Moral Judgments, Cognitivism and the Dispositional Nature of Belief: Why Moral Peer Intransigence is Intelligible

Abstract

Richard Rowland has recently argued that considerations based on moral disagreement between epistemic peers give us reason to think that cognitivism about moral judgments, i.e., the thesis that moral judgments are beliefs, is false. The novelty of Rowland’s argument is to tweak the problem descriptively, i.e., not focusing on what one ought to do, but on what disputants actually do in the light of peer disagreement. The basic idea is that moral peer disagreement is intelligible. However, if moral judgments were beliefs, and beliefs track perceived evidence, then moral peer disagreement would not be intelligible. Hence, moral judgments are not beliefs. The argument is both novel and interesting, but this paper argues that it fails to establish the conclusion. Beliefs are plausibly analyzed as constituted by dispositions to respond to what is perceived as evidence, but dispositions can always be interfered with. Provided a background explanation of why the disposition is not manifested, peer intransigence is quite intelligible.

1 Introduction

In a recent paper, Richard Rowland (2018) argues that considerations based on moral disagreement between epistemic peers give us reason to think that cognitivism about moral judgments, i.e., the thesis that moral judgments are beliefs, is false.¹ The peer disagreement debate mainly concerns the rational response to peer disagreement, e.g., whether one is rationally required to revise one’s belief on the disputed matter upon discovering that one’s peers disagree. The novelty of Rowland’s argument is to tweak the problem descriptively, i.e., not focusing on what one ought to do, but on what disputants actually do in the light of peer disagreement. The basic idea is that moral peer intransigence is intelligible. This is, in part, motivated by intuitions about cases. For example, Rowland considers the following case. Anna judges that torture is always forbidden. However, she knows that many of those she regards as her epistemic peers disagree. Despite acknowledging this, Rowland claims, it seems that Anna’s moral judgment about torture can be wholly unaffected. Moreover, as Rowland observes, there are a number of philosophers (Kalderon 2005; Kramer 2009, and Setiya 2012) who think that moral peer intransigent is intelligible. These considerations, Rowland argues, make the following claim intuitively plausible.

Moral Peer Intransigence is Intelligible. For any agent A, it is intelligible that both (a) A’s non-derivative moral judgment about the moral status of ϕ-ing and (b) A’s judgment about who her epistemic peers about the moral status of ϕ-ing could be simultaneously entirely intransigent in light of her judging that a significant number of her epistemic peers about the moral status of ϕ-ing disagree with her about the moral status of ϕ-ing. (267)

However, if moral judgments were beliefs, and beliefs track perceived evidence, then moral peer intransigence would not be intelligible because perceived peer disagreement is perceived evidence. Hence, moral judgments are not beliefs.² This is both a novel and interesting argument. In this paper, we nevertheless argue that peer intransigence is intelligible given that, which seems plausible, beliefs are understood as dispositional mental states. Peer intransigence is thus not a problem for cognitivism. Before we turn to the details, let us outline the different claims that the argument builds on.

• 1. Moral Peer Intransigence is Intelligible

• 2. Beliefs Track Perceived Evidence

• 3. Perceived Peer Disagreement is Perceived Evidence

(2) and (3) entails

• 4. Peer Intransigent Judgments are not Beliefs.

(1) and (4), in turn, entails

• 5. Non-derivative moral judgments are not beliefs.

This, Rowland argues, presents a dilemma for the cognitivist. The cognitivist must reject either (1) or (4), but “we have good reason not to reject either of these claims” (274). We will not dispute (1). In other words, moral peer intransigence is intelligible. Rather, we will argue that the cognitivist has good reasons to reject (4). In the discussion below, our focus will primarily be on (2). Moreover, explaining the grounds for rejecting (4) will also serve to explain why moral peer intransigence is not a problem for cognitivism.

2 Peer Disagreement and Perceived Evidence

If John were to consider whether there are people who disagree with him regarding what he believes, chances are that he would discover that there are peers that disagree with him on a host of different issues.³ The peer disagreement debate concerns the rational response to such discoveries. A classic example used to illustrate the phenomenon involves two parties splitting a bill at a restaurant, but end up with different sums. Given that they are peers, it seems that both parties acquire evidence to the effect that at least one of them have made a mathematical mistake. Peer disagreement is most often understood as a kind of “higher-order evidence.” Higher-order evidence is, broadly, evidence about one’s evidence: typically evidence that indicates that something has gone wrong in the belief-forming process behind one’s belief regarding p (see e.g., Christensen (2010) and Lasonen-Aarnio (2014) for discussion).

In the peer disagreement debate, there are mainly two opposing camps. According to conciliatory views, higher-order evidence makes one rationally required to suspend judgment or at least to significantly reduce one’s confidence regarding the truth of the disputed proposition. According to steadfast views, by contrast, higher-order evidence does not make one rationally required to suspend judgment.⁴ Nevertheless, Rowland claims that “it is generally presumed by all parties in the peer disagreement literature that Perceived Peer Disagreement is Perceived Evidence” (272).

Perceived Peer Disagreement is Perceived Evidence. When one comes to believe that someone one believes to be one’s epistemic peer regarding p disagrees with one regarding p, (other things equal) one comes to believe that there is (strong) evidence against one’s view regarding p that one did not previously have regarding p. (272)

There are two issues that we want to highlight here. First, even if it is true that the parties in the peer disagreement debate agree that peer disagreement is evidence, there is, as explained above, ample disagreement regarding what one ought to do when one discovers that a peer disagrees. According to steadfast views, there are cases where one ought to stick to one’s belief that p despite discovering that a peer disagrees regarding p, i.e., despite coming “to believe that there is (strong) evidence against one’s view regarding p that one did not previously have regarding p.” Interestingly, a steadfast view thus seems to presuppose that beliefs can be intransigent in the sense that Rowland questions – at least assuming that ought implies can. Second, it does not seem to matter whether a conciliatory or steadfast view is correct. Rather, what matters is whether an agent like Anna perceives of peer disagreement as evidence against her belief that p. Suppose that perceived peer disagreement regarding p is perceived evidence against p. Moreover, suppose that John believes that many peers disagree with him regarding p, but that his belief is completely intransigent. This does not strike us as strange at all. Indeed, we think that people are intransigent in this sense all the time.⁵ One explanation is that many people do not perceive peer disagreement as evidence – even if it actually is evidence. Similarly, a bad detective may fail to realize that John’s fingerprints on the murder weapon is evidence to the effect that John is the murderer.

Why is this important? Given Rowland’s claim, believing that a peer disagrees regarding p is to believe something that is evidence against p, but, as the examples above illustrate, perceiving evidence in this sense is consistent with failing to perceive it as evidence – again assuming that it is evidence. When Rowland describes Anna, we are told that she judges that torture is always forbidden and that she knows that many of those she regards as her epistemic peers disagree. However, Rowland does not explicitly say that Anna also believes that peer disagreement regarding the permissibility of torture is something that she perceives as strong evidence against her view. Again, there is nothing odd about Anna if she judges that torture is always wrong and believes that many of her peers disagree if she does not think of peer disagreement as evidence against her judgment. Notice also that whether e.g., a conciliationist view is true does not seem to matter. Rather, the only thing that matters is whether one perceives peer disagreement regarding p as evidence against p. In other words, Anna judges that torture is wrong. When she learns that many of her peers disagree, she comes to believe that there is strong evidence against judging that torture is wrong. However, the change in what she perceives of as evidence does nothing to change her judgment. In order for Rowland’s argument to get off the ground, it is a case like this that we must find intelligible. We will not dispute the intelligibility of Anna given such an interpretation. However, as we will try to explain below, we do not think the intelligibility of intransigent Anna shows that cognitivism is problematic.

3 Beliefs Do Not Invariably Track Perceived Evidence

In order to properly assess whether moral peer intransigence is a threat to cognitivism we need to closer examine the nature of belief. What makes a mental state a belief rather than some other attitude, e.g., a fantasy? An answer that seems to be gaining in popularity is that beliefs are attitudes that are evidence responsive in a distinct way.⁶ This idea can be understood as either a normative or descriptive claim. Rowland understands evidence responsiveness as descriptive.⁷

Beliefs Track Perceived Evidence. If one gains what one believes to be new evidence that bears on whether p, and one’s judgment regarding p (whether p) is a belief, then either (i) one adjusts one’s judgment regarding p in the light of and in line with this evidence or (ii) one adjusts one’s judgment that what one gained was in fact evidence bearing on whether p. (270)

If a mental state is not evidence responsive, i.e., if it does not actually track changes in what one perceives as evidence, then it is not a belief. Rowland taps three sources to this effect. The first concerns our reactions to cases like the following.

I used to believe that stomach ulcers were caused primarily by stress and diet; but when Warren and Marshall’s research on the Heliobacter pylori bacterium became widely known, I revised my belief to reflect this information. (Gendler 2010: 296)

A belief, in Tamar Gendler’s words, “is subject to immediate revision in the face of changes in our all-things-considered evidence” (Gendler 2010: 296).⁸ The second reason concerns epistemic akrasia. Rowland argues that the following thesis is plausible.⁹

No Akrasia. It is not intelligible that we might consciously believe that p whilst simultaneously consciously believing that our evidence does not justify us in believing that p. (270)

For example, to consciously believe that flying is really dangerous and simultaneously consciously believe that all the evidence shows that flying is not particularly dangerous seems to be a combination of beliefs that is not obviously intelligible.

The third consideration concerns deliberative transparency. The idea is that only certain considerations can be thought of as relevant to doxastic deliberation: “when we think about what to believe regarding the winner of the next Presidential election, we only think about considerations that bear on who the winner will in fact be, that is, evidence regarding who the winner is likely to be” (271). For example, that a certain candidate makes me happy or wears fashionable clothes are not considerations that are relevant to the truth of the matter and therefore not something that can be thought of as relevant to what to believe. Rather, it is only considerations that are recognized as relevant to the truth that matters. Indeed, such considerations, as Nishi Shah claims, “must be immediately seen as relevant to determining whether to believe that p“(Shah 2003: 453).¹⁰

We agree with Rowland that these considerations provide a firm rationale for thinking that evidence responsiveness is a distinguishing feature of belief. In brief, it is constitutive of a belief that it changes in response to changes in what is perceived as evidence. Nevertheless, we worry that the connection that Rowland needs for the argument to work will be too demanding. It is not sufficient, for example, that beliefs normally track what is perceived as evidence. Rather, what is required is that beliefs invariably track what is perceived as evidence. If one can have a belief that does not change in response to changes in what is perceived as evidence, moral peer intransigence does not show that moral judgments are beliefs. Hence, in order for the argument to work we must reformulate (2) as follows:¹¹

• 2*. Beliefs Invariably Track Perceived Evidence

In other words, it is impossible to have a belief that does not change in response to changes in what is perceived as evidence. The problem is that this claim is not obviously true. Some philosophers deny that responding to evidence is not constitutive of belief.¹² Instead, one may think that something like (1) sincere assertion, (2), action guidance, and/or (3) inferential promiscuity is all that is required. These responses would be easy ways of resisting the argument. However, as emphasized above, we think that the considerations Rowland rehearses are compelling.¹³ Evidence responsiveness is plausibly constitutive of belief. Moreover, the argument may be interpreted as mainly targeting cognitivists sympathetic to the claim in question. Nevertheless, the claim does not warrant thinking that beliefs invariably change in response to changes in what is perceived as evidence.

Before we explain why, let us make two brief points. First, we will focus only on the maintenance of beliefs – not their formation. For example, it seems quite possible to acquire a belief in many different ways that are not responsive to what is perceived as evidence, e.g., by virtue of wishful thinking or the like. Second, it is important to be precise about what “track what is perceived as evidence” requires. In some cases, tracking what is perceived as evidence may lead to abandoning the belief, but at other times one will merely become less certain in one’s belief. The latter, as we understand it, is sufficient for a belief to track perceived evidence. Intransigence thus requires that the belief does not change period.

Why think that (2*) is too demanding. One reason is based on intuitive examples like the following. John believes that he is a better than average driver. He has just read a paper that explains that most men overrate themselves qua drivers. John realizes that the considerations he has just learnt show that he has no reason to think that he is any better than the average driver. He thus comes to believe that the considerations provide evidence to the effect that he overrates himself qua driver. However, it seems conceivable that John’s belief is nevertheless left entirely unchanged. Not only does he not give it up, but his belief is just as strong as it was before he acquired the evidence against it.

If this is right, then it seems possible for an agent to have a belief that fails to track what is perceived as evidence. Of course, it may be argued that the example is problematic. First, it may be argued that John engages in some kind of rationalization that functions to explain away the evidential import of the evidence. Although we engage in rationalizations all the time, this is not what we have in mind. Rather, both beliefs remain fixed. Second, John’s “belief” may be interpreted as an evaluative judgment: John judges that he is a good driver. What we need is a non-evaluative or non-normative example. This problem is easily fixed. “Good driver” may be interpreted descriptively, e.g., what John believes is that he is statistically less likely to get into an accident than the average driver. Moreover, there are many similar cases. John also believes that his son is not capable of shoplifting. When presented with surveillance footage, he may grant there is evidence to the effect that his son is actually guilty of shoplifting. However, it still seems conceivable that there is no change in John’s belief in response to the change in what he perceives of as evidence against it. Indeed, we think that most people are familiar with examples where a person’s belief is intransigent in such a sense.

It may be objected that the considerations Rowland advances in support of Beliefs Track Perceived Evidence suggests otherwise. If John’s mental state is not invariably evidence responsive it cannot be a belief. It may be argued that doxastic deliberation, i.e., whether to believe that p is always determined and exclusively determined by the question of whether p is true. It is indeed plausible to think that the deliberative question is determined by the question of whether p is true where this end is guided by considerations regarding what is perceived as evidence, i.e., considerations that one perceives of as relevant to the truth of p. In other words, we cannot consciously appeal to considerations that we perceive of as irrelevant to the truth of p in doxastic deliberation. However, this is consistent with other forces working unconsciously in the background. For example, when John considers whether to believe that his son is guilty of shoplifting, the only considerations that are relevant to his deliberation are considerations that he thinks are relevant to the truth of the belief he holds. John cannot consciously consider that he wants his son to be innocent as a consideration that is relevant to what to believe (unless, of course, he for some reason thinks this is relevant to the truth of the matter). However, this does not rule out that non-evidential forces play a non-deliberative role in belief-maintenance (see e.g., Owens 2000 chapter 2). In other words, non-evidential considerations may work silently in the background making the evidential considerations impotent. Indeed, various non-evidential factors, e.g., self-image, values, hopes, desires, and so on, may plausibly non-deliberatively/non-transparently influence doxastic deliberation. For example, when John is confronted with what he sees as evidence against his belief that his son is innocent, it is likely to stir up various painful emotions. Those emotions can, in turn, prevent his belief from changing in response to changes in what is perceived as evidence. Similarly, assuming that being a good driver is central to John’s self-conception, perceiving evidence that threatens this will also activate various psychological defense mechanisms. Grace Helton, advances an example that is in important respects similar to the one’s advanced above.

Suppose you have a much-loved friend who is accused of stealing from you. You believe your friend to be innocent, despite the overwhelming evidence against her. In fact, your affect for your friend is so strong that you are psychology incapably of revising your belief in your friend’s innocence. (Helton 2018: 17–18)

The idea here, like the idea above, is that certain psychological features of an agent may explain why a belief does not change in response to changes in what is perceived as evidence. By contrast, if Grace walks into the kitchen believing that there is coffee made, but finds the coffee pot empty, it seems much less plausible to think that she will maintain her belief.¹⁴ The explanation is that it is very farfetched to think that the belief about coffee is somehow central or important to Grace. There will thus be no threat that triggers a psychological defense mechanism that prevents the belief from changing. In other words, certain beliefs are more difficult to give up because they are central to who we are, our values or the like. Indeed, Shah seems to allow for such factors to play a role in the maintenance of a belief, but a non-evidential factor, “exerts its influence only when it is masked” (Shah 2013: 315).

If this is right, then it seems that even if deliberative transparency is correct, i.e., that deliberation about p is exclusively determined by considerations that one perceives of as relevant to the truth of p, other factors may influence the maintenance of a belief unconsciously. The fact that a belief is central to John’s self-image or his values, may prevent what he perceives as truth-related considerations from changing his belief. This is just how the complicated human psyche works. A belief that does not change in response to changes in what is perceived as evidence may be odd, but it does not seem impossible or inconceivable. Moreover, this also explains why akratic combinations of mental states are possible. John believes that p, but simultaneously believes that the evidence does not justify the belief that p. However, as explained above, there may be various explanations of why John’s belief does not change in response to what he perceives as evidence that makes such combinations of attitudes intelligible.

4 The Dispositional Nature of Beliefs

We can think of the considerations above as drawing on an interpretative view of mind. However, these considerations can be supplemented with a dispositional account of beliefs. In metaethics, the explanation of what a belief is seldom goes beyond a very basic comparison to desires, viz., beliefs, by contrast to desires, have a mind-to-world direction of fit. In other words, they purport to represent the world (rather than to change the world to fit one’s desires).

It seems very plausible to think of beliefs, and other attitudes as dispositional mental states (see e.g., Schwitzgebel 2013). Indeed, a dispositional account of beliefs lines up nicely with the idea that beliefs aim to represent the world. To represent the world is roughly to aim at truth. However, we do not hit the target, i.e., truth, automatically. Rather, the means to the end is considerations we perceive of as evidence (or truth-indicative). A belief that p is thus plausibly constituted by a disposition to respond to considerations that are perceived as relevant to the truth of p.¹⁵ In metaethics, a view along these lines is suggested by Michael Smith. “A belief with the content that p tends to go out of existence in the presence of a perception with the content that not-p, whereas a desire tends to endure, disposing the subject in that state to bring about that p” (Smith 1994: 115). Of course, a mere perception with the content that not-p is not always enough as is evident from various optical illusions e.g., the perception of a “bent stick” in water or the Müller-Lyer illusion. Nevertheless, the basic idea, as we understand it, is that a belief (at least in part) is constituted by a disposition to respond to considerations that are perceived of as evidence. This is what distinguishes a belief that p from other attitudes, e.g., pretending that p.

But what is a disposition? This is a question that has many answers. Fortunately, for our purposes, we need only scratch the surface of this complicated and intriguing issue. A fragile glass, for example, is disposed to break when struck. A classic analysis of disposition is in terms of simple conditionals, viz., a fragile glass is disposed to break when struck if it would break if struck. A standard counterexample to such an analysis involves a glass protected by a support structure. When protected by a support structure, the glass is struck, but the glass does not break. Yet, given that the intrinsic properties of the glass remain, it does not seem to lose its fragility. This illustrates three important things. First, something can have a disposition although the disposition is not manifested. Second, a disposition can be interfered with, i.e., prevented from being manifested. Third, even if the “stimulus condition” is satisfied (the glass is struck), the disposition need not manifest itself (the glass does not break).

These brief remarks are important to understand the dispositional underpinnings of mental states. Start with desires. We can think of a desire as a disposition to bring about motivation. For example, John has a standing desire to exercise on Saturdays. During the week, this desire is mostly dormant. It does not issue in any motivation relevant to satisfy the desire. However, when John realizes that it is Saturday, he becomes motivated in various ways, e.g., he starts packing his bags and heads to the gym. Although the disposition to bring motivation about is constitutive of a desire, it can also be interfered with in different ways. Suppose, for example, that John has recently become severely depressed. This state may interfere with the normal causal chain leading from input to output and thus prevent John’s desire to exercise from manifesting itself. In other words, the desire has no motivational upshots. If John would not have been depressed, however, the desire would have motivated him in the usual ways.¹⁶

Beliefs plausibly mirror desires in this respect. Suppose that a belief is partly constituted by a disposition to change in response to changes in what is perceived as evidence. Just like the disposition to motivate can be interfered with in the case of desires, the disposition to change in response to changes in what is perceived of as evidence can be interfered with in the case of beliefs.

Consider John again. John’s beliefs usually change in response to changes in what he perceives as evidence. However, the examples considered above are examples where this does not happen. John believes that his son is innocent. Confronted with the surveillance footage, he does indeed come to believe that there is evidence against his belief. However, as argued above, we think that there are various ways in which the evidential belief can be rendered impotent. At this point, the interpretative view and the dispositional view come together nicely. One explanation of why John’s belief that his son is innocent is left unchanged despite John simultaneously believing that there is evidence that speaks against it is because the former belief is, for some reason, too painful to revise or give up. This leads to various unconscious defense mechanisms to kick in that functions to interfere with the normal causal chain leading from a change in what is perceived as evidence to a change in belief. In other words, just like depression may interfere with a desire manifesting itself in motivation, the threat to one of one’s most central beliefs may prevent it from responding to a change in what is perceived as evidence against it. As argued above, provided a certain background explanation, it seems plausible to think that John’s belief does not change in response to the change in what he perceives as evidence. By contrast to the interpretative view, the dispositional view provides a more theoretical explanation of how this is possible. It seems plausible to think that dispositions can be interfered with in different ways. This is just part of what it is to be a disposition. Hence, if a belief is a disposition to change in response to changes in perceived evidence, we should just expect this to be possible.

However, this does not mean that John’s belief may fail to track perceived evidence in just any old way. Rather, to believe something (just like it is to have some other attitude) “is to embody a certain broad-ranging actual and counterfactual pattern of activity and reactivity” (Schwitzgebel 2013: 76). First, in order to determine whether someone has a belief (or any other mental state) it does not suffice to merely look at what actually happens. Even if John’s mental state does not change in response to the change in what he perceives as a change in evidence, this does not show that his mental state is not a belief. In order to more adequately try to determine whether John’s mental state is a belief, we also have to speculate about various counterfactuals. The hypothesis is that John’s mental state is a belief and thus constituted by a disposition to respond to what is perceived of as evidence. The explanation of why John’s belief does not change in response to a change in what is perceived as evidence is, roughly, that certain beliefs are more central to who we are, what we value and the like. Like a fragile glass, some of these beliefs are protected by a kind of psychological support structure that makes them immune to change even in the light of perceiving a change in evidence. However, if John’s mental state were to lose its centrality or importance, we would no longer have a background explanation of why the disposition to change in response to changes in perceived evidence is interfered with. If John’s mental state would not change in such a situation, we would have reason to think that it is not a belief.

Even if we normally change our beliefs in response to changes in what is perceived of as evidence, it seems that there are circumstances where beliefs are intransigent. However, this requires some kind of background explanation. Such background explanations function to (1) give us a story that makes sense of interpreting someone as having a belief that does not play its normal role and (2) explain what it is that interferes with the disposition to change in response to changes in what is perceived as evidence. Consequently, if we understand beliefs as dispositional states and grant that dispositions can be interfered in different ways, then we have reason to think that beliefs usually track what is perceived as evidence, but that beliefs do not invariably track what is perceived as evidence.

Moreover, the dispositional account of belief can be used to explain the intuitions regarding the sources that Rowland uses to support Beliefs Track Perceived Evidence. A belief is (at least part) constituted by a disposition to respond to what is perceived of as evidence. This plausibly explains deliberative transparency, i.e., why doxastic deliberation is exclusively determined by considerations that one perceives of as relevant to the truth of p. Given the dispositional view, the explanation of why we only appeal to considerations that are relevant to the truth is that these are the kinds of considerations that the belief is disposed to respond to. A consideration, e.g., that a certain belief feels good, is not a consideration that is truth-indicative (unless one has a particular background theory according to which it is) and thus not a consideration that a belief is disposed to respond to – it is not the kind of stimulus condition that can bring a change about. This is why we cannot consciously appeal to any other considerations apart from considerations that are perceived of as relevant to the truth of the matter.

This is consistent with other forces influencing what we believe that are outside doxastic deliberation (and control). In particular, the threat to a belief that is central or important can elicit various defense mechanisms that prevent the change in what is perceived as evidence from bringing about a change in belief. In such a case, both beliefs remain fixed. These considerations also explain why akratic combinations of doxastic attitudes are intelligible. Of course, most of the time, akratic combinations of doxastic attitudes are unintelligible. However, if we are given an explanation of why the disposition to change in response to changes in what is perceived as evidence is interfered with, then we are also given an explanation of why an akratic combination of doxastic attitudes is both possible and intelligible.

Finally, Rowland seems sympathetic to Gendler who claims that beliefs are “subject to immediate revision in the fact of changes in our all-things-considered evidence” (Gendler 2010: 296). This is perhaps true for the large majority of our beliefs. However, it does not seem to be true for all beliefs. Schwitzgebel seems to agree: “[o]ur morally most important beliefs […] the ones that reflect our values, commitments, our enduring ways of viewing the world – they’re not like this. They change slowly, painfully, effortfully” (Schwitzgebel 2010: 547). Indeed, if our argument is right, some beliefs may even remain fixed despite a change in what is perceived as evidence. Again, if beliefs are constituted by dispositions to respond to what is perceived as evidence, we should expect that there are ways in which the disposition can be interfered with (just like depression can interfere with a desire causing motivation), e.g., due to various threats that set of various defense mechanisms.

The dispositional view advanced above explains why it is plausible to think that evidence responsiveness is constitutive of belief while this does not warrant thinking that beliefs invariably track what is perceived as evidence. Moreover, it explains most of the data and the intuitions that Rowland relies upon. Of course, we disagree that the nature of belief rules out akratic combinations of doxastic attitudes and that a belief changes immediately in response to changes in what is perceived as evidence. The human psyche, we believe, is much more complicated. Given the dispositional nature of belief (and other attitudes), it seems quite possible for a belief to remain fixed despite changes in what one perceives as evidence against the belief. Not only does this explain why the claim Rowland’s argument requires is false, it hopefully also provides the groundwork for a more adequate dispositional view of belief.

Let us finally, and briefly, return to consider moral judgments and intransigence. Let us suppose that moral judgments are beliefs. If the argument above is right, it is possible for a belief to remain fixed despite changes in what is perceived as evidence. Moreover, given some of the ideas advanced above, it should not be very surprising that moral beliefs (often) are intransigent. A moral belief differs in important respects from most beliefs, e.g., the belief that there is coffee made: our moral beliefs are important to us. They are beliefs that are central to who we are. They are beliefs that some people are willing to die for. It therefore should not be very surprising that they are beliefs that, when threatened, elicit various psychological defense mechanisms that function like a support structure. Sometimes these defense mechanisms may interfere with the normal causal chain leading from a change in what is perceived as evidence to a change in belief. By contrast, the belief that coffee is made is not of any importance and will not set off any defense mechanisms, but immediately change in response to changes in what is perceived as evidence. Consequently, given a dispositionalist view of beliefs along the lines advanced above, the cognitivist can explain why it seems plausible to think that evidence responsiveness is constitutive of belief while arguing that moral peer intransigence is not a problem for cognitivism.

5 Concluding Remarks

Although we have argued that Rowland’s argument against cognitivism fails, we nevertheless think that the argument highlights something important that has largely gone unnoticed in the metaethical debate that is worth emphasizing. Cognitivists argue that moral judgments are beliefs, but the nature of belief is seldom explained – it rarely goes beyond saying that beliefs have a mind-to-world direction of fit. We think that a dispositional view of beliefs makes sense of this metaphor (see also Smith 1994) and that a disposition to change in response to changes in what is perceived as evidence is the key to understand the nature of belief. Moreover, as we have argued, although evidence responsiveness (in the pertinent sense) is constitutive of belief, it is not inconceivable that a belief remains fixed despite changes in what is perceived as evidence. If a belief is constituted by, in part, a disposition to change in response to changes in evidence, this disposition may be interfered with in different ways.

It is nonetheless not obvious that there is a perfect match between moral judgments and beliefs. First, we need to more thoroughly investigate the nature of belief and the connection to evidence responsiveness. In particular, cognitivists need to be more explicit about how they think about beliefs. Second, if evidence responsiveness is constitutive of belief, which seems plausible, we need to more thoroughly examine if moral judgments behave like beliefs in the relevant respects. What Rowland would have had to show is that it would be possible to remain intransigent even if there is no background story that explains why the disposition to change in response to changes in evidence is interfered with. Again, it is not enough to merely consider the actual patterns of reactions, but we must also speculate about counterfactual patterns of reactions. If moral judgments are beliefs, and the explanation of the intransigence has to do with the importance of the belief to the agent, we have to speculate about what would happen if the belief would lose its importance. If it is conceivable that an agent maintains his or her moral judgment even in the absence of some psychological background story, it seems that moral judgments are intransigent in the sense that beliefs are not: moral judgments would be intransigent in a much more radical sense that challenge the cognitivist’s identification of moral judgment with belief. Rowland’s article is a nice start to investigating the nature of moral judgments, but it is merely the first step towards what we think is a new and interesting way to examine the nature of moral judgments and the relative merits of cognitivism and non-cognitivism.¹⁷