–29).2. This objection was famously pressed by Mackie (1977, 38): ‘If we were aware of[objective values], it would have to be by some special faculty of moral perceptionor intuition, utterly different from our ordinary ways of knowing everything else.’Some philosophers seem to believe that it is sufficient to answer this objection just todeny that our immediate normative or moral judgments arise from any such ‘specialfaculty’. This line has become common among philosophers who seek to rehabilitatea broadly intuitionist view, according to which certain basic normative propositionsare self-evident—that is, one is justified in believing these propositions purely in virtueof one’s ‘understanding’ of those propositions. For this view, see e.g. Shafer-Landau(2003, Ch. 11), Crisp (2002, 57–59), and Stratton-Lake (2002b, 22). But this seemsnot to meet the explanatory demand: how exactly does one’s ‘understanding’ of anormative proposition justify one in believing it? Why is one entitled to regard one’sapparent intuitions as reliable? So far, the intuitionists have not given much in theway of an answer to these questions.3. I try to defend this claim in a forthcoming paper (Wedgwood forthcoming).

how we know what ought to be 65

According to this claim, normative truths are part of whatmake mental states what they are: in effect, normative truthsare already built into our minds, simply because they are mindsat all; so it is not surprising that there is less of a problem aboutour ‘access’ to such irreducible normative truths than there mightseem.

In this paper, I shall only give an account of the epistemologyof normative propositions of a certain specific sort—namely,propositions about which mental states or attitudes are ‘correct’and which are ‘rational’. It may be that several other normativeconcepts are definable in terms of the concepts of a mental state’sbeing ‘correct’ or ‘rational’; for example, it may be that theconcept of something’s being admirable is just definable as theconcept of its being correct to admire the thing in question. Ifthat is so, then an account of how we can know, or have rationalor justified beliefs in, propositions about which mental states arecorrect or rational would also be able to explain how we canknow a wider range of normative propositions as well. But inthis paper, I shall not be able to investigate whether these othernormative concepts are definable in this way; I shall just focus onthe epistemology of beliefs about which mental states are correctand which are rational.

II

A Normative Theory of Mental States. The claim that ‘theintentional is normative’ has been interpreted in many ways.4

Here I shall outline a certain specific version of that claim. (To fixideas, the version that I shall outline is slightly more specific thanI actually need in order to support my epistemological accountof normative beliefs. I shall explain later how my argument couldbe generalized to apply to some other versions of this claim.)

Intentional mental states can be classified along twodimensions. First, they have some sort of content, which I shallregard as in many cases composed of concepts. Thus, the content

4. See especially Brandom (1994). In a broadly similar way, Morris (1992) offersan ‘evaluative theory of [intentional] content’. For an argument for a normativetheory of linguistic meaning, see Lance and Hawthorne (1997). For some discussionof Brandom, see especially McDowell (1997), Gibbard (1996), and Rosen (1997 and2001).

66 ralph wedgwood

of the belief that snow is white (and of the hope or the fear thatsnow is white) is the content ‘snow is white’, which is composedby means of predicating the concept ‘ . . . is white’ of the concept‘snow’. Second, these intentional mental states involve a certainmental relation to that content; in many cases this mental relationcould be called an attitude (such as the attitude of belief or hopeor fear) towards that content.

According to the version of the claim that ‘the intentional isnormative’ that I shall outline here, accounts of the nature ofthe various concepts, and of the various mental relations andattitudes that thinkers can have towards contents, are both to begiven in terms of the normative principles that apply to mentalstates involving those concepts or attitudes.

More specifically, according to this version of the claim that‘the intentional is normative’, the nature of each concept isgiven both by the principle that defines when beliefs involvingthat concept are correct, and also by certain basic principlesof rationality that apply to that concept. One main differencebetween the notion of a ‘correct’ attitude and the notion ofa ‘rational’ way of thinking is the following. Whether or notan attitude is correct is typically determined by some relationbetween that attitude and the external world. By contrast,whether or not a way of thinking is rational, for a thinker ata given time, is determined purely by the essential nature of thatway of thinking and its relation to the antecedent mental statesthat the thinker has at that time.

Assuming that a belief is correct if and only if the content ofthe belief is true, the conditions under which beliefs involvinga concept are correct would in effect determine the concept’ssemantic value—the contribution that the concept makes to thetruth conditions of contents in which it appears. On the otherhand, the basic principles that specify certain ways of using theconcept as rational (or specify certain other ways of using theconcept as irrational) would determine what we could call—to adapt a term from Frege (1892, 25)—the concept’s cognitivesignificance.

Thus, for example, the nature of logical concepts, like ‘or’and ‘not’, might be given both by their semantic values—theircontribution to the truth conditions of contents in which theyappear—and also by the basic principle that it is rational to


make inferences in accordance with certain fundamental rules ofinference for these logical concepts. The nature of the concept‘ . . . is yellow’ might be given both by the concept’s semanticvalue—the property of yellowness that the concept stands for—and also by the basic principle that it is rational to make ajudgment applying this concept to some perceptually presentedobject if one has a visual experience that represents that objectin a certain distinctive way (and one has no positive reason forthinking that one is not perceiving properly in the circumstances).

According to the version of the idea that ‘the intentional isnormative’ that I am outlining here, it is not just concepts whosenature is given by the normative principles that apply to them;the same is true of the various types of attitude as well. Thus,for example, perhaps the nature of the attitude of belief is givenboth by the principle that a belief is correct if and only if thecontent of the belief is true, and by some related principle aboutwhich ways of forming or revising beliefs count as rational. Forexample, this related principle might be the principle that a wayof forming beliefs counts as rational just in case given one’santecedent mental states, it is sufficiently likely (at least in somesense of the term ‘likely’) that any beliefs that are formed in thatway are correct.

This pair of principles illustrates a general relationship thatseems to hold between the notions of correctness and rationality.In general, it is rational to form a mental state in a certain way ifand only if, given one’s antecedent mental states, it is sufficientlylikely (in the relevant sense, whatever exactly that may be) thatthe mental state in question is correct. There are notorious dif-ficulties (which cannot be investigated here) about how best tointerpret terms like ‘likely’ in various contexts. But I propose tointerpret this occurrence of ‘likely’ in such a way that if a propo-sition p is likely, given one’s antecedent mental states, then therewill be something strange and unusual about any case in whichall of those antecedent mental states are correct, but p is not true.

This approach could also be applied to other types of attitudeas well. Thus, perhaps the nature of the attitude of admirationis given by the following two principles. First, it is correct toadmire something if and only if the object of one’s admirationreally is admirable in some suitable way. Secondly, it is rationalto admire something if and only if, given one’s antecedent mental

68 ralph wedgwood

states, it is sufficiently likely that the object of one’s admirationreally is admirable. A similar story may also be true of the moralemotions. For example, it may be that it is correct to have themoral emotion of disapproval towards an action if and only ifthe action in question really is morally wrong, and it is rational todisapprove of an action if and only if it is sufficiently likely, givenone’s antecedent mental states, that the action is wrong. Theseprinciples, about when the attitude of disapproval is correct andwhen it is rational, may be an essential part of what makesdisapproval the kind of attitude that it is.

This then, roughly, is the core of this normative theoryof intentional mental states—an account of the nature of thevarious concepts, and of the various attitudes that we can havetowards intentional contents, in terms of the normative principlesthat apply to certain mental states involving these concepts orattitudes. But this core will need to be surrounded by a shellthat explains what it is for a thinker to possess those concepts,or to be capable of those attitudes. If this is the correct accountof the nature of these concepts ‘or’ and ‘ . . . is yellow’, and ofsuch attitudes as belief or admiration or disapproval, what hasto be true of a thinker if she is to possess these concepts or to becapable of having these attitudes?

One plausible answer to this question is that the thinker musthave some dispositions that constitute an appropriate sort ofsensitivity to these normative principles that give the nature ofthe mental states in question.5 In most cases, it will be moreplausible to suppose that the thinker must have a dispositionto conform to the principles of rationality that give the natureof the relevant concepts or types of attitude than that she musthave a disposition to conform to the corresponding principles ofcorrectness. As I pointed out earlier, whether or not a mentalstate is correct is typically determined by the relation betweenthat mental state and the external world; and one’s dispositionsto have many sorts of mental state do not respond directly tothe external world, but only to one’s antecedent mental states.

5. In effect, this is a principle of interpretive charity rather like the principle that wasfamously advocated by Davidson (2001, Essay 9). We could invoke such a principlewithout accepting Davidson’s full-blown ‘interpretivism’. As Lewis (1974) suggests,the reference to interpretation could just be taken as a way of dramatizing what isobjectively constitutive of the intentional states in question.


Nonetheless, as I noted earlier, there seems to be a generalrelationship between correctness and rationality: it is rational forone to form a mental state in a certain way if and only if, givenone’s antecedent mental states, it is sufficiently likely that themental state in question is correct. So, if one has a dispositionto conform to a principle of rationality that applies to mentalstates of a certain type, this disposition can be regarded as anindirect way of being sensitive to the conditions under whichmental states of that type are correct.

Specifically, then, the sort of disposition that a thinker musthave, if she is to possess a given concept, is a disposition tothink in ways that the principle of rationality that figures inthe account of the nature of that concept specifies as rational.Such principles typically take the form of specifying some set ofantecedent mental states such that—according to this principle—it is rational to respond to being in those antecedent mental statesby forming a certain further mental state. Then, the dispositionthat one must have, in order to possess the concept, will be adisposition to respond to one’s actually being in those antecedentmental states by forming that further mental state. So, forexample, the thinker must be disposed to form a judgmentapplying the concept ‘ . . . is yellow’ to a perceptually presentedobject whenever she has a visual experience that represents theobject in the relevant way (at least if the question of the object’scolour arises, and the thinker has no positive evidence that sheis not perceiving properly in the circumstances).

A similar point would apply to the principles of rationalitythat figure, not in an account of the nature of a concept, but inan account of the nature of some type of attitude. Suppose forexample that the basic principle of rationality that applies to theattitude of admiration is the principle that I suggested earlier—the principle that an attitude of admiration is rational if and onlyif, given all one’s antecedent mental states, it is sufficiently likelythat the object of one’s admiration really is admirable. Nowattitudes of admiration, like many other emotions, are typicallyreactions to some antecedent beliefs or experiences concerningthe object of admiration.6 (As it is sometimes said, such emotions

6. This is what Hume meant by claiming that the passions are ‘secondary impressions’rather than ‘original impressions’ (1739, II.i.1)—or as he also put it, ‘impressions of

70 ralph wedgwood

have ‘cognitive bases’.) I hear my friend Charles playing thepiano, and I react to this experience with a feeling of admirationfor his playing. The disposition that I manifest in reacting in thisway might be a disposition to respond to antecedent beliefs orexperiences that make it highly likely that a certain object reallyis admirable, by having an attitude of admiration for that object.Perhaps having this disposition is essential to the capacity for theattitude of admiration at all.

This is not to say that when I manifest this disposition,the concept ‘admirable’ actually features in the content of myexperience of Charles’s piano playing, or in the content ofmy emotion of admiration. I may have no thoughts explicitlyinvolving the concept ‘admirable’ at all: my attitude of admi-ration is focused on his playing itself; and my experience is justan experience as of his playing Bach’s French Suites, say, in acertain way.7 Nonetheless, it may be that as a matter of fact,given all my antecedent mental states, this experience makes itsufficiently likely that Charles’s playing really is admirable; andit is because my experience has this feature that I manifest thisdisposition, and respond to this experience (directly and withoutfurther ado) by having a corresponding attitude of admiration.

The claim that a thinker has a disposition of this kind does notimply that this disposition will be manifested in every possiblecase. A disposition can be defined by means of a function fromstimulus conditions to corresponding response conditions. Butthe claim that an object has the disposition does not imply thatwhenever the object is in one of the relevant stimulus conditions,it will always go into the response condition onto which thefunction in question maps that stimulus condition. It impliesonly that ceteris paribus, or normally, when the object is in one ofthese stimulus conditions, it will also go into the correspondingresponse condition. But the circumstances can fail to be normal,

reflection’ rather than ‘impressions of sensation’ (1739, I.i.2). It also seems to be whatLocke meant by identifying passions with ‘internal’ rather than ‘bodily’ sensations(1690, II.20.3). We should add, however, that such an emotion is not merely causallytriggered by the belief or experience to which it is a reaction; it is a psychologicallyintelligible response to that belief or experience (we might even say that the emotionis motivated by that belief or experience).7. Thus, Johnston (2001b, 228) is mistaken in his claim that ‘for Wedgwood, affectivestates can only be consequential upon basic evaluative beliefs.’


and cetera can fail to be paria. When this happens, there willpresumably be some explanation of what the interfering factorswere that caused things to go awry in this way. It is only whenall such interfering factors are absent that the disposition mustbe manifested.

Thus, this proposal—that everyone who is capable of having acertain type of mental state must have a disposition to a certainsort of rational thinking involving mental states of that type—does not imply that we are always rational. It does not even implythat irrational thinking is less common than rational thinking:it implies only that when we do not conform to these very basicrequirements of rationality, the situation was abnormal (ceterawere not paria), and so there must be some explanation of whatwent awry in that case.8 Indeed, it seems possible, in extremecircumstances, for there to be an agent, or perhaps even anentire community of agents, for whom the circumstances arealways abnormal, so that interfering factors always prevent thesedispositions from being manifested. Nonetheless, even thoughthese dispositions can be blocked or inhibited in this way,according to the proposal that I am outlining here, a thinkermust at least have these dispositions in order to be capable ofthe types of mental state in question.

Moreover, in addition to the various ways in which one’sdispositions towards the basic kinds of rational thinking can beblocked or inhibited by interfering mental factors, one may alsohave many other mental dispositions, which may be dispositionsthat it is sometimes irrational to manifest. (Indeed, we seem tohave quite stable dispositions towards various forms of irrationaland fallacious thinking.) This too is quite compatible with itsbeing the case that in general, one also has a general dispositiontowards certain basic kinds of rational thinking.

III

Primitively Rational Ways of Forming Beliefs. My main topichere is how we can know, or even form rational or justifiedbeliefs in, normative propositions (such as propositions about

8. Thus, I believe that this proposal is not vulnerable to the sorts of objections thatare raised by Stein (1996).

72 ralph wedgwood

which mental states are correct or rational). Later on, I shallsketch a certain way in which one might form beliefs in suchnormative propositions, and I shall argue that it is rationalto form normative beliefs in this way—and indeed that underfavourable conditions, beliefs formed in this way may count asknowledge.

In this section, however, I shall propose some thesesconcerning rational belief. These theses are suggested byconsidering why it is rational to form beliefs by taking one’ssensory experience at face value. As I shall use the phrase, ‘takingone’s sensory experiences at face value’ involves forming a beliefin a proposition p directly on the basis of a sensory experienceas of p’s being the case (so long as one has no reason to thinkthat one is not perceiving properly in the circumstances).9

I suggested above that it is rational to form a certain belief ina certain way if and only if it is sufficiently likely, given one’santecedent mental states, that the content of any belief that isformed in that way is true. Applied to the case of beliefs thatare based on sensory experience, this suggestion implies thatif it is rational to form a belief in a proposition p by takingone’s sensory experience at face value, then it must in somesense be ‘likely’, given one’s antecedent mental states, and giventhat one forms the belief in this way, that p is true. While thissuggestion seems plausible, it may well be that the relevant notionof ‘likelihood’ can only be explained at least partly in terms ofwhat it is rational to believe. But then this suggestion would nothelp us to explain why it is rational to form beliefs by takingone’s sensory experience at face value. For this reason, I shallnot rely on this suggestion in trying to explain why this way offorming beliefs is rational.

In some cases, it is rational to form a belief in a certain waybecause it is independently or antecedently rational for one tobelieve that forming that belief in that way is reliable—that is,it is something that could not easily happen that such a beliefformed in that way would fail to be true. For example, this seems

9. We may also have to require that the sensory experience in question should‘basically represent’ the proposition p. In some sense, an experience can represent anobject as a cathode ray tube, but such an experience may not suffice all by itself tomake it rational for one to form the belief that the object in question is a cathoderay tube. For this point, see Peacocke (2004, 65–69).


to be the case with beliefs that are based on the use of measuringinstruments. It is rational for me to form a belief about thetemperature of the air around me on the basis of reading athermometer; but this is only because it is already rational—independently of any beliefs that I form on the basis of thisthermometer—for me to believe that the thermometer is reliable.

However it would lead to an obvious infinite regress if wesupposed that it cannot be rational to form a belief in any wayunless it is already independently or antecedently rational tobelieve that it is reliable to form that belief in that way.10 Theremust be certain ways of forming beliefs that are just basicallyor primitively rational—that is, rational but not as a result ofits being independently or antecedently rational to regard thoseways of forming beliefs as reliable. It is plausible that it is in thissense basically or primitively rational to form beliefs by takingone’s sensory experiences at face value.

If a primitively rational way of forming beliefs is not rationalbecause it is independently rational to regard it as reliable,then why is it rational? Some philosophers might suggest thatthe explanation is simply that the disposition to form beliefsby taking one’s experience at face value is a deeply-entrenchedcognitive disposition, which plays a pivotal role in our wholepractice of forming and revising beliefs.11 But I do not see whythere could not be quite irrational belief-forming dispositionsthat are deeply entrenched in precisely this way. So I doubt thatthis sort of explanation can be sufficient.

By definition, a primitively rational way of forming beliefsis not rational because it is independently rational to regardthat way of forming beliefs as reliable. This seems to show thatsuch a primitively rational way of forming beliefs is not rationalbecause of any empirical evidence in favour of that way offorming beliefs. This seems to make it plausible that this wayof forming beliefs is simply necessarily rational, purely in virtueof the essential nature of that way of forming beliefs and of thevarious concepts and mental states that it involves. But how canthe nature of these concepts and mental states explain why it isrational to form beliefs by taking one’s experience at face value?

10. For this point, see Pryor (2000).11. For this sort of approach, see Alston (1991 and 1993).

74 ralph wedgwood

Part of the story may be that a disposition towards formingbeliefs in this way is constitutive of possessing some of theconcepts, or of having the capacity for some of the types ofmental state, that are involved in this way of forming beliefs. Butit is hard to see how this can be the whole story. According to thenormative theory of mental states that I am working with, thefact that the disposition is constitutive of such a basic mentalcapacity is itself explained by the fact that it is a dispositiontowards a basic form of rational thinking. So it is hard to seehow the fact that this way of thinking is rational could itself beentirely explained by the fact that the disposition to engage inthis sort of thinking is constitutive of some such basic mentalcapacity.12

What seems to be required is that there should be someconnection between the practice of forming beliefs by takingexperience at face value and the truth. But it cannot be requiredthat beliefs formed in this way should actually be true, sinceit seems that it is equally rational for one to form beliefs bytaking one’s sensory experiences at face value even if one isbeing undetectably deceived by an evil demon or a manipulativeneuroscientist, so that most of the beliefs that one forms bytaking one’s experiences at face value are not true at all.Indeed, there could even be a possible world in which the vastmajority of sensory experiences are produced by such deceivingneuroscientists; so it may not even be necessary that the practiceof taking experience at face value is generally reliable.

Nonetheless, it seems plausible that some such connection tothe truth is part of what makes it primitively rational to formbeliefs by taking experience at face value. As I have suggested,any such primitively rational way of forming beliefs is necessarilyrational, purely in virtue of the nature of the concepts and mentalstates involved. Thus, if this connection to the truth is part ofwhat makes this a primitively rational way of forming beliefs, thisconnection to the truth must hold necessarily, purely in virtue ofthe nature of concepts and mental states involved.

What sort of connection to the truth could this be? I proposethat the practice of forming beliefs by taking one’s sensory

12. In this way, I must amend the story that I sketched in an earlier paper of mine(Wedgwood 1999).


experiences at face value has the following essential connectionwith the truth. It may be that it is essential to experiences that anysubject has some disposition to have experiences that veridicallyrepresent certain aspects of her environment. More precisely,for every possible subject of experience, there is a range ofpropositions such that for every proposition p within that range,under normal conditions, the subject will respond to being in asituation in which p is the case by having a sensory experienceas of p’s being the case. Even if this disposition is blocked orinhibited by other factors (such as the machinations of an evildemon), the subject may still have some disposition to haveveridical experiences of this sort. Indeed, the fact that a mentalstate is of the kind that would be involved in one’s manifestingthis disposition in response to one’s being in a situation in whichp is the case may be an essential part of what makes this mentalstate count as a sensory experience as of p’s being the case inthe first place.13 Whenever one’s experience does consist in themanifestation of this essential disposition, then the content ofthe experience will be true.

It may be that this connection to the truth is part of theexplanation of why it is rational for one to form beliefs by takingone’s sensory experiences at face value (so long as one has nopositive reason to think that one is not perceiving properly in thecircumstances). We could not function as agents at all unless wehad some beliefs about our immediate environment. So it wouldbe unreasonable to demand that we should not form any suchbeliefs at all. It would also be unreasonable to demand that weshould not form any beliefs about our immediate environment inany way unless we have independent or antecedent reasons forregarding that way of forming beliefs as reliable; we could haveno such independent reason to regard a way of forming suchbeliefs as reliable unless there were some way of forming suchbeliefs that was simply primitively or basically rational. So theonly reasonable demand to make here is that we should not formbeliefs in a given way in spite of not having any independentor antecedent reason to regard that way of forming beliefs asreliable, unless that way of forming beliefs has some essential

13. Compare Peacocke’s (2004, 69) idea that experiences are ‘instance-individuatedwith respect to certain of their contents’.

76 ralph wedgwood

connection to the truth. If this proposal about the nature ofexperiences is correct, there is an essential connection betweenour sensory experiences and the truth. So this may help to explainwhy it is primitively rational to form beliefs by taking one’ssensory experiences at face value.

IV

The Source of Normative Intuitions. It clearly follows from theversion of the claim that ‘the intentional is normative’ that Ioutlined in the second section that the dispositions that areessential to the capacity for a given sort of mental state arereliable indicators of the truth of certain normative propositionsconcerning mental states of that sort. The disposition thatis essential to possessing the concept ‘or’, for example, is adisposition to use the concept in rational ways. In this way, theseessential dispositions are reliable indicators of the truth aboutwhich ways of thinking are rational. This is the crucial featureof this theory of the nature of intentional states for my purposes.Any other theory that also has this implication would serve thepurposes of my proposal about the epistemology of normativebelief equally well.

Moreover, in a more indirect way, these essential dispositionsare also reliable indicators of the truth about which mental statesare correct. For example, I suggested above that the dispositionthat is essential to the capacity for admiration is a disposition torespond to a set of antecedent mental states that makes it highlylikely that it is correct to admire a certain object, by havingan attitude of admiration for that object. Thus, if this essentialdisposition leads one to respond to a certain set of antecedentmental states about a certain object by having an attitude ofadmiration for that object, then this is a reliable indicator ofthe fact that at least under normal conditions, if all of thoseantecedent mental states are correct, then it really is correct toadmire the object in question.

In forming and revising one’s normative beliefs, it is possible tobe sensitive to these reliable indicators of normative truths. Ineffect, one can take the manifestations of one’s mental dispo-sitions as prima facie evidence of the corresponding normative


truths. One can take facts about what one is disposed to believeas evidence for what it is rational for one to believe, facts aboutwhat one is disposed to admire as evidence for what it is correctto admire, and so on. To form one’s normative beliefs in thisway is not to base one’s normative beliefs on any consciousjudgments about one’s mental dispositions. It is, rather, forone’s normative beliefs to be directly sensitive to the contoursof those mental dispositions themselves.

There are at least two ways in which one’s normative beliefsmay be directly sensitive to the contours of these mentaldispositions. First, one may actually be in the conditions thattrigger a manifestation of the disposition, and one may form thebelief directly in response to that manifestation of the disposition.For example, suppose that I feel an emotion of admiration onreading about someone’s action in the newspaper. Since thisemotion is whole-hearted, and not clouded by any countervailingemotions, and since I assume that the antecedent mental stateson which my admiration is based are correct, and that conditionsare otherwise normal (that is, there is nothing otherwise unusualabout this case), this emotion may directly incline me toform the belief that it is correct to admire the action inquestion.

This way of forming normative beliefs, however, is limitedto those cases in which one is actually in the conditions thattrigger the manifestation of the disposition in question. It maybe possible to be sensitive to the contours of our mentaldispositions even if we are not in conditions that actually triggerthe manifestation of those dispositions. As several philosophershave recently claimed, we seem to have the capacity for asort of ‘off-line simulation’ of the operation of our mentaldispositions. We can imaginatively simulate being in certainmental states—having certain beliefs or experiences, say—andthen we can respond in a way that simulates the way in whichwe are disposed to respond to actually being in those mentalstates.14

14. This idea of the ‘off-line simulation of our mental dispositions’ has been invokedby several philosophers to explain how we ascribe mental states to others. See inparticular Goldman (1989) and Heal (2003). (My own views would be considerablycloser to Heal’s than to Goldman’s, as is no doubt suggested by my sympathy forthe normative theory of intentional states.)

78 ralph wedgwood

For example, suppose that you imagine a certain act (say,you imagine a doctor who kills one healthy patient to use hisorgans to save five patients who would otherwise die). Imaginingthis act is a sort of ‘make-believe’: it involves a simulation ofthe state of believing that someone has performed an act ofthis kind. Then you might respond to this imagining with asimulation of the moral emotion that you are disposed to havein response to actually believing that a doctor has acted in thisway. In the case of imagining this act in particular, you will mostlikely respond with a simulation of the moral emotion of whole-hearted disapproval, unclouded by any countervailing emotionsor inclinations. Then it is possible that this sort of simulation ofthe emotion of such whole-hearted disapproval will also inclineyou to form the belief that at least under normal conditions, itis correct to disapprove of any action of this kind.

If that claim is true, then it may be plausible that this is whatour so-called ‘moral intuitions’ actually consist in—either actualmoral emotions, or the offline simulation of our dispositions formoral emotions. In this way, we can give a clear account of whereour ‘moral intuitions’ come from; we do not have to postulateany inscrutable mechanism for becoming aware of moraltruths.

This kind of simulation of the operations of one’s emotionaldispositions need not involve actually thinking about one’s ownemotions, or indeed representing emotions in any way. Indeed,strictly speaking, it need not be an imagination of an emotion—just as visualizing an uninhabited island involves simulating avisual experience of the island, but it does not involve imaginingthe occurrence of a visual experience of the island (after all, oneis precisely imagining the island as uninhabited).15 When oneimagines an act, and responds with a simulation of an emotionof moral disapproval towards that act, one’s attention is whollyfocused on the imagined act (not on one’s own reactions towardsthe act).

15. Compare Williams’ (1973) point that imagining oneself assassinating the PrimeMinister might typically involve, not only entertaining the thought ‘I am nowassassinating the Prime Minister’, but also producing within oneself mental images(based mostly on visual memory images) of 10 Downing St, of Tony Blair’s strickenbody, and so on. One is simulating visual experiences of assassinating the PrimeMinister, not thinking about such experiences.


A similar account could be given of our intuitions of whatit is rational to choose. Suppose that I am wondering aboutwhat it is rational to choose in a certain situation—say, in thechoice situation of Newcomb’s problem.16 I imagine the situationin question by simulating the beliefs and desires that define thatsituation; and I may then respond with a simulation of the choicethat I am disposed to make in response to actually having thebeliefs and desires in question. This response, which in my casereflects my disposition to be a ‘two-boxer’ in situations of thiskind, also disposes me to judge that it is rational to make thischoice in this situation. So this may be what my intuition that itis rational to be a ‘two-boxer’ consists in—the offline simulationof my dispositions for making choices. A similar account couldbe given of our intuitions of what it is rational to believe.17

In general, according to this account precisely the samedispositions underlie both one’s intuitions of rationality andone’s intuitions of correctness. When one manifests or simulatesthe operation of a mental disposition that responds to a certainset of antecedent mental states by forming a certain new mentalstate, this will incline one to form two different beliefs (if therelevant question arises). The first of these two beliefs is the beliefthat it is rational to form to that new mental state in responseto being in those antecedent mental states, while the second isthe belief that at least under normal conditions, if all of thoseantecedent mental states are correct, then the new mental statein question is also correct. When the manifestation or simulationof the disposition inclines one to form the former belief, it is anintuition of rationality; when it inclines one to form the latterbelief, it is an intuition of correctness.

This proposal, that we can base our normative beliefs onintuitions, where intuitions are interpreted as consisting in either

16. For a thorough discussion of Newcomb’s problem, see Joyce (1999).17. It is as the source of our intuitions about what it is rational to believe that thedispositions that are essential to possessing various particular concepts play theirmain role as a source of our normative beliefs. (Our beliefs about what it is correctto believe arise in the following simple way: whole-heartedly believing a proposition pinclines one to believe that it is correct to believe p—just as whole-heartedly admiringan object x inclines one to believe that that it is correct to admire x; and so if onesimulates believing a proposition p, one will respond with a simulation of believingthat it is correct to believe p, thus leading one to accept the inference from p to ‘Itis correct to believe p’.)

80 ralph wedgwood

the manifestations or the offline simulation of our mentaldispositions, has some advantages over another proposal thatsome philosophers have made—namely, the proposal that wecan base our normative and evaluative beliefs directly on ouremotions or affective states.18 First, my proposal allows fora uniform treatment of beliefs about what it is rational tobelieve (which are surely not based on any emotions or affectivestates), and of beliefs about what it is correct or appropriate toadmire or disapprove of. Second, it can accommodate the factthat much of our thinking about normative matters focuses onimaginary cases, which (since we do not believe these cases to beactual) rarely inspire any actual emotions or affective states onour part.

V

The Quest for Reflective Equilibrium. Even though my proposaldiffers from the proposal that we base our evaluative beliefsdirectly on our emotions, it faces some of the same objections. AsI have already pointed out, even if it is true that these ‘essential’mental dispositions are reliable indicators of correspondingnormative truths, these dispositions may in many cases beblocked or inhibited by interfering factors. Moreover, we mayalso have many other dispositions to use a given concept, orto have attitudes of a given type, that are not essential tothe capacity for that concept or attitude-type; and many ofthese other dispositions may be entirely unreliable. Forexample, people’s moral emotions are notoriously unreliable.Slave owners might have the most whole-hearted attitudes ofmoral disapproval towards slaves who defy their masters.19

In general, some of our mental dispositions are mostdefinitely not reliable indicators of corresponding normativetruths.

So if our ‘intuitions’ consist in simulations of the operations ofour mental dispositions, or in the manifestation of those disposi-tions, some of these intuitions will be unreliable. Moreover,

18. See especially Tappolet (2000) and Johnston (2001a).19. Compare for example Mill (1869, Ch. 1, para. 6).


we seem to have no way of responding directly to thedifference between our essential and our non-essential mentaldispositions—nor, more generally, to the difference betweenreliable and unreliable dispositions. So, we have no way ofresponding directly to the difference between reliable andunreliable intuitions.

However, it may still be possible to correct for the errorsthat may infect our normative intuitions—in effect, by revisingone’s beliefs so that one approaches what John Rawls called‘reflective equilibrium’.20 First, we can try to canvas as manyintuitions as we can, and then try to fit all these intuitionstogether into a coherent systematic set of normative beliefs.Our intuitions themselves suggest that some of these normativepropositions demand explanation; so a coherent overall set ofnormative beliefs will have to give at least some outline of anexplanation of some of these propositions in terms of other suchnormative propositions. In this way, we can be led to reject someof our intuitions if doing so leads to a more coherent overall setof normative beliefs. When we reject some of our initial pre-theoretical intuitions in this way, we will also be led to viewsabout which sorts of empirical psychological explanations of anintuition are most likely to cast doubt on its reliability. Then, wecan carry out psychological investigations of our own intuitions,to see whether any of them are best explained in ways that shouldlead us to view them with suspicion.

This method of revising one’s normative beliefs by seekingRawlsian ‘reflective equilibrium’ will not succeed in correctingfor the errors in one’s intuitions if one’s initial pre-theoreticalintuitions are so unreliable that further reflection based onthese intuitions will only lead one deeper into error. It willsucceed only if one’s initial intuitions are at least sufficientlyreliable that a process of reflection based on these intuitionswill eventually lead one closer to the truth. But I do not seehow one can have any a priori guarantee that this is how thingsare in one’s own case—any more than one can have an a prioriguarantee that one’s sensory experiences are veridical perceptionsrather than hallucinations that an evil demon is feeding intoone’s mind. But there seems to be no alternative method

20. See Rawls (1971, Sec. 9).

82 ralph wedgwood

available that does not ultimately depend on one’s pre-theoreticalnormative intuitions in this way (just as there is no non-empiricalmethod for detecting optical illusions and hallucinations andthe like).

Nonetheless, it seems to me that it is perfectly rational for usto form normative beliefs by means of this method. We need tohave some normative beliefs if we are to act as reflective agentsat all. It would be quite unreasonable to demand that we shouldsuspend judgment about all normative propositions unless wehave some infallible method of reaching true normative beliefs.It would also be unreasonable to demand that we shouldonly form normative beliefs by methods that we have someindependent or antecedent reason for regarding as reliable, sinceunless there is some basic or primitively rational way of formingnormative beliefs, we could never have any such independent orantecedent reason for regarding any way of forming normativebeliefs as reliable in the first place.

The only reasonable demand to make here is that we shouldnot form normative beliefs in a certain way in absence of anyindependent or antecedent reason to regard that way of formingnormative beliefs as reliable, unless that way of forming beliefshas some essential connection to the truth. If my account iscorrect, there is a general essential connection between ournormative intuitions and the truth: these normative intuitionsarise from our underlying mental dispositions, and some ofthese dispositions are guaranteed, by the essential nature of theconcepts and mental states involved, to be reliable indicatorsof corresponding normative truths. Since this connection to thetruth is guaranteed by the essential nature of the concepts andmental states that are involved in this way of forming beliefs,it may well be that this way of forming normative beliefs isrational.

Moreover, in certain favourable cases, the beliefs that areformed in this way may count as knowledge. Suppose thatone arrives at a normative belief, and the normative intuitionson which one relies in arriving at this belief consist purely insimulations of the operations of these ‘essential’ dispositions. Inthat case—according to my account—all these intuitions will betrue. Suppose that the belief that one arrives at in this way is bothrational and reliable, in the sense that it could not easily happen


that one should follow that method in those circumstances butfail to form a belief that is true. Then this normative belief willplausibly count as knowledge.21

Some philosophers will object to my proposal that there is aparallel between the explanation of why it is rational to formnormative beliefs by relying on one’s normative intuitions andthe explanation of why it is rational to form beliefs by takingone’s experiences at face value. According to these philosophers,this proposal is vitiated by certain disanalogies between sensoryexperiences and normative intuitions. For example, whilecognitive scientists often explain our sensory experiences on thebasis of the very facts that make those experiences veridical,there is no serious program of attempting to explain ournormative intuitions on the basis of the corresponding normativefacts.

In fact, however, even if there is a disanalogy here, it isnot clear that it undermines the proposal that I am making.According to my proposal, what we need to explain therationality of relying on normative intuitions is some accountof how these intuitions have some sort of essential connectionto the truth—not necessarily a scientific practice of explainingthese intuitions on the basis of the corresponding normativefacts. Moreover, it is not even clear that there is a disanalogyhere. Perhaps it is a crucial (though no doubt often merelyimplicit) assumption of many psychological explanations thatwe often form a certain belief or attitude precisely because ourantecedent mental states make it rational for us to do so. Ofcourse, such explanations would presuppose the existence ofnormative facts (such as the fact that certain antecedent mentalstates make it rational for one to form a certain new belief),and the existence of such normative facts is philosophicallycontroversial; and according to my proposal, the only primitivelyrational way of coming know such facts relies on thevery intuitions that are in question. But there is no disanalogyhere with sensory experiences. The cognitive scientists’explanations of our experiences presuppose the existence ofa material world, which is philosophically controversial (it is

21. I have said more about this conception of knowledge in an earlier paper(Wedgwood 2002).

84 ralph wedgwood

denied by Berkeleyan idealists), and the primitively rational wayof coming to know facts about the material world is by relyingon the very experiences that are in question. This surely doesnot cast doubt on the rationality of forming beliefs by takingexperiences at face value. So the analogous fact also does notcast doubt on the rationality of forming beliefs by relying onour normative intuitions.

This is not to say that there are no important disanalogiesbetween sensory experiences and normative intuitions. Oneimportant disanalogy comes out particularly clearly in thecase of normative intuitions that concern purely imaginaryexamples, and consist in the off-line simulation of ourunderlying mental dispositions. In these cases, the belief thatthe intuition inclines one to form is a general belief—such asthe belief that in any normal case of such-and-such a kind, itis correct to admire an action of such-and-such a kind.More fundamentally, the normative truths that are reliablyindicated by the ‘essential’ dispositions (which one must havein order to possess the relevant concepts or to be capableof the relevant types of attitude) are also general normativetruths about types of conditions under which various typesof mental state are correct or rational. In this way, ournormative intuitions are crucially different from our sensoryexperiences, whose contents are brutely particular—that is,they concern how things are in a particular situation at aparticular time. In my view, any plausible account of theepistemology of normative beliefs must capture this crucialdifference between normative intuitions and sensory experiences.Nonetheless, this does not rule out the limited sort of analogybetween the two ways of forming beliefs that I have invokedabove.

Many further problems will have to be solved before thisapproach can give a fully satisfactory epistemology of normativebeliefs. For example, we need to explain whether this methodof forming normative beliefs is empirical or a priori; andwe need to explain how this method could produce rationalbeliefs in spite of the widespread disagreement that surroundsnormative questions. I cannot hope to address these furtherquestions here. But if I am right that this approach hassome degree of plausibility, then it would be worthwhile


trying to see if this approach can address those furtherproblems.22

Merton CollegeOxford OX1 [email protected]

References

Alston, W. P. (1991): Perceiving God: The Epistemology of Religious Experience(Ithaca, NY: Cornell University Press).

Alston, W. P. (1993): The Reliability of Sense Perception (Ithaca, NY: CornellUniversity Press).

Brandom, R. (1994): Making It Explicit (Cambridge, MA: Harvard University Press).Crisp, R. (2002): ‘Sidgwick and the Boundaries of Intuitionism’, in P. Stratton-Lake

(2002a).Davidson, D. (2001): Inquiries into Truth and Interpretation, revised edition (Oxford:

Clarendon Press).Frege, G. (1892). ‘Uber Sinn und Bedeutung’, Zeitschrift fur Philosophie und

philosophische Kritik, NF 100: 25–50.Gibbard, A. (1996): ‘Thought, Norms, and Discursive Practice’, Philosophy and

Phenomenological Research 56: 699–717.Goldman, A. (1989): ‘Interpretation Psychologized’, Mind and Language 4: 161–185.Heal, J. (2003). Mind, Reason and Imagination (Cambridge: Cambridge University

Press).Hume, D. (1739): A Treatise of Human Nature, Books I–II (London: John Noon).Joyce, J. M. (1999): Foundations of Causal Decision Theory (Cambridge: Cambridge

University Press).Lance, M., and Hawthorne, J. (1997): The Grammar of Meaning (Cambridge:

Cambridge University Press).Lewis, D. (1974): ‘Radical Interpretation’, Synthese 23: 331–344.Locke, J. (1690): Essay concerning the Human Understanding (London).Johnston, M. (2001a): ‘The Authority of Affect’, Philosophy and Phenomenological

Research 63: 181–214.Johnston, M. (2001b): ‘Is Affect Always Mere Effect?’, Philosophy and Phenomeno-

logical Research 63: 225–228.Mackie, J. L. (1977): Ethics: Inventing Right and Wrong (Harmondsworth: Penguin).McDowell, J. H. (1997): ‘Brandom on Representation and Inference’, Philosophy and

Phenomenological Research 58: 157–162.Mill, J. S. (1869): On Liberty (London: Longmans, Green, Reader, and Dyer).Morris, M. (1992): The Good and the True (Oxford: Clarendon Press).Peacocke, C. (2004): The Realm of Reason (Oxford: Clarendon Press).

22. Earlier versions of this paper were presented at the University of York, theUniversity of Oslo, a workshop on Moral Epistemology at the University ofEdinburgh, and a conference on Metaphysics and Epistemology at the CroatianInter-University Centre in Dubrovnik. I am grateful to those audiences, and also toJohn Broome, Roger Crisp, Michael Dummett, Dorothy Edgington, Daniel Isaacsonand David Wiggins, for their helpful comments on those earlier drafts. I wrote thefinal draft of this paper while being supported by a Research Leave Award from theUK Arts and Humanities Research Board (AHRB), to whom I should also like torecord my gratitude.

86 ralph wedgwood

Pryor, J. (2000): ‘The Skeptic and the Dogmatist’, Nous 34: 517–549.Rawls, J. (1971): A Theory of Justice (Cambridge, MA: Harvard University Press).Rosen, G. (1997): ‘Who Makes the Rules Around Here?’ Philosophy and

Phenomenological Research 57: 163–171.Rosen, G. (2001): ‘Brandom on Modality, Normativity and Intentionality’,

Philosophy and Phenomenological Research 63: 611–623.Shafer-Landau, R. (2003). Moral Realism: A Defence (Oxford: Oxford University

Press).Stein, E. (1996): Without Good Reason: The Rationality Debate in Philosophy and

Cognitive Science (Oxford: Clarendon Press).Stratton-Lake, P., ed. (2002a): Ethical Intuitionism: Re-evaluations (Oxford: Claren-

don Press).Stratton-Lake, P. (2002b): ‘Introduction’, in P. Stratton-Lake (2002a).Tappolet, C. (2000): Emotions et valeurs (Paris: Presses Universitaires de France).Wedgwood, R. (1999): ‘The A Priori Rules of Rationality’, Philosophy and

Phenomenological Research 59: 113–131.Wedgwood, R. (2001): ‘Conceptual Role Semantics for Moral Terms’, Philosophical

Review 110: 1–30.Wedgwood, R. (2002): ‘The Aim of Belief ’, Philosophical Perspectives 16: 267–297.Wedgwood, R. (forthcoming): ‘The Normativity of the Intentional’, in Brian

McLaughlin and Ansgar Beckermann, eds., The Oxford Handbook of thePhilosophy of Mind (Oxford: Clarendon Press).

Williams, B. (1973): ‘Imagination and the Self’, in his Problems of the Self(Cambridge: Cambridge University Press).

iii-how we know what ought to be

Documents