Erkenntnis Volume 79 Issue 1 2014 [Doi 10.1007%2Fs10670-013-9485-9] Cowie, Christopher -- Epistemic...

19
Epistemic Disagreement and Practical Disagreement Christopher Cowie Received: 19 March 2012 / Accepted: 7 April 2013 Ó Springer Science+Business Media Dordrecht 2013 Abstract It is often thought that the correct metaphysics and epistemology of reasons will be broadly unified across different kinds of reason: reasons for belief, and reasons for action. This approach is sometimes thought to be undermined by the contrasting natures of belief and of action: whereas belief appears to have the ‘constitutive aim’ of truth (or knowledge), action does not appear to have any such constitutive aim. I develop this disanalogy into a novel challenge to metanormative approaches by thinking about disagreement. The constitutive aim of belief can play a role in adjudicating epistemic disagreements for which there is no analogue in practical disagreements. Consequently, we have more reason, all else being equal, to expect convergence in epistemic judgment than in practical judgment. This repre- sents a prima facie challenge to the metanormative theorist because the extent of (suitably specified) disagreement in an area of thought is of prima facie significance for the metaphysics of that area of thought. It is often thought that the correct metaphysics and epistemology of reasons will be broadly unified across different kinds of reason: reasons for belief, and reasons for action. A number of authors have recently attempted to develop such a unified or ‘metanormative’ approach. 1 There is much to recommend this approach: reasons for belief and reasons for action are, after all, both reasons. But there are also disanalogies between reasons for belief and reasons for action that might be thought to undermine it. One disanalogy sometimes mentioned in the literature concerns the contrasting natures of belief and of action. It is claimed that whereas belief appears C. Cowie (&) Fitzwilliam College, University of Cambridge, Storeys’ Way, Cambridge CB3 0DG, UK e-mail: [email protected] 1 ‘Metanormativity’ is, admittedly, a broad church. Those who explicitly self-identify as ‘metanorm- ativists’ include Chrisman (2010), and especially Enoch (2011a), Ch. 4. Many more views might reasonably be thought to fall under this classification, including (e.g.) Scanlon (1998), Gibbard (2003), Wedgwood (2002), Parfit (2011), Skorupski (2011). 123 Erkenn DOI 10.1007/s10670-013-9485-9

description

moral

Transcript of Erkenntnis Volume 79 Issue 1 2014 [Doi 10.1007%2Fs10670-013-9485-9] Cowie, Christopher -- Epistemic...

  • Epistemic Disagreement and Practical Disagreement

    Christopher Cowie

    Received: 19 March 2012 / Accepted: 7 April 2013

    Springer Science+Business Media Dordrecht 2013

    Abstract It is often thought that the correct metaphysics and epistemology ofreasons will be broadly unified across different kinds of reason: reasons for belief,

    and reasons for action. This approach is sometimes thought to be undermined by the

    contrasting natures of belief and of action: whereas belief appears to have the

    constitutive aim of truth (or knowledge), action does not appear to have any such

    constitutive aim. I develop this disanalogy into a novel challenge to metanormative

    approaches by thinking about disagreement. The constitutive aim of belief can play

    a role in adjudicating epistemic disagreements for which there is no analogue in

    practical disagreements. Consequently, we have more reason, all else being equal, to

    expect convergence in epistemic judgment than in practical judgment. This repre-

    sents a prima facie challenge to the metanormative theorist because the extent of

    (suitably specified) disagreement in an area of thought is of prima facie significance

    for the metaphysics of that area of thought.

    It is often thought that the correct metaphysics and epistemology of reasons will be

    broadly unified across different kinds of reason: reasons for belief, and reasons for

    action. A number of authors have recently attempted to develop such a unified or

    metanormative approach.1 There is much to recommend this approach: reasons for

    belief and reasons for action are, after all, both reasons. But there are also

    disanalogies between reasons for belief and reasons for action that might be thought

    to undermine it. One disanalogy sometimes mentioned in the literature concerns the

    contrasting natures of belief and of action. It is claimed that whereas belief appears

    C. Cowie (&)Fitzwilliam College, University of Cambridge, Storeys Way, Cambridge CB3 0DG, UK

    e-mail: [email protected]

    1 Metanormativity is, admittedly, a broad church. Those who explicitly self-identify as metanorm-

    ativists include Chrisman (2010), and especially Enoch (2011a), Ch. 4. Many more views might

    reasonably be thought to fall under this classification, including (e.g.) Scanlon (1998), Gibbard (2003),

    Wedgwood (2002), Parfit (2011), Skorupski (2011).

    123

    Erkenn

    DOI 10.1007/s10670-013-9485-9

  • to have the constitutive aim of truth (or knowledge), action does not appear to

    have any such constitutive aim.2 In what follows, I develop this disanalogy into a

    novel challenge to metanormative approaches.

    I develop the argument by thinking about disagreement. The constitutive aim of

    belief can play a role in adjudicating epistemic disagreements for which there is no

    analogue in practical disagreements. Consequently, we have more reason, all else being

    equal, to expect convergence in epistemic judgment than in practical judgment. This

    represents a prima facie challenge to the metanormative theorist. It represents a prima

    facie challenge because the extent of (suitably specified) disagreement in an area of

    thought is of prima facie significance for the metaphysics of that area of thought.

    The argument proceeds as follows. Firstly, I set out a requirement for

    convergence in normative judgmentthe shared criteria requirement. Secondly, I

    identify a sense in which this requirement is met for epistemic judgment but not for

    practical judgment. Thirdly, I explain this in terms of the disanalogy in the

    constitutive aims of belief and action. Fourthly, I respond to objections.

    1 Disagreement and Shared Criteria

    What is the source of normative disagreement, or disagreement about what reasons

    one has? In this section it is claimed that the source of normative disagreement is, in

    part, a lack of shared normative criteria for adjudicating normative disagreements.

    This appears to apply to both practical, and epistemic, normative disagreements.

    Begin by thinking about practical disagreement: disagreement about what one has

    reason to do. Two sources of practical disagreement are often highlighted in the

    literature. The first is ignorance of non-normative matters. An example illustrates. Two

    disputants may agree that the most reasonable course of action is that which maximises

    welfare. But they may nonetheless disagree on which course of action is most reasonable

    if one or more is ignorant as to which course of action in fact maximises welfare. A

    second obvious source of disagreement is a failing in the procedural rationality of one or

    more of the disputants.3 For example, two disputants may agree both that the most

    reasonable course of action is that which maximises welfare, and that a particular course

    of action, A, maximises welfare. But unless they are sufficiently rational in combining

    their beliefs (so as to conclude that the appropriate course of action is A), then we should

    not expect them to reach agreement on what the appropriate course of action is.

    That disputants are well-informed and (procedurally) rational may be necessary

    for convergence in practical judgment. But it is not obviously sufficient. It is also

    necessary that disputants share normative criteria for adjudicating disputes. In the

    above example, the disputants share the normative criterion that the required course

    2 This is expressed in the context of scepticism about companions in guilt arguments between morality

    and epistemology (such as that undertaken in Cuneo 2007) in Lillehammer Lillehammer and Hallvard

    (2007), and p. 170. FitzPatrick (2009), p. 757. See also Darwall (2003), pp. 483488, Tersman (2006),

    p. 96.3 I have claimed that disagreement often has its source in failings of procedural rationality. But it might

    be claimed that disagreements also often have their source in failings of substantive rationality (that is, infailure to respond to the reasons that one has). See Fn. 7.

    C. Cowie

    123

  • of action is that which maximises welfare. If they did not share this criterion then

    factual agreement and (procedural) rationality would be insufficient to yield

    convergence in their judgment about which course of action was required. This

    gives rise to an intuitive requirement on convergence in normative judgment. For

    disputants to converge (non-accidentally) in their normative judgments, they must

    share some normative criteria for determining when those judgments are correct.4

    Ill call this the shared criteria requirement: it is necessary for two (or more)disputants to (non-accidentally) converge in their judgments that they share

    common normative criteria for determining which judgment is correct. The above

    example illustrates this in an over-simplified sense. In reality, disputes will not

    always rest on simple normative principles (e.g. it is reasonable to do whatever

    maximises welfare). Rather, disputes will rest on a number of different background

    principles and particular judgments of greater and lesser degrees of determinacy and

    importance. The fewer the criteria that disputants share, or the less determinate or

    important these criteria are, the less convergence we should expect.

    One of the sources of philosophical interest in practical disagreement is that it

    appears that frequently, the shared criteria requirement is met to an insufficient degree

    to ensure convergence, even amongst apparently rational and well-informed

    disputants. Disputants lack sufficient basic shared normative criteria to resolve

    disputes about, for example, abortion, medical research, and gun control, not to

    mention in genuinely inter-cultural practical disputes.5 This gives rise to arguments for

    scepticism about practical, especially moral, judgment, and to abductive arguments

    for metaphysical anti-realism about practical, and again especially moral, judgment.6

    And whilst the soundness of these arguments is sometimes contested, they nonetheless

    represent an important prima facie challenge to the scope of our practical knowledge,

    and to straightforwardly realistic views of the metaphysics of practical normativity.7

    4 Some philosophers are optimistic that informed and procedurally rational disputants would converge in

    a substantial proportion of their practical judgments (e.g. Brink 1989, pp. 197210; Parfit 2011, p. 543,

    and see also Jackson 1998, Ch. 5). But there is good reason to be sceptical of this e.g. Mackie (1977),

    p. 37, Enoch (2011a), p. 191. I discuss this further in Objections and Replies below.5 Elga (2007), (Sec. 1213) claims that disputants about the cluster of issues related to the permissibility

    of abortion typically lack sufficient shared normative criteria for convergence. See Kornblith (2010) for

    the view (in response to Elga) that whereas disputants may share criteria in these cases, criteria are not

    shared in cases of inter-cultural practical disagreements.6 Arguments from disagreement to metaphysical conclusions take a number of forms, e.g.: (1) abductive

    arguments against moral realism, and for moral constructivism (e.g. Harman 1996, 2000); (2) arguments from

    judgment dependent theories of moral truth to moral error theories (see e.g. Tersman 2004, Ch. 4); (3)

    arguments from conceptual requirements on convergence to moral error theories (see e.g. Lillehammer 2004).7 For the most thorough discussion of, and rejection of, arguments from disagreement to metaphysical

    anti-realist conclusions, see Enoch (2009, 2011a) Ch. 8. One of the most obvious reasons for rejecting any

    arguments of this form concerns the use of rationality (Ibid., pp. 210211). I have claimed that

    disagreements amongst procedurally rational and well-informed agents pose a prima facie challenge toclaims to moral knowledge, and to straightforwardly realistic accounts of practical normativity. But it

    might be objected that it is only disagreements amongst substantially rational agents that have thisconsequence. And it might be thought that substantially rational agents would share normative criteria

    (or at least that we have no compelling reason to think otherwise). My response is concessive. If

    well-informed and substantially rational agents were to disagree, this would be sufficient for rejectingclaims to moral knowledge, and to straightforwardly realistic accounts of practical normativity. By

    contrast, I have merely claimed that disagreement amongst well-informed and procedurally rational

    Epistemic Disagreement and Practical Disagreement

    123

  • My concern is with whether epistemic disagreements pose an analogous prima facie

    challenge to epistemic knowledge, and to straightforwardly realistic views of the

    metaphysics of epistemic normativity. Do apparently rational and informed parties lack

    shared criteria for resolving disputes about what they have reason to believe in anything

    like the sense that their practical counterparts appear to? On a first pass, perhaps they

    seem to. Suppose that we understand epistemic disagreements as disagreements about

    whether a source of evidence provides a reason to believe a conclusion. Such

    disagreements can often be traced to a lack of shared criteria for adjudicating whether a

    source of evidence is reason-providing. For example, my friend Stuart thinks that he has

    reason to believe whatever The Bible says. I dont. These are fairly fundamental

    commitments for Stuart and I, and we appear to lack sufficient shared criteria to make

    progress in resolving our disagreement. I use the criterion of naturalistic plausibility to

    assess whether theres reason to believe what the Bible says. And so, I dont take there to

    be strong reason to believe much of what it says. But Stuart rejects the criterion of

    naturalistic plausibility precisely because it would lead him to deny that there is much

    reason to believe what the Bible says. Our criteria are in conflict. A fortiori, the shared

    criteria requirement is not met to a sufficient degree to allow for convergence.

    2 Formal and Substantial Criteria

    In this section I raise the possibility that, in fact, practical and epistemic

    disagreement are not as similar as the above argument suggests. There is a sense in

    which the shared criteria requirement may be met for epistemic disputes, but not for

    practical disputes. And as a consequence, if all else is equal, we should expect a

    greater degree of convergence in epistemic judgment than in practical judgment.

    Consider the following rather abstract sense in which criteria requirement might

    be thought to be shared for epistemic disputes: judgments about reasons for belief

    are answerable to the truth of the beliefs that would result from following them. If

    Stuart thought that believing whatever the Bible says was likely to lead him to form

    a lot of false beliefs, then he wouldnt take himself to have much reason to believe

    whatever the Bible says. And if I thought that believing whatever the Bible says was

    likely to lead me to form a lot of true beliefs, then I would alter my epistemic

    judgment accordingly too. Failure of either of us to revise our epistemic judgment in

    this way would likely belie some irrationality or conceptual incompetency on our

    part. We are, then, operating with the following, admittedly rough, shared criterion

    for assessing epistemic judgments: a source of evidence provides reason for

    believing a conclusion to the extent that it increases the likelihood that the

    conclusion believed is true rather than false. For short, Ill say that we would share

    the criterion of the reliability for truth of a source of evidence in assessing thereason for belief provided by that source of evidence.

    The reliability for truth criterion clearly admits of extensions and precisifications,

    but the present formulation is (as I hope will become clear) sufficient for present

    Footnote 7 continued

    agents poses a prima facie challenge to claims to moral knowledge, and to straightforwardly realisticaccounts of practical normativity. This may satisfy Enoch (see Ibid., p. 213, Fn. 72).

    C. Cowie

    123

  • purposes. The important point is that plausibly, the truth-values of ones subsequent

    beliefs can function as a shared criterion of assessment for epistemic judgments (at

    least amongst the rational and conceptually competent).8 I defend this claim further

    in the following section. For present, however, I focus on introducing a challenge to

    the view that there is an analogous criterion for assessing practical judgment.

    I begin by taking Wedgwoods discussion of the action-guiding mental state

    choice as a foil for thinking about practical reasons and the criteria for assessing

    them. Is there some criterion, analogous to the reliability for truth criterion, which

    could be used to assess judgments about choice? Wedgwood works with

    choiceworthiness as an analogue to truth. It is plausible that (rational and

    conceptually competent) disputants would agree that whether one has reason to

    choose some course of action depends on the likelihood that the course of action is

    choiceworthy. Much as with the discussion of belief and truth above, one who

    denies this belies some form of irrationality or conceptual incompetence. So, it may

    appear that just as reliability for truth is an (admittedly abstract) shared criterion for

    adjudicating epistemic disputes, so reliability for choiceworthiness is an analogousshared criterion for adjudicating practical disputes.

    In fact though, reliability for choiceworthiness isnt analogous to reliability for

    truth. Reliability for choiceworthiness is much weaker, or less informative, than

    reliability for truth. We can see this by employing a distinction made by Velleman.

    Velleman draws a distinction between two different ways in which the aim of a

    practice can be specified.9 The aim of a practice can be specified either formally or

    substantially. Velleman uses the example of chess to illustrate. The formal aim of

    chess is winning. The substantial aim is what winning in chess consists in: check-

    mating the opponents king. More generally, the substantial aim of a practice is

    whatever achieving its formal aim consists in. The terminology of aiming is

    particular to Vellemans concerns, but the formal/substantial distinction can

    usefully be applied to show how truth and choiceworthiness are dis-analogous.

    They are disanalogous in that choiceworthiness is a merely formal norm on choice,

    but truth is a substantial norm on belief. That choiceworthiness is a merely formal

    norm on choice is intuitive. Choiceworthy simply means whatever one ought to

    choose. But truth is not merely a formal characterisation of the norm that governs

    belief. The formal norm that governs belief is whatever one ought to believe. But

    true isnt simply a placeholder for whatever one ought to believe.10 So, truth

    states a substantialalbeit very abstractnorm on belief. It follows that whereas

    reliability for choiceworthiness is a merely formal criterion for assessing judgments

    about what one has reason to choose, reliability for truth is an informative or

    substantial criterion for assessing judgments about what one has reason to believe.11

    8 I shant claim that sharing this criterion is sufficient for adjudicating all epistemic disputes. My claim

    defended further below and in dealing with objections and repliesis merely that it is of some help.9 Velleman (1996), p. 700. My presentation follows (in part) the discussed of Velleman in Wedgwood

    (2002), pp. 1012.10 Not unless (a particular variety of) pragmatism about belief holds as a conceptual truth. I assume it

    does not.11 Compare Darwall (2003), p. 485.

    Epistemic Disagreement and Practical Disagreement

    123

  • This is relevant for thinking about disagreement. All else equal, disputants are

    more likely to converge in their judgment if they share a substantial criterion of

    assessment of the disputed judgment than if they merely share a formal criterion.

    This isnt to say that disputants who agree on a substantial criterion of assessment

    for a disputed judgment will always have sufficient resource to converge on the very

    same judgment. They may not. But they will still be better placed than those who do

    not. Compare chess: players are more likely to agree who has won if they agree that

    winning is achieved by check-mating the opponents king than if they dont (even if

    they dont agree on exactly how castling works).

    3 Correctness Conditions

    I have suggested that there might be a substantial criterion of assessment of epistemic

    judgment in terms of the reliability for truth criterion. And I have gestured toward,

    though not yet argued for, the view that there is only a formal criterion of assessment of

    practical judgment. In this section I present the arguments for these claims.

    The argument is based on thinking about the correctness conditions of mentalstates. The very general norms on mental states under discussiontruth and

    choiceworthinessare explicable in terms of the correctness conditions of those

    states. Correctness is a normative property of a mental state that applies to that state

    merely insofar as it is the kind of thing that it is. More precisely, the correctness

    conditions of a mental state specify the conditions under which a mental state goes

    right or gets things right merely insofar as it is the kind of thing that it is.12 The

    correctness norm is important in the present context because it allows us to identify

    general norms on mental states just by reflecting on the nature of those states. I shall

    claim that whereas reflection on the nature of belief allows one to see that a belief

    goes right or is correct when it is true, reflection on the nature of choice only

    allows one to see that a choice goes right or is correct when it is choiceworthy.

    This explains why there is a substantial general norm on belieftruthon which

    we can expect agreement amongst the conceptually competent, but only a formal

    general norm on choicechoiceworthinesson which we can expect agreement

    amongst the conceptually competent.

    Begin with belief. How is it possible to arrive at the view that truth is a norm on

    belief simply by reflecting on the nature of belief? The correctness conditions of a

    belief state the conditions under which a belief goes right just insofar as it is the kind

    of thing that it is. So in order to work out the conditions under which a belief is

    correct, it is necessary to ask when a belief goes right merely insofar as it is the kind

    of thing that it is. Belief is a representational state. It represents the world as being

    thus-and-so. But it is not merely a representational state. It is a state which purports

    to represent veridically. In this it is distinguished from other states which arerepresentational, but which dont purport to veridicality in their representation (e.g.

    imagination, supposition, etc.).13 Given that belief is the kind of state which

    12 Wedgwood (2002), p. 267. See also Thomson (2008), p. 83.13 A point stressed by Velleman (1992), p. 12.

    C. Cowie

    123

  • purports to represent veridically, it is possible to state the correctness conditions of

    belief. A belief is correct iff it represents veridically. Or, more simply:

    Belief-Correctness: (For any agent, S, any proposition, P,) Ss belief that P iscorrect iff P is true.

    This quick sketch, and particularly some of the details, may of course be

    contested. I discuss some worries below. But the basic idea is intuitive and

    appealing. Truth is normative for belief because of the nature of belief itself.14 And

    so, anyone who is fully competent with the concept of belief can come to know that

    truth is normative for belief. Conceptual competence with belief suffices for some

    very general knowledge of the norms on belief.

    Before turing to the discussion of the correctness-conditions of choice, it is worth

    developing the account a little further in the context of belief. The exact normative status

    of correctness is contested. But on a number of plausible interpretations, the reliability

    for truth criterion for assessing epistemic judgments falls out of Belief-Correctness

    almost directly. Take Thomsons account as an example (though a number of other

    similar models would serve my purpose).15 Thomson explains how the correctness

    conditions of a mental state are reason-providing for that state. According to Thomsons

    account, a fact provides a reason to hold a mental state with a particular content iff that

    fact is evidence that it would be correct to hold in that mental state. Or:

    Reasons for Mental States: For any agent S, any proposition, P: A fact, F, is areason, R, for S to hold some mental state A toward P iff F is evidence that P

    satisfies the correctness conditions of A.16

    This provides a plausible schema for the reasons for holding any kind of mental

    state. Reasons for specific kinds of mental states are then specified by substituting in

    the correctness conditions of that state. Substituting in the correctness conditions of

    belief yields an account of reasons for belief which should be recognisable as the

    reliability for truth criterion:

    14 It might be objected that reasons for belief cant be explained solely in terms of the nature of belief for

    reasons familiar from David Enochs Agency Schmagency and Schmagency Revisited (Enoch 2006;

    Enoch 2011b). As applied to belief, the worry would take roughly the following form: one doesnt have

    reason to believe what its correct for one to believe unless one has reason to be a believer (as opposed to

    a schmeliever) in the first place. I dont attempt to provide a full response to Enochs worry here. I simply

    note that Enochs worry applies to reasons for holding any mental state (e.g. belief and choice). And anysatisfactory response must also, I think, apply with equal generality. So, if a satisfactory response to

    Enochs worry can be found, then, provided that my (independent) argument (below) for a disanalogy in

    the correctness conditions of belief and choice is sound, there will be a subsequent disanalogy between

    reasons for belief and reasons for choice. I leave the provision of a full response to Enochs worry to

    another paper, or another author.15 Especially Wedgwood (2007), pp. 154158. See also Kearns and Starr (2009).16 Thomson (2008), pp. 130134. This formulation might be objected to if read from right-to-left.

    According to some philosophers one doesnt have reason to believe a conclusion for which one has

    (sufficient) evidence unless one has some interest in so believing (see e.g. Leite 2007; Steglich-Petersen

    2011). For counter-argument see Kelly (2003, 2007). In any case, it is highly likely that agents who

    disagree about a proposition do have an interest in it (if for no other reason than that they are disagreeing

    about it). For the more radical view that agents have reasons to believe conclusions that they dont have

    evidence for the truth of, see Sect. 5.1 below.

    Epistemic Disagreement and Practical Disagreement

    123

  • Reasons for Belief: (For any agent S, any proposition, P) A fact, F, is areason, R, for S to believe that P iff F is evidence that P is true.

    Can the approach sketched above be used to establish a substantial criterion for

    assessing practical judgments? My claim is that it cannot. Reflection on the nature

    of choice only allows us to establish a formal (not a substantial) norm on choosing:

    choiceworthiness. To see this, suppose that we ask what the correctness conditions

    of choice are. The correctness conditions of choice specify the conditions under

    which ones choice goes right merely insofar as it is an instance of the kind of

    thing that it is. But when does a choice go right as the kind of thing that it is?

    Following the discussion of the previous section, we can, at a minimum, accept that

    a choice goes right as the kind of thing that it is iff the course of action chosen is

    choiceworthy. This allows us to state the following norm of correctness on choice:

    Choice-Correctness: (For any agent, S, and any action, V,) It is correct for Sto choose to V iff V-ing is choice-worthy.

    And so, following Thomsons suggested relation between correctness and reasons

    we can deduce an approximation of reliability for choiceworthiness as follows:

    Reasons for Choice: (For any agent S, and any action, V,) A fact, F, is areason, R, for S to choose to V iff F is evidence that V-ing is choice-worthy.

    But is it possible to do any better than this? Is it possible, merely by reflecting

    rationally on the nature of choice, to arrive at a substantial (not merely formal)

    characterisation of choiceworthiness? Whilst it might be possible to arrive at some

    constraints on choiceworthiness by reflecting on the nature of choice (e.g. that

    choosing to simultaneously perform to incompatible courses of action is never

    choiceworthy), I am sceptical of whether we can arrive at a more informative

    characterisation of choiceworthiness (analogous to the role that truth plays in

    relation to belief) simply by reflecting rationally on the nature of choice.

    One intuitive reason for scepticism is similar in outline to the reasoning behind

    the open question argument. It appears that one can always question, without

    belying a lack of rationality or conceptual competence, whether a course of action

    possessed of some specific non-normative property is in fact choiceworthy in virtue

    of its possession of that property. And the burden of proof is on one who wishes to

    show otherwise. If this is the case, then the burden of proof is on one who wishes to

    show that there is some substantial characterisation of choiceworthiness on which

    we should expect rational and conceptually competent agents to agree.17 And it is

    surely reasonable to be sceptical of whether this burden could be discharged if we

    think about the process by which one would go about working out a substantial

    characterisation of choiceworthiness. It would require engaging in ordinary first

    order practical justification, or reflective equilibrium.18 This process consists in

    17 The significance of the open question argument for thinking about the disanalogy between reasons for

    belief and reasons for action is pursued in greater length in Heathwood (2009).18 Wedgwood (2003), p. 27. Compare Dworkin (2011), Ch. 8. According to Dworkin, there are no

    conceptual truths in, or about, morality. All first-order and meta-ethical moral judgments are themselves

    moral judgmentsjudgments whose justification rests on first-order moral theorising.

    C. Cowie

    123

  • reflecting on ones existing judgments about what is choiceworthy, both in the form

    of particular judgments and general principles, and coming to a conclusion on this

    basis. And there is no reason to expect disputants to settle on the same equilibrium

    point. There are two reasons for this. Firstly, different agents may start with

    commitments to different practical principles and intuitions. And secondly, even

    those who start with similar intuitions and principles may weigh them differently,

    thereby arriving at different equilibria. So, reflection on the process by which one

    would go about working out a substantial characterisation of choiceworthiness does

    not encourage optimism for convergence in that judgment.

    There is a second kind of reason, sometimes mentioned in the literature, for

    scepticism of convergence on a substantial characterisation of choiceworthiness. It

    is that, in contrast to belief, there are many kinds of (good) reason that one can

    adduce for choosing. Consider, for example, Scanlon:

    since believing is believing to be true, the only kind of reason that one canhave for believing something is a reason for thinking it true Reasons foraction, on the other hand, are plural.19

    Scanlon puts the point in terms of reasons for action rather than in terms of

    reasons for choice. But I take it that the same point applies to choice too (I discuss

    this further in the following section). Reasons for choice are plural: there are many

    different dimensions along which a consideration might count in favour of choosing

    some course of action. A consideration might count in favour of choosing some

    course of action on account of its pragmatic properties, its moral properties, its

    aesthetic properties, its comic properties, its legal properties, and so on.

    This plurality of reasons for choice might be thought to undermine the prospect

    of a substantial characterisation of choiceworthiness in two ways. Firstly, it is

    reasonable to be sceptical of whether there really are all-things-considered reasons

    for choice encompassing these many different dimensions. Consider, for example,

    the possibility that at least some of the kinds of reason provided by these different

    dimensions are incommensurable.20 It would be a consequence that there are at least

    some cases in which there is no uniquely correct answer as to which course of action

    is choiceworthy. But even if there are all-things considered reasons for choice we

    might be sceptical of whether disputants would agree on them. Different agents mayweight the reasons provided by different dimensions of assessment differently. This

    could lead to disagreement concerning which course of action is choiceworthy and it

    is not obvious that any such differences would belie rational failings or conceptual

    incompetencies on behalf of the disputants.

    4 Extending the Argument: Action

    I have argued that whereas conceptual competence may be sufficient to arrive at

    substantial criterion of assessment for judgments about reasons for belief, the

    19 Scanlon (1998), p. 36.20 A possibility discussed in e.g. Chang (1997).

    Epistemic Disagreement and Practical Disagreement

    123

  • analogous claim cannot be made for judgments about reasons for choice. In this

    section I extend the argument from choice to action and other action-guiding

    mental states.

    It might be thought that the argument presented above turns on the selection of

    choice as the practical analogue of belief. Specifically, it might be thought that

    there is a substantial criterion governing judgments about action on which weshould expect disputants to agree. But this would be mistaken. The argument

    sketched above generalises. To show this, suppose that we ask whether there is

    some norm which governs action and which would could be used as a criterion for

    assessing judgments about reasons for action. It might be thought that there is: the

    good. If so, then the good stands to action as truth stands to belief. I dont think,

    however, that this is right. To see this, note that there are two ways in which we

    might understand the good. One obvious way to understand it is simply as a place-

    holder for whatever one ought to do. So understood, we might reasonably expect

    agreement that the good normatively governs actions. Any dissenters would belie a

    conceptual incompetence. But if we think of the good in this way, then it is merely a

    formal, not a substantial norm on action. And so it is not analogous to the role that

    truth plays with respect to belief. Suppose, though, that we understand the good in

    more substantial terms, not merely as a place-holder. This involves thinking of the

    good as embodying some substantial first-order normative commitmentsfor

    example, some form of consequentialism. Insofar as we do this, the good is a

    substantial, not merely a formal, norm on action. And so it may ground a substantial

    criterion of assessment for judgments about reasons for action. However, we no

    longer have reason to expect agreement on that criterion. Much as with substantial

    characterisations of choiceworthiness, defending a substantial conception of the

    good would require engaging in first-order practical justification that goes beyond

    mere reflection on the concept of action. And, as claimed above, we should be

    sceptical of agreement on conclusions that must be reached by such a process.

    Indeed, it is worth noting that we might think that a similar argumentative

    strategy applies to, for example, moral emotions such as shame, guilt, regret, and so

    on. There are purely formal normative relations between shame and shamefulness,

    guilt and guiltworthiness, trust and trustworthiness. But as these are formal, not

    substantial relations, they are not analogous to the role that truth plays relative to

    belief. One might propose some substantial characterisation of shameworthiness,

    guiltworthiness, trustworthiness, and so on. But, it would be necessary to engage in

    first-order practical justification in order to arrive at such a substantial character-

    isation. And so we should be sceptical of agreement on that characterisation.

    This response may not satisfy. It might be objected that the response given above

    misses an important point. I have claimed that mere rational reflection on the nature

    of action-guiding mental states (e.g. choice) yields only a formal characterisation of

    the norms that govern these states. Many authors working on the correctness

    conditions of action-guiding mental states would agree.21 But there are a number of

    views in the literature according to which action itself does have a substantial

    constitutive aim that can be ascertained just by thinking about the nature of those

    21 e.g. Wedgwood (2003), Thomson (2008) (e.g. p. 132), Shah (2008), Shah and Evans (2012).

    C. Cowie

    123

  • states. Consider, for example, the views of David Velleman and Alan Millar.

    According to (at least one time slice of) Velleman the constitutive aim of action is

    autonomy, or the conscious control of ones action.22 And according to Millar the

    constitutive aim of action is that the action have some end, point, or purpose.23

    These are both substantial, or informative views of the constitutive aim of action.

    And so they might be thought to represent a challenge to my argument.

    In fact, I dont think these views do represent a challenge, nor would their authors

    intend them to. To see this, it is necessary to be clear on the difference between

    correctness as a property of mental states, and constitutive aims as Millar andVelleman are understanding them. Correctness for mental states is a normative

    property. It states when a mental state goes right as the kind of thing that it is. This

    should not be confused with the constitutive aim of action in the sense that

    Velleman and Millar are using it. The constitutive aim of action (in the sense that

    they are understanding it) is not necessarily a normative property. It doesnt state the

    conditions under which an action goes right as the kind of thing that it is. Rather, it

    states what is intrinsic to all intentional actions.24 In claiming that action

    constitutively aims at purposiveness or autonomy, these authors should be thought

    of as claiming that autonomy or purposiveness are properties that are intrinsic to

    action. They are properties such that to the extent that any putative action does not

    possess them, it is not a bona fide action.

    We can see the difference between correctness conditions and constitutive aims

    clearly by contrasting the consequences of ones action failing to meet its

    correctness condition with the consequences of ones action failing to meet its

    constitutive aim. If ones action fails to meet its correctness conditioni.e. if that

    action is not worthy of being performedthen one is worthy of criticism for

    performing that action. One is worthy of criticism because one has performed an

    action that one ought not to have performed. But the consequence of ones action

    failing to meet its constitutive aimfailing to be autonomous, or purposiveis

    rather different. It isnt necessarily that one is blameworthy at all. Rather, the

    consequence is that one hasnt performed a full-blooded intentional action at all.

    This highlights the sense in which whereas correctness is a normative property, a

    constitutive aim is not.25

    Given that the constitutive aim of a mental state neednt be normative for that

    mental state, the views of Velleman and Millar arent in competition with my claim

    that the correctness conditions of action (and action-guiding mental states) are

    uninformative. The views of Velleman and Millar would represent a challenge to

    the view that action doesnt have a substantial constitutive aim. But I have not

    claimed that action does not have a substantial constitutive aim (in the sense that

    they are using it). It may. I have only claimed that action (or choice) doesnt have a

    22 Velleman (1996), p. 719.23 Millar (2004), p. 68.24 Ibid.25 This point is nicely made by Millar: an action may achieve the constitutive aim whilst being subject

    to criticism in all sorts of ways that are not explicable just in terms of the constitutive aim (Millar 2004,

    p. 68).

    Epistemic Disagreement and Practical Disagreement

    123

  • substantial correctness condition. And their views do not (and are not, I think,

    intended to) represent a challenge to this.

    5 Objections and Replies

    5.1 Objection 1

    The first objection concerns the significance of my conclusion for the scope of

    disagreement that we should expect in the epistemic and the practical domains

    respectively. It might be objected that the existence of a substantial norm on a

    mental state is neither necessary nor sufficient for convergence in judgments about

    reasons for holding that mental state. Note firstly why the existence of a substantial

    criterion of assessment is not necessary. Focus once again on the practicaljudgment. According to some authors we should expect convergence on at least

    some fixed points in their practical judgment, from which more general agreement

    may follow.26 Disputants might agree, for instance, that some particular courses of

    action (e.g. torturing children) are not choiceworthy. In this case, they will share

    some criteria for resolving disputes. And yet their agreement is not a consequence of

    any prior agreement on a substantial norm governing choice. Rather, it is a

    consequence of agreement on particular cases. Note secondly that the existence of a

    substantial norm governing a mental state is not sufficient. To see this, focus on themental state of belief. It is surely possible for disputants to agree on the reliability

    for truth criterion for resolving an epistemic dispute without thereby having the

    means to resolve that dispute. This will be the case if they disagree on what the

    evidence available to them supports. The disagreement discussed above between

    Stuart and myself over the status of the Bible is a relevant example.

    My response to this objection isnt to contest these claims. I concede that the

    existence of a substantial norm governing a mental state is neither necessary nor

    sufficient for convergence in judgments about reasons for holding that mental state.

    However, properly understood, this concession should not be thought of as

    undermining my argument. This concession would undermine my argument if the

    intended conclusion of my argument was:

    (C1): In order for disputants to converge in their judgments on whether they

    have reason to hold some mental state, it is necessary and sufficient that they

    agree on some general, substantial criterion for assessing judgments about

    reasons for holding that state.

    But my intended conclusion is not as strong as this. Rather, my intention is to

    draw a weaker conclusion which is compatible with the worries stated above. It is:

    (C2): All else equal, we should be more confident that disputants will

    converge in their judgments on whether they have reason to hold certain

    26 See, for example, Pettit (2001), p. 123.

    C. Cowie

    123

  • mental states if they agree on some general substantial criterion for assessing

    judgments about reasons for holding that state, than if they do not.

    C2 is weaker than C1 in two ways. Firstly, the conclusion is qualified by an all

    else equal clause. It is only if all else is equal that we should expect the disanalogy

    in the correctness conditions to result in a disanalogy in the degree of convergence.

    If, for example, there are more fixed points of convergence in practical judgment

    than in epistemic judgment, then all else may not be equal. Secondly, the conclusion

    is comparative (i.e. it is of the form we should expect a greater degree ofconvergence in x than y). It does not state that we should expect convergence in

    epistemic judgment, or that we shouldnt. Rather, it states that (all else equal) we

    should be more expectant of convergence in the former than in the latter.

    This weakened conclusion is no longer susceptible to the worries raised above.

    But a sceptic may worry that the weakened conclusion fails to represent much of a

    challenge to the metanormative theorist. The sceptic may concede that, all else

    equal, we should expect some positive comparative degree of convergence inepistemic judgment as a result of sharing the reliability for truth criterion. But he

    may claim that the degree is very small indeed. Reliability for truth is, after all, a

    highly abstract criterion, and is not of much use in resolving many epistemic

    disputessuch as that between Stuart and myself. There are two lines of response to

    this sceptical worry. The first line of response is direct. The reliability for truth

    criterion may not settle all epistemic disputes. But it already does a lot of work inhelping to resolve many epistemic disputes. To see this, note that if epistemic

    disputants did not share this criterion, then we ought not to expect them to resolve

    their disputes even in those cases in which they come to agree on how reliable or

    unreliable for truth a certain epistemic judgment is. But this is precisely how

    ordinary, everyday, epistemic disputes are, with great frequency, resolved. The

    sceptical worry may be accused of overlooking this. The second line of response is

    more dialectically focussed. It is that the weakened conclusion implies that there is

    prima facie reason to expect the degree of convergence in epistemic judgment and

    practical judgment to differ. And the metanormative theorist must at least get his

    hands dirty with the arguments to show that the difference in degrees of

    convergence is insufficiently significant to trouble him.

    5.2 Objection 2

    The second objection concerns the correctness conditions of belief. The objection is

    that although the correctness-condition of belief ground some reasons for belief, it is

    not the only source of reasons for belief. Rosens Jingle Bells example helps to

    illustrate the worry.27 There is a correct way to play Jingle Bellsby playing the

    right notes in the right order in the right time. But it may be that one has reason to

    play Jingle Bells incorrectlyfor comic purposes, or to teach a pupil, for example.

    So, we shouldnt assume that one ought to play Jingle-Bells correctly. Analogously,

    perhaps, for belief. There may be reasons for belief that are not exhausted by the

    27 Example from Rosen, in Shah (2003).

    Epistemic Disagreement and Practical Disagreement

    123

  • correctness conditions of belief. Specifically, there may be pragmatic reasons for

    belief.

    One response to this challenge is concessive. Provided that it is accepted that the

    reliability for truth of an epistemic judgment provides some reason to believe thatjudgment, some progress has been made toward showing that practical disagreement

    is problematic in a way that epistemic disagreement is not. This is because we have

    found no criterion of assessment for practical judgment that would be accepted on

    the basis of mere reflection on the relevant concepts. So, the fact that there is some(merely conceptually based) substantial criterion for assessing epistemic judg-

    menteven if there are also some pragmatic criteria of assessmentconstitutes

    progress toward my intended conclusion.

    Nevertheless, this concession would weaken my argument. And it is in any case,

    I think, possible to avoid it. The objection above relies on the existence of non-

    evidential or pragmatic reasons for belief. And the existence of pragmatic reasons

    for belief is questionable.28 One obvious response to the claim that there are

    pragmatic reasons for belief is that such reasons are merely motivating reasons for

    belief. They are not normative reasons for belief. According to this response, it is

    true that one can be caused to form a belief for pragmatic reasonsone can think

    wishfully, for examplebut it is not true that ones belief can ever be justified bypragmatic reasons. Only evidential reasons can justify ones belief. If this obvious

    response is correct, then it is possible to admit that there are pragmatic reasons for

    belief, but to deny that that this undermines my central claim. My central claim is

    that truth is the only norm on belief. I am concerned with normative, not motivating,reasons.

    This therapeutic argument can be extended. It is possible to admit that

    pragmatic concerns provide normative reasons for states very much like belief, just

    not for belief itself. So, for example, pragmatic considerations may provide

    normative reasons for supposing a proposition, or imagining a proposition. But theyneednt thereby provide normative reasons to believe that proposition. The fact thatI desire that P may indeed provide me with a normative reason to suppose that p.

    But it neednt thereby provide me with a normative reason to believe that p. Onceagain, it is possible to make sense of why one might think that there are normative

    pragmatic reasons for belief without admitting that there in fact are any.

    Obviously, this isnt a knockdown argument against (normative) pragmatic

    reasons for belief. That is a project beyond the scope of this paper. But it goes a long

    way to undermining the motivations for thinking that there are pragmatic reasons for

    belief. So, it supports the view that the only reasons for believing a conclusion take

    the form of evidence that the belief is correct.

    28 There are a number of arguments against the existence of pragmatic reasons for belief in the literature.

    For example, Thomson has argued that pragmatic reasons for belief represent a category mistake: if

    there were pragmatic reasons for belief then one could believe whatever one wants to; but one cant

    believe whatever one wants to; so there are not pragmatic reasons for belief. Thomson (2008), p. 138.

    Relatedly, Pamela Hieronymi has argued that pragmatism about belief explains the impossibility of

    believing at will Hieronymi (2006). See also Shah and Vellemans argument that the psychological

    phenomenon of the transparency of belief is best explained by evidentialism. See Shah (2003), Shah and

    Velleman (2005).

    C. Cowie

    123

  • 5.3 Objection 3

    The third objection concerns my claim that there is no substantial general norm on

    action, or action-guiding mental states on which we should expect agreement. It

    might be objected that this claim is too strong: there must be some substantialgeneral norm on action, or action-guiding mental states. This source of resistance

    might be motivated by interpretative constraints associated with Davidson.

    According to Davidson, in order for radical interpretation to be possible, we must

    assume that the desires of any agent we are interpreting overlap (to some degree)

    with ours.29 But if my argument is sound, then one might worry that there is no

    reason to assume that the desires (or as I have said, choices) of other agents would

    overlap with ours at all. So, we would have no reason to think that radical

    interpretation is possible. But radical interpretation is possible. So, my argument is

    unsound. We can express this objection in terms of the following argument:

    1. If there is no substantial general norm on desire, then one could rationally

    regard anything as desirable;

    2. If one could rationally regard anything as desirable, then there is no reason to

    think that two rational disputants would overlap at all in their desires;

    3. It is necessary for two rational disputants to be able to interpret one another that

    there is some overlap in their desires;

    4. Any two rational disputants could interpret one another;

    5. So, there is some substantial, general norm on desire.

    How should we respond to this argument? One obvious response is to reject

    premise 4. We shouldnt assume that any two rational disputants could interpret one

    another. In a sense, Davidson would agree. We certainly shouldnt assume that any

    two rational disputants could interpret one another perfectly. However, Davidsondoes seem to be committed to at least the possibility of partial interpretative success.

    And he also seems to be committed to the claim that at least some overlap in desire

    is required for even partial interpretative success. So, Davidson might alter premises

    3 and 4 to:

    3*. It is necessary for two rational disputants to be able to partially interpret oneanother that there is some overlap in their desires.

    4*. Any two rational disputants could partially interpret one another.

    How should we respond to this modified argument? It is possible to reject either

    3* or 4*.30 Suppose, however, that we grant both 3* and 4*. The resulting argument

    still fails as it doesnt entail 5. To see this, note how weak 3* is. According to 3*, all

    rational disputants must have some degree of overlap in their desires. Specifically,

    they must have sufficient overlap in desires for partial interpretation to be possible.

    29 We must assume that he is a lover of the good (as well as a believer of the truth). A theme discussed

    in Davidson (1986, 1995).30 Tersman (2004) rejects 3*. He argues that in order for interpretation to be possible, it is only necessary

    that agents share certain logical (i.e. decision-theoretic) and functional properties of their desires. It is not

    necessary that there is any overlap in the contents of their desires. Tersman (2006) extends this modelexplicitly to moral judgments.

    Epistemic Disagreement and Practical Disagreement

    123

  • But this is too weak to imply 5 (i.e. to imply that there must be some substantial

    general norm on desire). Establishing 5 would require the stronger claim:

    3**. It is necessary for rational disputants to be able to partially interpret one

    another that they share a substantial general criterion for assessing desires.

    And there is no reason to think either that Davidson makes this claim, or that he

    or anyone else would be entitled to make it. So, there is no reason to think that the

    argument from interpretative constraints entails 5. So, there is no reason to think

    that it undermines my overall argument.

    This response might seem unsatisfactory. My argument still leaves something

    unexplained. There is, as it happens, a good deal of overlap in judgments about what

    is and is not desireable, across times and places. But if desire is not governed by a

    substantial general norm, then this might appear mysterious. If there isnt a

    substantial general norm on desire, then why should we expect there to be any

    convergence in judgments about desireability at all? My response is twofold. Firstly,

    although it is true that there is overlap in judgment about desireability, there is also a

    great deal of disagreement. This shouldnt be under-estimated. Secondly, it is

    possible to explain what overlap there is without supposing that desire is governed

    by any substantial general norm. It is, as Tersman notes, possible to explain overlap

    in desire in terms of our common social and biological heritage rather than in

    meaning-theoretic terms.31

    6 Conclusion

    I have claimed that, owing to a disanalogy between the correctness conditions of

    belief and action-guiding mental states, we have reason to expect a greater degree of

    convergence in judgment about reasons for belief than in judgment about reasons

    for action. And I have drawn the conclusion that this represents a prima facie

    challenge to unified or metanormative approaches to reasons for belief and reasons

    for action.

    How strong is this challenge? Im not sure. I concede that it is possible for a

    metanormative theorist to deny the force of the challenge in a number of ways.

    Consider the following three options. Firstly, one might deny the premise that the

    scope of disagreement in an area of thought is of significance for the metaphysics of

    that area of thought. This is a response that some metanormative theorists will be

    happy to take, others not. Secondly, one might deny the force of the challenge by

    rejecting the approach to normativity taken above: namely, that reasons for mental

    states can be grounded in the nature of those states. Again, this is a response

    approach that some metanormative theorists will be happy to take, others not.

    Thirdly, and perhaps most importantly, a metanormative theorist may regard the

    considerations in favour of metanormative approaches to outweigh arguments

    against. There are after all, considerable advantages to the metanormative approach:

    the shared normativity of reasons for action and reasons for belief, and the difficulty

    31 Tersman (2004), p. 258.

    C. Cowie

    123

  • (in some cases) of cleanly distinguishing between those concepts and/or judgments

    that belong to the realm of theoretical normativity as opposed to the realm of

    practical normativity.32 How one reacts to my argument will, then, depend on ones

    prior commitments, and on the results of ones philosophical scorekeeping. There

    are some metanormative theorists who, given their prior commitments, could make

    use of any one of the three responses suggested above in order to reject the

    argument that I have provided (David Enoch is the best example). But none of these

    responses are mandatory for metanormative theorists. For some, their prior

    commitments may be such that the argument I have presented provides food for

    thought. If so, then it is an argument worth making.

    I shall conclude on a more positive note by briefly noting an interesting

    connection between the argument that I have offered and a related argument for a

    different conclusion elsewhere in the literature. Darwall offers an argument much

    like the argument that I have offered above.33 He claims that belief is normatively

    governed by truth. And he explains this in terms of the nature of the concept of

    belief. Furthermore, he contrasts this with the norms governing mental states that

    are relevant for practical judgment (he focuses on desire). Such norms are, he

    claims, uninformative or merely formal. The resemblance between Darwalls

    argument and the argument that I have offered should be clear. However, the use to

    which we put these thoughts differs.

    I have argued that these thoughts imply a disanalogy in practical and epistemic

    disagreement. Darwall also takes these thoughts to imply a significant disanalogy.

    But he is not concerned with disagreement. His concern is with freedom. He

    claims that whereas reasons for belief are constrained by objective truth and

    probability, reasons for acting can be grounded nowhere but within norms of free

    practical reflection itself. Darwalls thought is that the nature of belief, determines

    what can count as a reason for belief (i.e. evidence for the truth of that belief). But,

    by contrast, the nature of action-guiding mental states doesnt completely determine

    what can count as a reason for action. Hence, conceptually competent agents have a

    freedom in working out what to do that they lack in working ought what to believe.

    These conclusionsmine concerning disagreement and Darwalls concerning

    freedommight be regarded as corollaries of one another. The nature of action-

    guiding mental states doesnt restrict the practical judgment of conceptually

    competent agents in anything like the sense that the nature of belief restricts

    epistemic judgment. A comparative practical freedom may be one consequence of

    this lack of restriction. Likely disagreement is another. That these two features of

    practical judgmentfreedom and the scope of expected disagreementare

    explained by the same facts is an interesting result.

    References

    Brink, D. (1989). Moral realism and the foundations of ethics. Cambridge: CUP.

    32 For more on this point see e.g. Cuneo (2007), pp. 7780, and Enoch (2011a), p. 241.33 Darwall (2003), pp. 483488.

    Epistemic Disagreement and Practical Disagreement

    123

  • Chang, R. (1997). Introduction. In R. Chang (Ed.), Incommensurability, incomparability, and practicalreason. Cambridge: Harvard University Press.

    Chrisman, M. (2010). From epistemic expressivism to epistemic inferentialism. In D. Haddock, A. Millar,

    & A. Pritchard (Eds.), Social Epistemology. Cambridge: OUP.Cuneo, T. (2007). The normative web. New York: OUP.Darwall, S. (2003). Moore, normativity, and intrinsic value. Ethics, 113, 463489.Davidson, D. (1986). The objectivity of values. In M. Cavell (Ed.), Collected papers vol. 4. problems of

    rationality. New York: OUP.Davidson, D. (1995). The interpersonal comparison of values. In M. Cavell (Ed.), Collected papers vol. 4.

    problems of rationality. New York: OUP.Dworkin, R. (2011). Justice for hedgehogs. Harvard: Harvard University Press.Elga, A. (2007). Reflection and disagreement. Nous, 41(3), 478502.Enoch, D. (2006). Agency, Schmagency: Why normativity wont come from what is constitutive of

    action. Philosophical Review, 115, 169198.Enoch, D. (2009). How is moral disagreement a problem for realism. Journal of Ethics, 13(1), 1550.Enoch, D. (2011a). Taking morality seriously. New York: OUP.Enoch, D. (2011b). Schmagency revisited. In M. Brady (Ed.), New waves in metaethics. New York:

    Palgrave Macmillan.

    Evans, M., & Shah, N. (2012). Mental agency and metaethics. In R. Shafer-Landau (Ed.), Oxford studiesin metaethics (Vol. 7). Oxford: OUP.

    FitzPatrick, W. (2009). Recent work on ethical realism. Analysis, 69(4), 746760.Gibbard, A. (2003). Thinking how to live. Cambridge, MA: Harvard University Press.Harman, G. (1996). Moral relativism. In G. Harman & J. J. Thompson (Eds.), Moral relativism and moral

    objectivity. Cambridge, MA: Blackwell.Harman, G. (2000). Moral relativism defended. In G. Harman (Ed.), Explaining value: And other essays

    in moral philosophy. Oxford: Clarendon Press.Heathwood, C. (2009). Moral and epistemic open question arguments. Philosophical Books, 50, 8398.Hieronymi, P. (2006). Controlling attitudes. Pacific Philosophical Quarterly, 87(1), 4574.Jackson, F. (1998). From metaphysics to ethics. Oxford: Clarendon.Kearns, S., & Starr, D. (2009). Reasons as evidence. Oxford Studies in Metaethics, 4, 215242.Kelly, T. (2003). Epistemic rationality as instrumental rationality: A critique. Philosophy and

    Phenomenological Research, 66(3), 612640.Kelly, T. (2007). Evidence and normativity: Reply to Leite. Philosophy and Phenomenological Research,

    75(2), 465474.Kornblith, H. (2010). Belief in the face of controversy. In R. Feldman & T. A. Warfield (Eds.),

    Disagreement. Oxford: OUP.Leite, A. (2007). Epistemic instrumentalism and reasons for belief: A reply to Tom Kellys Epistemic

    Rationality as Instrumental Rationality: A Critique. Philosophy and Phenomenological Research,75(2), 456464.

    Lillehammer, H. (2004). Moral error theory. Proceedings of the Aristotelian Society, 104(2), 93109.Lillehammer, H. (2007). Companions in guilt: Argument for ethical objectivity. Basingstoke: Palgrave

    MacMillan.

    Mackie, J. (1977). Ethics: Inventing right and wrong. New York: Penguin.Millar, A. (2004). Understanding people: Normativity and rationalizing explanation. Oxford: OUP.Parfit, D. (2011). On what matters. New York: OUP.Pettit, P. (2001). Two sources of morality. Social Philosophy and Policy, 2, 102128.Scanlon, T. M. (1998). What we owe to each other. Cambridge, MA: Harvard University Press.Shah, N. (2003). How truth governs belief. Philosophical Review, 112(4), 447482.Shah, N. (2008). How action governs intention. Philosophers Imprint, 8, 119.Shah, N., & Velleman, D. (2005). Doxastic deliberation. Philosophical Review, 114(4), P497P534.Skorupski, J. (2011). The domain of reasons. New York: OUP.Steglich-Petersen, A. (2011). How to be a teleologist about epistemic reasons. In A. E. Reisner & A.

    Steglich-Peterson (Eds.), Reasons for belief. Cambridge: Cambridge University Press.Tersman, F. (2004). Are we lovers of the good? Synthese, 138(2), p247p260.Tersman, F. (2006). Moral disagreement. New York: Cambridge University Press.Thomson, J. J. (2008). Normativity. New York: Open Court.Velleman, D. (1992). The guise of the good. Nous, 26(1), 326.Velleman, D. (1996). The possibility of practical reason. Ethics, 106(4), 694726.

    C. Cowie

    123

  • Wedgwood, R. (2002). The aim of belief. Philosophical Perspectives, 16(s), 267297.Wedgwood, R. (2003). Choosing rationally and choosing correctly. In S. Stroud & C. Tappolet (Eds.),

    Weakness of will and practical irrationality. Oxford: Oxford University Press.Wedgwood, R. (2007). The nature of normativity. Oxford: OUP.

    Epistemic Disagreement and Practical Disagreement

    123

    Epistemic Disagreement and Practical DisagreementAbstractDisagreement and Shared CriteriaFormal and Substantial CriteriaCorrectness ConditionsExtending the Argument: ActionObjections and RepliesObjection 1Objection 2Objection 3

    ConclusionReferences