Socio-epistemic reasons for moral compromises

Socio-epistemic Grounds for Moral Compromises Socio‐epistemic grounds for moral compromises1 by Klemens Kappel Second Draft, September 2014. Any comment welcome 1. Epistemic grounds for moral compromises An influential view in political philosophy assumes that when we have deep and persistent disagreements about complex moral issues and these affect the common policies that we might adopt, we should at least in some cases be more willing to enter a compromise. My 2 aim is to explore this familiar idea in more detail. When does moral disagreement provide this sort socio-epistemic reason for moral compromise, and when not? As a part of the discussion, I will discuss the claims recently made by Simon C. May that we never epistemic grounds for compromise (May 2005). I argue that while May's claims are plausible, they only apply to a strong notion of compromise. We should, however, also be interested in weaker notions that are not affected by May's argument. It might be helpful to fix the topic for discussion. Suppose that Adele and Betty disagree about a particular moral issue, and therefore also disagree about which policy to adopt in a particular area. On their respective moral grounds, Adele prefers policy X, and Betty prefers policy Z. These two policies are 1 The paper was presented at a workshop on moral compromise held at University of Copenhagen, August 18-9, 2014. Thanks to Simon C. May, Daniel Weinstock, Fabian Wendt, Xavier Landes, Karin Joench-Clausen, Christian Rostbøll and others present for stimulating discussion. Special thanks to Martin Marchman Andersen for conversations about these issues. 2 References .... 1 Socio-epistemic Grounds for Moral Compromises incompatible and seem far removed from one another. But Y is a compromise position that is somehow located between X and Z in the space of possibilities. Adele and Betty have discussed and exchanged arguments extensively, and that they are entirely familiar with each other's position. Both have now formed stable views that are not about to change. All possibilities of including new policy options on the plate, or of devising policy options that meet more desiderata from both parties, have been explored to no avail. The 3 question is if there are reasons for Adele and Betty to prefer some compromise position Y? Surely, Adele and Betty may have instrumental reasons to compromise, say because they both want an ongoing collaboration and want to preserve peace and stability. In many cases instrumental concerns would provide a very strong or even overriding reason for accepting a compromise. However, like (May 2005) and others, I want to set pragmatic or instrumental reasons aside, and consider just principled compromises, that is, compromises that are not motivated by instrumental concerns. As indicated, my main focus will be whether Adele and Betty could have epistemic grounds for compromise. By epistemic grounds for compromise, I loosely have in mind something like this: the fact that Adele and Betty disagree, and know this, should somehow affect their confidence in their pre-compromise view and this should dispose them to find a compromise more attractive. Later I will try to specify various epistemic grounds for compromises in more detail, and I will distinguish different types of epistemic grounds for various types of compromises. 3 See Weinstock on the possibility of what he calls integrative and substitutive compromises (Weinstock 2013). Here and throughout I consider reasons for compromises relative to a fixed set of option. So if integrative or substitutive compromises are available, they have already been included on the list of policy options. 2 Socio-epistemic Grounds for Moral Compromises 2. Strong and weak notions of principled moral compromises First, we need to consider the very notion of a principled moral compromise. Here is how May describes the notion of a pragmatic reason to compromise: 'Consider Jane who endorses a position that is, in actual fact, the best balancing of values and interests on an issue (although perhaps only after a long period of philosophical reflection, practical experience, and political deliberation). Assume that the issue remains the subject of intractable disagreement in her society. This fact of moral disagreement does not give rise to any reason for moral correction, since we are assuming that Jane already endorses the best position on the issue. But if the distribution of beliefs and power is such that Jane is unlikely to realize any of her goals without some modification of her position then she has a pragmatic reason to accept a moral compromise.' (May 2005, 319). By contrast here is how May characterized the notion of a principled compromise: 'even if Jane were in a position to successfully implement her political position in its entirety, say after leading her political party to a decisive electoral victory, some think that she may still have principled reason for moral compromise with the dissenting minority independent of any pragmatic consideration of the future political fallout.' (May 2005, 321) I want to suggest that this notion of a principled compromise that is too narrow to cover everything we might be interested in regarding moral compromises. Generalizing a bit, the notion of a principled 3 Socio-epistemic Grounds for Moral Compromises compromise suggested by the above seems to be the fixed by the following: Adele and Betty need to make a choice between policy options (X, Y, Z). Adele prefers X, and Betty and prefers Z. Y is a compromise position. Adele has in fact (and what she takes to be) the all things considered best view on the issue. A fully legitimate political process has given Adele policy X. Betty fully acknowledges the legitimacy of the procedure, and accepts to abide by X, even though she thinks that Z would be a better or more just policy. So, policy X will prevail. Or at least, X will prevail unless Adele decides that she wants to concede Y as compromise to Betty. Adele is in no way forced to do so, and no instrumental reasons favor her doing so. Adele has a principled reason to compromise if and only if Adele has a reason to offer Y as compromise policy to Betty in these circumstances. Call this the strong notion of principled reason to compromise. According to the strong notion, for there to be a principled reason for compromise it has to be the case that even when some policy is thought to be the all things considered right option, there should nevertheless be a non-instrumental reason to prefer an inferior option as a compromise. The obvious question is how there can be a noninstrumental reason to prefer something other than the all things considered morally best? Clearly, when Adele has gone through all relevant considerations, she has also duly considered facts such as her disagreement with Betty, and she has taken this into account in whatever way this should be done. How could Adele then think that yet an additional consideration should sway her to suggest a policy Y, which she deems inferior to the best policy X? So, we soon arrive at the result that there can never be principled reasons for 4 Socio-epistemic Grounds for Moral Compromises compromise. But this is because this strong notion of a having a noninstrumental reason for a compromise verges on the incoherent. We are to imagine having a non-instrumental reason for doing something that reason indicates is not all things considered best. 4 However, I suggest that this results for adopting a too narrow way of looking at the issues. I want instead to propose another way of thinking of principled reasons for compromises that is initially more felicitous to their coherency. We can refer to this as the weak notion of principled reasons for compromise: Adele and Betty are about to make a policy choice between the options on the table. Adele is inclined to thinking that X is the best policy, and Betty to thinking that Z is the best. There is no commonly recognized legitimate decision procedure to settle the question (or there might in principle exist such procedures, but Adele and Betty do not have a common understanding or agreement of what they are). Either of them could try to win out in the political battle. Or they could try to expand the conflict beyond ordinary politics, or threaten to do so, in order to make their preferred policy prevail. Or they could seek a compromise. Surely, there are strong and even overriding instrumental reasons for going for a compromise. But are there also noninstrumental reasons for Adele and Betty to prefer a compromise position Y? Could it be a non-instrumental good-making feature of policy Y that it constitutes a compromise between disagreeing agents in such circumstances? 4 I am not saying that it is a conceptual truth that there cannot be principled reasons for compromises on the strong notion. All I am saying is that there is some conceptual tension here. 5 Socio-epistemic Grounds for Moral Compromises This way of setting up the issue raises the following broader question: when Adele considers the options, and initially prefers X, could she end op thinking that Y is the all things considered best option, where what makes Y all things considered best is in part noninstrumental good-making features of Y being a compromise? Or even weaker: could Adele think that some position is in one noninstrumental sense better in virtue of being a compromise? Are there non-instrumental reasons that sometimes favor compromises as such, independently of the good consequences that agreeing on a compromise may generate? Obviously, this way of characterizing principled reason for compromise is not in all respects precise, depending as it does not a distinction between instrumental/non-instrumental good making features of a policy option, or the similar distinction between instrumental and non-instrumental reasons for preferring a policy option. But I hope that it is clear enough for the arguments below to be developed. One might object that what I have specified now is really a situation in which Adele might have a principled reason to accept a compromise, rather than a case in which Adele could take her disagreement with Betty as a reason for revising her view about what is all things considered best. So, in May's apt terminology, the disagreement is a cause for correction, not for compromise. A compromise requires a residual of disagreement, it requires accepting a policy option that one continues to find morally inferior to one's preferred alternative. It is entirely correct, I believe, that this is one way of talking about compromise, and one that may underlie the strong notion of a principled reason for compromise. However, I am not sure that the word 'compromise' has such a precise meaning that we can say that the strong but not the weak notion is correct. In this end this would be a terminological question, I believe. It is a matter of what stipulation of the word 'compromise' we chose to work with in philosophy. And no matter what we decide here, the following 6 Socio-epistemic Grounds for Moral Compromises question that should be of general interest: When we face recalcitrant disagreements about what is morally best, how should this affect what we end up thinking is the all things considered best policy to pursue? It seems natural to ask this question in terms of compromises, and this requires that we use the weak notion of compromise, not the strong notion. Henceforth, when I talk about compromises, I will have the weak notion of compromises in mind, unless otherwise indicated. Some putative principled reasons for compromises are purely moral. They recommend compromise for pure moral reasons, say because adopting compromise positions better accommodates moral aims such as inclusion, respect, reciprocity, civic friendship, or desiderata of procedural fairness or legitimacy (see May 2005 and May 2013, for lucid discussions). I will comment on purely moral grounds for compromise below, but most of the paper will be concerned with epistemic grounds for compromise. How should epistemic features of disagreement affect what we should think all things considered best policy option to pursue? 3. What is a compromise? Before proceeding, I would like to add some further remarks about the structure of moral compromises. We have been assuming, quite schematically, that X and Z are policy options placed at extreme ends of a spectrum, and that Y is a compromise position somewhere between X and Z. Note that Y and Z are policy options, not moral views. I assume throughout that preferences for policy options are motivated by more basic moral concerns. This is why it is appropriate to talk about moral compromises, though what we compromise about are policy options, not our basic moral commitments. We have tacitly been assuming that our two agents involved in compromising have the following ranking of the options: Adele X>Y>Z 7 Socio-epistemic Grounds for Moral Compromises Betty Z>Y>X Now, this ordering of extremes and compromises provides an intuitively plausible constraint on what counts as compromises. 5 However, there are a couple of things to be noted. In many realistic cases there will be indefinitely many positions that qualify as compromise positions. Consider a disagreement over progressive taxation. Some argue that there should be a steep progressive taxation, say 80% for very high incomes, and their basic motivation is moral (after all, those who earn a lot do so in great part because of luck). Others are in favor of a regressive taxation, so that high earners effectively pay a lesser fraction of their income in tax (say because of views about selfownership and entitlement to whatever your talents and circumstances can earn you). Obviously, there are indefinitely many positions in between these two extremes, including some that are arbitrarily close to the extremes. Note two implications of this. For one think, it is hard to imagine that there could be a non-instrumental reason for a compromise that identifies one particular compromise position. Even if there is a compelling reason for accepting some compromise, such a reason might be entirely unsuited for identifying one particular compromise rather than another. For second, we clearly need the notion of a fair compromise, where this is not simply a compromise between two positions that are both reasonable, but also a compromise that is in some sense located in a fair interval between two extremes. As compromise positions can be arbitrarily close to the extremes, not any compromise is a fair compromise. However, when we seek compromises, we seek fair compromises, not just positions that in a mathematical sense constitute a compromise. But what could provide a metric for fairness of compromises? Recall that the disagreeing parties initially disagree 5 See the more elaborate definition in .... 8 Socio-epistemic Grounds for Moral Compromises about the underlying moral commitments. This is why they are attracted to extreme positions, so presumably they cannot agree to use any of these moral commitments as a metric of fair compromises. What one party to the compromise may think of as a major sacrifice incurred to forge a compromise may not seem so for the other. Should we measure the distance in terms of how psychologically hard it is for people to accept a compromise? A compromise would then be located in the fair interval if and only if the two parties are about equally psychologically disturbed by adopting the compromise. This seems hardly plausible for a number of reasons. Such a procedure would be vulnerable to a type of expensive preferences problem. And think of three-person cases, where one individual is already located in what is in the fair spectrum for the two others, so this individual does not have to sacrifice anything. Why would that be fair? We cannot solve this problem here. The point to note is that even there are compelling reasons for compromise, they may not enable us to determine fair compromises. In the examples above we assumed that compromises can be ordered on a single scale where the two extreme positions are also represented. Call this a scalar compromise. Other cases of compromise positions that come to mind don't obviously permit representation on a scale. Consider the problem of abortion, involving proponents of pro-life polices on the one hand, and pro-choice on the other. One type of compromise position could aim a limiting the number of abortions performed - such compromise positions can be represented on a scale. Another could impose various limits at the stage in pregnancy at which the abortion can be legally performed. This could be represented on a time scale. Again, this is a scale, but an entirely different one from a scale representing numbers of abortions performed. A very different type of compromises could restrict the cost of abortion procedures to those who are in favor of pro-choice policies. This again, would be a sort of compromise between pro- 9 Socio-epistemic Grounds for Moral Compromises choice and pro-life positions, but it is not a compromise that easily lends itself to representation on a scale. 6 This last form of compromise is a version of what one might call a liberal compromise. As is familiar, when we cannot agree on a contested issue, we can sometimes devise a policy that seeks to permits everyone affected to live in accordance with what they themselves consider the correct moral view. For another illustration, think of the controversy regarding the use of genetically modified crops. Roughly, some are in favor of a very restrictive legislation or even a ban, making it very costly or impossible to market such products. Others disagree and see no particular problem in the use of genetically modified crop. There is a well-known middle position that recommends labeling GM-products. In my terms, this would qualify as a liberal compromise. It is easy to see that there are liberal compromises to range of moral conflicts - think of same sex marriages, physician-assisted death, or vegetarianism. When a liberal compromise exists it may seem to be a much better solution to a conflict than any other type of compromise one might think of - at least this would be how it appears to the liberal minded. So, for the liberal, a liberal compromise might not even count as a compromise position, but simply as the correct noncompromise position. For those opposed, a liberal compromise may not seem a compromise at all, but as an extreme position. The important point here, however, is that there might be principled reasons for liberal solutions that makes them count as compromises that I have adopted, in so far as they are perceived to be superior in part because they are a non-instrumental response to the fact that we disagree about underlying moral commitments. This obviously increases the complexity of the problem of identifying reasons for 6 Some of the more realistic ones are discussed by May; here I am just interested in illustrating a formal point 10 Socio-epistemic Grounds for Moral Compromises preferring one particular compromise rather than another, and for saying what a fair compromise is. 7 Whatever we want to say about this latter problem, the fact remains that many moral conflicts and policy disagreements do not lend themselves to liberal compromises in the sense I have in mind: think of death penalty, climate policy, gun control, taxation, freedom of speech, and immigration. These are policy areas in which we cannot get very far by liberal compromises (which is not to say that there then necessarily exists a scalar compromise or any other sort of compromise in these cases). We will later discuss different types of grounds that there may be for scalar and liberal compromises. The important things to have in mind are these. First, there is often a wide spectrum of compromise positions, and often no obvious way in which these possibilities can be ordered, or in which the space of fair compromises can be determined. Second, there are at least two rather different types of compromises, scalar compromises and liberal compromises. As we will see, the grounds for adopting scalar 8 compromises and liberal compromises can be very different in ways that matter for our discussion of epistemic grounds for compromises. 4. May on epistemic grounds for moral compromise In this section I discuss May's reasons for rejecting that there are epistemic grounds for compromises (May 2005, p. 338-340). I then propose a different way of focusing on the epistemic grounds for compromise which utilizes the weak notion of moral compromise. Assume that Jane, a proponent of a pro-choice policy, in fact endorses the all things considered best policy, which moreover is 7 These classes of compromises - liberal and scalar - could be combined in various ways, making the problem of determining the fairness of compromises even more difficult. 8 See also (Weinstock 2013, 539ff) on various other types of compromises. 11 Socio-epistemic Grounds for Moral Compromises about to be adopted due to a legitimate decision procedure (May 2005, 338). May asks if Jane nonetheless would have some reason to propose a compromise because of the complexity and the controversy surrounding the abortion question? As May says: 'Principled compromise may be the best way to acknowledge our fallibility and the limitations of our ability to discern moral truth' (May 2005, 339). May presses two reasons why Jane's recognition of the complexity of the dispute, and of her own fallibility, is not a proper ground for compromise. First, if acknowledging the complexity of the issues and the fact of disagreement should make Jane less certain about her own position, it should make her equally uncertain about any proposed compromise position. 'A policy that splits the moral difference between two opposing yet reasonable viewpoints need be no more self-evident than the viewpoints themselves' (May 2005, 339). So, compromises will not be epistemically better than precompromise positions, and in virtue of being compromises they will inevitably seem morally worse. Actually, a somewhat stronger conclusion would seem warranted. Suppose that Jane has already adjusted her confidence in her view as a response to the disagreement that she knows surrounds the abortion question. Assume that Jane is still rather confident that her view is correct, though she is now somewhat less confident than she would have been, were it not for the disagreement. Then Jane would seem to have a reason to trust any compromise position less than her current position. While she should surely recognize that her current position might be wrong, she should suspect that compromise positions are more likely wrong. And they are in no respect morally better. So, compromise positions would seem worse in both epistemic and moral respects. Second, May points out that the complexity of the issues and the recalcitrant disagreement, and Jane's resultant reduced confidence in her view, do not constitute first order reasons that Jane's view is wrong, or needs correction. A first-order reason is a 12 Socio-epistemic Grounds for Moral Compromises moral reason to think that some part of a moral view is wrong. A commitment to the value of autonomy is a first-order reason to think that the rejection of the moral permissibility of physician-assisted death on request is a mistake. But Jane's acknowledgement of her fallibility as a moral thinker is not a first-order reason to think that her view needs correction. It may, to be sure, be a reason for doubting her own view more than she would do if did not consider herself fallible. But, as May remarks, 'this doubt is normatively inert', (May 2005, 339). In another paper, May mentions an additional reason to question whether disagreement as such (as distinct from mere complexity) is a reason for compromise: 'A difficulty with this approach [that disagreement is a reason for compromise] is to explain why the need for epistemic modesty is conditional on the presence of disagreement and not simply illustrated by it. For instance, if uncertainty about an empirical assumption warrants some modification in a party's position, it should arguably do so whether or not the parties disagree' (May 2013, 6). As I will discuss more fully in the next section, I do not think that this is plausible. Actual disagreement, as distinct from mere complexity, does make a difference for the credence we should assign to basic moral commitments. The upshot is that if Jane has fully taken the complexity and the disagreement into account, and has adjusted her views accordingly, then there are no further epistemic grounds for compromise. This, it seems to me, is exactly what we get if we focus on the strong notion of a principled reason for compromise. According to this notion, there is a principled reason for compromise only if there is a non-instrumental reason for deviating from the all things considered best option. Acknowledged disagreement is not such a reason: when we have taken disagreement into account in what we take to be the best option, continued disagreement does not constitute yet a reason to prefer an inferior option. 13 Socio-epistemic Grounds for Moral Compromises While this is correct, it obscures the question of how epistemic features of disagreement should affect what we ought to think is the best option all things considered. This is the question I want to focus on here. Do compromise positions somehow stand out as better in non-instrumental respect on epistemic grounds because we initially disagree moral views and preferred policy options? In order to focus on this problem, I want to address two questions in the sections that follow. The first question is whether moral disagreement is a reason to be less confident in one's moral views. The second is whether reduced confidence in one's moral views as a result of a disagreement is a reason to find a compromise position more attractive than it would otherwise be? This question assumes the weak notion of compromise. I suggest that affirmative answers to these two questions have been operative in the literature, though the details of the epistemic grounds for compromise have never been stated explicitly. I want to explore some of the complexities of both questions. 5. Is moral disagreement a reason to reduce confidence? Assume that an agent can assign levels of credence to moral propositions between 0 and 1, where 1 signifies full confidence in the truth of the proposition, and 0 full confidence in the falsity of the proposition. A confidence level of 0.5 is somewhere in between, and indicates that one is as confident in the truth of the proposition as in the falsity. There are issues about how to interpret this, and I will discuss some other options below, but for now we can leave these complications aside. 9 The question before us is this. Prior to engaging in her disagreement with Betty, Adele is very confident that her position on some moral issue is correct. Now she reflects on the fact that Betty, who appears to be as thoughtful and sincere as she herself is, and 9 For a brief explanation and more references, see (Goldman 1999). 14 Socio-epistemic Grounds for Moral Compromises who is fully acquainted with all the arguments and the relevant evidence, sharply disagrees. Should this affect Adele's confidence in her view? The answer to this question crucially depends, I think, on details of the structure of the disagreement between Adele and Betty. What I say here is in great part motivated by David Christensen's recent work on disagreement and higher order evidence, see (Christensen 2007; Christensen 2009; Christensen 2011).A full view of the epistemology of moral disagreement cannot be elaborated here, but I want to suggest a couple of things to be used later in our discussion. First distinguish between what we might call local disagreement on the one hand, and comprehensive disagreement on the other. Adele and Betty have a local disagreement when their basic disagreement is about one single moral proposition. Suppose that Adele has a particular strongly felt moral intuition, say that criminal offenders deserve punishment. Betty has an equally strong felt moral intuition that criminals do not per se deserve punishment. Other differences between Adele and Betty are either irrelevant for this issue, or they are direct consequences of their conflicting moral intuitions. Suppose that Adele and Betty now realize that this moral intuition the only thing that separates them. What should they do? In my view, they should both be less confident about the veracity of their respective intuitions. Consider the situation as Adele sees it. She knows that she has a particular moral intuition, the origin of which she knows very little about. She also knows that Betty, who is in many respects just like her, has an incompatible intuition. Adele has no particular higher order reason to believe that her intuition is correct, and Betty's wrong. So, the natural reaction for her is to reduce her confidence that the intuition reflects a true moral 15 Socio-epistemic Grounds for Moral Compromises proposition. She might still believe the propositional content of the 10 intuition, but now with lesser confidence. Betty's situation is, of course, similar. Most actual moral disagreements are not local in this sense - rather they involve larger complexes of considered intuitions, reflections about ethical principles of various levels of generality, counter-examples, explanations of why some intuitions might be misleading, reflections of theoretical virtues of coherent views and so on. Comprehensive disagreements engage all the familiar features involved in reflective equilibrium. Obviously, there is a spectrum of cases from very local to very comprehensive disagreements, but here I can only consider the extreme cases. Suppose that Adele has a fully articulated comprehensive moral disagreement with Betty. By this I mean that Adele has made clear to herself what fundamental normative assumptions her position depends upon, and which arguments (if any) or basic moral intuitions (if any) they depend upon, she is fully aware of the possible counter-examples or counter-arguments to these normative assumptions, and what the counter-arguments to the counterarguments are. So, for any putative first order moral reason that her view is wrong that Adele knows of, she also knows how to deflect or rebut that reason. So, Adele has, we can imagine, a philosophically impeccable moral view. Let us imagine that Betty is in the same position as regards her disagreement with Adele, and that they both know about the symmetry. 10 11 In recent papers, Wedgwood argues to the contrary (Wedgwood 2007; Wedgwood 2010). I think that his arguments fail, however. 11 The case May have in mind, involving Jane who after careful philosophical reflection of the problem of abortion, including the pro-life arguments, endorses a sophisticated pro-choice position, could well be an instances of what I call a comprehensive disagreement. 16 Socio-epistemic Grounds for Moral Compromises Now, what should Adele do? Should she reduce her confidence in her view? After all, she knows that Betty is as thoughtful and reflective as herself, and she knows that Betty can defend her position just as well as Adele can defend her own. Here is what one might suggest as a reason for thinking that Adele should not reduce her confidence. As I have described the case, Adele is not confronted with undefeated first order reasons to think that the moral position she espouses is wrong or in need of adjustment. Adele is, to be sure, aware of many first order reasons that suggests revision, but she knows how to respond to them. I want to propose that we should nonetheless reduce our confidence in such cases, at least sometimes. The reason is that we do not understand well enough how we employ the method of reflective equilibrium (or any other method used to generate moral belief). Basically, we do not understand where our basic moral intuitions come from, or how we should trade off the aim of preserving those intuitions with concerns for simplicity and power and other theoretical commitments, that often suggests revising untutored intuitions. Moral philosophers develop highly sophisticated views of the way theoretical positions in ethical theory and political philosophy are supported by certain basic intuitions, by various arguments and fundamental principles, how they are best accounted for by positing moral principles of varying degree of complexity, how these may be subjected to counter-examples and counter-arguments, and how one might respond to such difficulties. But what we do not understand very well is how the balance of evidence provided by all of this adds up. 12 So, we should not, I suggest, be too confident that we are right in complex moral questions where one sophisticated 12 For an elaborate skeptical view about the possibility of being steadfast in philosophical disagreements, see (Frances 2013). For skepticial remarks about the meta-justification of reflective equilibrium, see .... 17 Socio-epistemic Grounds for Moral Compromises comprehensive view confronts another, even when these disagreements are fully articulated at the first-order level. This is not because we don't understand the dialectic involved, but because we don't have a firm grip on how it all adds up. So, in these cases we should reason as follows. Since we disagree, at least one of us got it wrong. At least one of us weighed the evidence in the wrong way, placed a too high premium of simplicity over conservation of intuition, or fed the wrong set of moral intuitions into the machinery to begin with. However, none of us is in a position to say with any confidence that we did so in the right way, and that the other is wrong. We should therefore reduce our confidence in our view. 13 14 13 Note that there also seems to be a phenomenon such as contrastive credence. I may be very confident that P rather than Q, but not that P1 rather than P2, where P1 and P2 are moral principles that are rather similar in terms of content and justification. Think for example of the contrast between two familiar forms of egalitarianism, prioritarianism and egalitarianism proper. This suggest that we should sometimes reduce contrastive confidence between closely related moral principles, and be willing to compromise regarding policies that depend on them, but not between principles far apart, though our epistemic credentials in a way remain the same. 14 In discussion May suggested that the proposed view has the implausible implication that when we disagree we have no reason to go on trying to persuade one another. I don't think that this is a prolem. First, on the view proposed if one should reduce confidence as a response to disagreements, one should do so only when one is convinced that the disagreement is of a certain qualified sort. One can reasonably believe this only after prolonged discussion. Second, in real life disagreements, new evidence often comes in, and this may unsettle the balance, so to speak. 18 Socio-epistemic Grounds for Moral Compromises Other cases are different, however. Sometimes we have good reasons to think that the source from where our opponent derives his or her moral convictions is unlikely to be reliable. Suppose that our opponent derive them from a book, or an authority the reliability we have reason to question, or if their moral sensibilities seem abnormally limited, say because they are obsessed with a particular moral principle, and they want to derive everything from that principle. In these cases (that are obviously also common) we might be justified in attributing less (or no) weight to the fact that we disagree. The fact of disagreement weighs less; it provides less of a reason to think that we are mistaken in our own view. Recall that May suggests in passing that is it is not the disagreement as such rather than the complexity of the issues which should make the parties 'acknowledge their fallibility regarding both empirical and moral questions.' (May 2013, p. 6). What the above suggests is that the fact of disagreement matters, not the mere complexity of the issues. It is the actual disagreement that should alert us; the actuality of error, not just the possibility of one. If issues are highly complex, but many thinkers independently of one another, and from different backgrounds nonetheless converge on the same view this should make us increase our confidence, despite the acknowledged complexity of the issues. 6. Is reduced confidence a reason for compromising? Assume that local as well as comprehensive moral disagreements are sometimes reasons to reduce one's confidence in moral commitments. Is this is a reason for compromise? When one, as a result of deliberating about a moral disagreement, becomes less confident that one is right, should one become more favorably inclined towards one or more policy options that are compromise positions relatively to one's position before one deliberated about the disagreement? I want to argue that in a range of cases this is not so. But there are other types of cases in which compromises appear very 19 Socio-epistemic Grounds for Moral Compromises attractive on epistemic grounds. Moreover, if we grant certain further assumptions that I will identify later, there might be fairly large class of cases. It will be useful to work with a more elaborate case in order to better exemplify possible epistemic grounds for compromises. My focus is the structure of epistemic grounds for compromises, not the plausibility of these views themselves. With this in mind assume that Adele believes (C) The autonomy of women is a weighty moral concern, as are the negative social consequences of a restrictive access to abortion. Adele also holds (P) The early has fetus has no independent moral standing (because it does not have sentience, or preferences regarding it's future or present, and is not numerically identical to any later individual). Consequently other moral concerns can easily outweigh moral concern for the early fetus. Adele believes that C and P strongly support policy X, which is prochoice policy. Betty's view is different. While she fully agrees with C, she rejects P and accepts (Q) The early fetus enjoys full moral standing, at least as concerns termination of its existence right from conception. This moral factor ultimately outweighs concern for women's autonomy, as well and concern for negative social consequences of a more restrictive access to abortion. 20 Socio-epistemic Grounds for Moral Compromises Betty believes that taken together C and Q strongly supports policy Z, which is a pro-life policy. Note some features about the case. First, Adele and Bettye share a set of moral commitments, though there are also steep disagreements. Second, Adele and Betty disagree about contrary moral propositions. P implies not-Q, and Q implies not-P. So both views cannot be true, though they could both be false. I suggest that both features are common in value-dependent policy disagreements. At any rate, both features play an important role the discussion below. Suppose that prior to engaging in the disagreement Adele is very confident about P, and Betty is just as confident about Q. As a result, they are both very confident in their preferred policies, X and Z. Assume that as a result of learning of their disagreement, Adele and Betty are both rationally adjusting their confidences in P and Q. Should Adele and Betty now be more inclined towards some compromise policy Y? Consider first three types of cases in which reduced confidence in moral commitments does not seem to be a ground for compromises ((a), (b) and (c)): (a) Reduced but still high confidence After reflecting on the fact that they persistently disagree about P and Q, Adele and Betty still rationally believe that their respective view is likely to be correct, although they are now less confident than the used to be. Assume, for example, that they now both assign credence 0.6 to their respective moral views P and Q, whereas prior to thinking about their disagreement they assigned these views credence 0.8. It is difficult to see why Adele or Betty should now view a compromise position Y as attractive in any way. Although they have reduced their confidence in their own position, they should have still less confidence in any position different from their pre-compromise position. 21 Socio-epistemic Grounds for Moral Compromises (b) Maximally reduced, but asymmetrical confidence After deliberating and taking into account the fact that they disagree, Adele and Betty both ends up assigning only 0.5 credence to their own moral commitments. Should they now compromise? Even this need not be the case. Suppose that while each of them have lost confidence in their own view, they still think it is overwhelmingly likely that the opponent's view is false. So, Adele might be impressed by her disagreement with Betty, and reduce her confidence in her own view to the lowest possible short of being more inclined to simply rejecting it. Yet, Adele might still believe that Betty's view is very likely to be wrong. This is possible on the assumption noted above that P and Q are contrary propositions, not contradictory. How can Adele become less convinced that her own view is correct, while not being more inclined to think that Betty is right? One answer is that what may make Adele rationally reduce her confidence in her own view need not be that she think that Betty's view is likely to be correct. Rather, it could be that she notes that Betty, who seem rational and willing to consider the possibilities carefully does not seem to see the force of Adele's view. This may make Adele wonder if she has somehow made a mistake, even if she still thinks that Betty's own view is quite implausible. The case might be symmetrical, of course; Betty could also remain convinced that Adele's view is false, while reducing confidence in her own view. In this type of case it is difficult to see why Adele should have epistemic ground for compromise between her own and Betty's policy positions. This is so in particular if we consider scalar compromises, where the compromise position is somehow located in the space between the extreme positions. Why should Adele be inclined towards a compromise between her own policy position that she barely trusts, and Betty's, which Adele is convinced to be wrong? (c) Symmetrical maximally reduced confidence Suppose then that Adele and Betty both reduce their confidence to 0.5, and that they both assign that same level of credence to the views 22 Socio-epistemic Grounds for Moral Compromises of their opponent. So, Adele and Betty assign credence 0.5 to both P 15 and Q. Should Adele and Betty now be more inclined towards a possible compromise between policy X and Z? Assume first that Y is a scalar compromise between X and Z. It is not clear, at least in the abstract, why Y would appear attractive. Why should Adele and Betty prefer Y to X or Z? Recall that Adele and Betty now no longer have any particular confidence in their original moral views P and Q. In so far as the extreme policy positions X and Z are in great part motivated by P and Q, why would they now be inclined to Y, assuming that Y is a scalar compromise between X and Z? Why should the compromise position between these two extremes seem attractive, or even more attractive, than their pre-compromise views? It is hard to see. So, so even if a scalar compromise between X and Z exists, it is not clear why it would seem attractive for Adele and Betty. It would seem much more appealing for Adele and Betty to set aside both P and Q and try do decide the matter on other grounds, in particular on the ground of C, which is a moral commitment they share. This could in principle land them in a scalar compromise, but as I shall now suggest, it is might be more natural to view these types of cases as grounds for liberal compromises. Let us now look more closely at this possibility. (d) Reduced confidence and common ground: a case for liberal compromise. Assume that Adele and Betty both assign credence 0.5 to P and Q respectively, whereas they still agree on C. When deliberating about policy options, why not suspend judgment about P and Q entirely? Why not take P and Q out of the equation, so to speak, and discuss what policy is best on grounds of the normative assumptions they 15 Adele and Betty cannot coherently assign credences higher than 0.5 to each other's views in this symmetrical way, since we have assumed their views to be contraries. 23 Socio-epistemic Grounds for Moral Compromises share, namely C? Suppose they agree on a policy motivated in this way, or that they each now prefer a different policy than they did before. Note that compromises motivated in this way would have a significant epistemic motivation. Adele and Betty maximally reduce their confidence in P and Q and decide for a compromise policy in the light of the moral commitments that they are still confident about. This implies that Adele and Betty must consider the compromise position epistemically well motivated compared to the pre-compromise positions X and Z, since they now consider the moral views P and Q to be barely epistemically qualified. Moreover, note that from Adele's and Betty's point of view, a compromise policy thus motivated will also seem morally superior to X and Z. After all, Adele and Betty now suspend trust in the moral views that initially led them to prefer X and Z respectively, so they would refrain from using P and Q in evaluating the compromise position. Earlier in the discussion we noted that compromise positions may seem both morally and epistemically inferior to precompromise positions. This view, however, was based on the strong notion of a compromise. Once we focus the weak notion of compromises, and consider compromises that are motivated by resorting to common ground, compromises may be both epistemically and morally better than pre-compromise position, judged from the perspective of the compromising agents. 16 Note that if there are pure moral reasons to compromise, these would compete against other reason in favor of non- 16 Sometimes, of course, compromises made from common ground are hopeless. This happens when the common ground is too slim, or too peripheral to our main concerns to permit sensible policy options to be chosen. 24 Socio-epistemic Grounds for Moral Compromises compromise views. But if our confidence in these other reasons 17 vanishes as a result of disagreement, then the pure moral reasons would assume greater relative weight. Indirectly, epistemic grounds matters for this to happen. Epistemic epistemic grounds for compromises might be inert by themselves (as we saw in (a), (b) and (c)), but nonetheless pave the way for pure moral reasons for compromise, assuming they exist. Before proceeding, I want to consider an objection. The epistemic ground for arises when we reduce our trust in our moral commitments as result of disagreement. We have seen that this ground is sufficient only when Adele and Betty both undertake a very drastic reduction of confidence in their disputed moral beliefs. However, it may seem that such massive reduction in credence is hardly ever warranted as a response to the sort of moral disagreement that we actually face in our society. Note for example that most people form quite stable moral views, and do not seem to feel a strong rational pressure to change their views, despite the controversies about these moral views they cannot fail to know about. Note also that we don't often blame one another for upholding our core moral commitments when they are known to be subject to long standing disagreement. If this objection is on the right track, the massive reduction in confidence required for compromise would be rationally required only in a small fraction of actual cases, if ever. In consequence, the epistemic grounds for compromises that I have discussed so far are either severely limited in scope, or they implausibly exaggerate the extent to which we should reduce confidence as a response to moral disagreement. I suggest here one immediate response to this worry, and then discuss to other much more far reaching possibilities in the 17 I am inclined to be skeptical about the view that there are pure moral reasons for compromise, mainly because of the arguments in May's papers. 25 Socio-epistemic Grounds for Moral Compromises next section. I want to suggest that when we intuitively judge it unlikely that we are often required to reduce our confidence when facing moral disagreement, there are two sources of bias that may influence our judgment. First, as philosophers, there is a risk that we tend to focus on cases where the epistemic work has been done. We are biased because as philosophers we are inclined to think about individuals who have reflected carefully about their moral commitments, and who are anyway not likely to be highly confident in moral beliefs that they know are subject to qualified controversy. In part this is because they have already revised their moral beliefs extensively, and they may also in many cases have also reduced their confidence in moral beliefs that are still controversial. Second, we might be inclined to factor in the psychological cost of reducing confidence in basic moral commitments. If such costs are high, a reduction in trust might be epistemically warranted, but not prudentially warranted. Again, this may influence our intuitive judgments about when such reduction is warranted. To avoid these biases, we would need to consider cases where people adopt strongly held without much deep reflection about the justification of these views, or what the epistemic implication of controversy may imply, and yet they know that they are subject to controversy. Such cases are easy to come by when one considers the actual history of socially prevalent moral views that are quite unreflective, and sustained without much concern for reasons and consistency. I suggest that it is not implausible that holders of such views ought to be much less confident than they actually are, or were. Also, we would need to consider cases where we assume that the psychological cost of reducing confidence in deeply felt moral beliefs are nonetheless low (such cases would have to be quite contrived, of course). I suggest that when we consider such cases, we would be less inclined to judge that one is not required to reduce confidence. 26 Socio-epistemic Grounds for Moral Compromises Anyway, I now want to set this aside, and consider two other ways in which epistemic grounds for compromises may become significant even with less drastic reductions of confidence than those we have considered so far ((e) and (f)). (e) The reduced weight principle Suppose that Adele and Betty are rationally compelled to reduce their confidence in P and Q somewhat, though not to the point of 0.5. Adele and Betty still have most confidence in their own view, though less than before they deliberated about their disagreement. As discussed above, in such a case there would seem to no epistemic ground for compromise; even with reduced confidence in their respective moral grounds, they need not change their views about what policy is best. But consider now two questions. First, should Adele and Betty attach less weight to their moral commitments, now that they have less confidence in them? Second, if so, could this be a reason for a compromise position becoming more attractive? The first question concerns whether we should accept: The reduced weight principle. When A assigns lesser credence to the truth of a moral principle, then A should assign lesser weight to the moral factors identified by that principle. 18 18 As stated, the reduced weight principle presupposes that a familiar form of externalism about moral reasons is false. According to this form of externalism, whatever moral reasons or factors exert the same force on us, whether we know about them or not. If we grant the truth of this or similar forms of externalism, the reduced weight principle would have to be restated in terms of subjective moral obligation, or subjective rightness, where subjective obligation or rightness is a function of what subjects reasonably believe. 27 Socio-epistemic Grounds for Moral Compromises Is the reduced weigh principle plausible? Some reflection suggests that it is, though a full discussion is beyond the scope of this paper. If one is fully confident in the truth of a moral principle, then one should surely assign full weight to the moral factors it identifies; not doing so would seem irrational. If one is fully confident that a principle is false, one should, of course, assign the factors identified no weight at all. But suppose one assigns a principle some credence higher than 0.5, though one is not fully confident that it is true, what should one do? It would be odd to assign it no weight, but also peculiar to assign it full weight as if one believed the principle fully. So, some reduction in weight would seem to be mandated when confidence in truth is reduced. When one is less confident in the validity of a purported moral reason or factor, then that reason or factor should have less traction in one's moral thinking. If the reduced weight principle is correct, we seem to have yet an epistemic ground for compromises, provided that disagreement is sometimes a reason to reduce confidence in one's moral commitments. When our moral commitments are subject to disagreement that should make us less confident in their truth, the moral reasons identified should assume less weight in the total array of reasons, even when we still on balance have more confidence that our moral commitments are correct than not. This will give more relative weight to undisputed moral commitments, and this will at least in some cases rationally incline us in the direction of compromise positions. (f) Reduction of credence, suspension of belief, and accepting to suspend belief Consider again the objection that it seems implausible to hold that we are often required to abandon deeply held moral convictions as a response to moral disagreement. As suggested, one thought that might move us here is consideration of the psychological burden that reducing confidence in moral beliefs may impose on people. Even 28 Socio-epistemic Grounds for Moral Compromises when required by epistemic rationality that one decrease confidence in one's moral commitments, we may be psychologically incapable of doing so, or unwilling to do so, because the psychological costs are too large in terms of emotional burdens and loss of the identityforming role that some identity-sustaining role that some moral beliefs sometimes have. If this is right, then moral beliefs sometimes do not respond well to evidence, including evidence that one might have made an error in adopting the belief, and this also holds when it is the fact of disagreement that constitute this evidence. This may simply the way that moral convictions work in normal human beings. Whatever the correct psychological account, assume that Adele is psychologically incapable of giving up certain of her moral commitment P, though she is involved in a highly qualified disagreement with Betty that actually warrants a drastic reduction of her confidence in P. Now, in this situation Adele may be in an epistemic position that might well rationally compel her to reduce confidence in P, or even give it up, though she is psychologically unable to do so. Adele's moral beliefs subsist because they respond not only to evidence, but also to the psychological costs of abandoning them. So, assume that Adele cannot give up her strong commitment to P without prohibitive cost. Even if this is so, I want to suggest that Adele can do something else. In her deliberation with Betty about what policy to adopt, Adele can accept what I will call a suspension of her judgment that P. In Cohen's sense, accepting a proposition is distinct from believing it, and it amounts to treating it as true for some practical purpose, irrespective of whether one believes it or not. Acceptance 19 of propositions is, arguably, common in science and in bureaucracies, where acceptances of propositions that one doesn't really 19 believe may be pragmatically necessary for various See (Cohen 1989) 29 Socio-epistemic Grounds for Moral Compromises collaborative projects. Now, clearly acceptance is a special sort of propositional attitude, and this attitude could also be applied suspensions, as it were. Just as one can accept a proposition, or accept it's negation, one can accept to set aside both a proposition and its negation for some practical purpose, and this is what I mean by accepting to suspend judgment about some proposition. So, even if Adele cannot bring herself to change her belief in P, she might nonetheless decide to accept a suspension of her belief in P. So, in her dealings with Betty about what policy to adopt, Adele could decide to disregard the disputed moral belief P, as well as the negation of P. She would act as if she didn't have that moral belief, nor a belief regarding its negation. Adele could, of course, expect that Betty would do the same with regard to Q, or Adele could make her own acceptance of suspension of belief about P conditional on Betty's acceptance of a suspension of belief about Q. Why should Adele do accept to suspend belief about P? Well, Adele could acknowledge that though she is psychologically incapable of suspending her first order moral belief in P, a significant reduction of credence in the proposition that P is nonetheless epistemically warranted. She might treat this as a reason for accepting a suspension, as a sort of surrogate of the revision in credence that is too costly for her to undertake. She could also cite reasons like the reduced weight principle for why accepting a suspension of P is warranted. If adopted by both Adele and Betty, acceptance of suspension would create room for liberal compromises as we were considering in (d), and maybe also other types of compromises. Again, such compromises would be partly epistemically grounded, motivated as they are by the epistemic effects of qualified disagreement. Of course, accepting a suspension of belief would reflect that Adele in some sense distrusts her own moral beliefs, or her capacity to properly revise moral confidence in her beliefs. Although Adele remains firm in her moral convictions, she nonetheless from a 30 Socio-epistemic Grounds for Moral Compromises higher order perspective takes notice of the reasons to worry about her level of confidence. Adele adopts a sort of higher order skeptical attitude towards her own capacity to correctly adjust her credence in first order moral beliefs. Obviously, this indicates some tension in Adele's moral outlook, but it does not strike me as an psychologically implausible or incoherent stance to take. Note here that acceptance of suspensions might be a more significant attitude for our willingness to compromise than low degrees of credence in one's moral beliefs, and more significant than suspension of belief. Note first that assigning a proposition a 0.5 credence is not the same as suspending belief in that proposition as well as it's negation. In some cases, is doesn't matter whether an agent assigns a moral proposition 0.5 credence, or simply has suspended belief about it. But there are contexts in which it makes a huge difference. To illustrate, suppose that Betty as a result of her disagreement with Adele becomes agnostic about the moral status of the fetus, that is suspends belief about Q. So, Adele has no beliefs about the matter; she neither believes that the fetus is worthy of moral concern for its own sake, or believes that it is not. Betty should then, it seems, not be opposed to a pro-choice policy, given that she still believes C (assuming that C, on its own favor a pro-choice policy). Compare this to a case in which Betty does not suspend belief about Q, but rather assigns a 0.5 credence to the truth of Q. Betty might think that while surely uncertain, there is still a 50% likelihood that it is correct. If this is how Betty reasons she could still be strongly opposed to a pro-choice policy - her view could be that such a policy would have a significant risk of condoning very large numbers of seriously wrong actions. So, even low credence in a moral proposition may operate differently than suspension of belief in that proposition (this is on the assumption that the reduced weight principle is false, that is, on the assumption that even if we think there is only a 50% change that a moral principle is correct, we should still assign the moral factors it identifies full weight). 31 Socio-epistemic Grounds for Moral Compromises So suspending a moral belief may in some cases be a much more significant step than merely reducing one's confidence in the truth of that principle (particularly when one assumes that the reduced weight principle is false). Now as mentioned, maybe we are only rarely capable of suspending belief in moral commitments, even when we are rationally required to do so. But acceptances of suspensions are easier to come by - we only need to decide a particular course of action, not to change our beliefs. And we may more often be warranted in accepting suspensions of some of our moral beliefs. But acceptances of suspensions of beliefs in moral commitments have the same implications for what policies we should prefer as actual suspensions of belief in those commitments, and these implications are sometimes much more far reaching than mere reductions in confidence. There is another feature of acceptances of suspensionsn that is worth noting. It is often regarded as an important feature of compromises that compromisers continue to disagree. Not all the epistemic grounds for compromise that I have identified above preserve this feature, and some might then object that we are then no longer talking about compromises. But if Adele and Betty decide to compromise because they both accept to suspend belief in moral propositions that are controversial between them, then they may retain these beliefs. They have accepted to suspend belief for the purpose of their interaction, but they may not have given up these beliefs entirely. So, Adele and Betty could therefore still point to a residual disagreement that makes the compromise painful. 7. Concluding remarks My main aim in this paper has been to explore the socio-epistemic grounds for moral compromise. When we disagree about preferred policies because we fail to agree about what to think about the underlying moral issues, should we then be more inclined to compromise? I have detailed types of cases in which there are such 32 Socio-epistemic Grounds for Moral Compromises socio-epistemic grounds for compromises, and others in which there are not, and I have suggested how the scope of socio-epistemic reasons for compromises may be broadened if we also accept the reduced weight principle, and agree to consider not only beliefs but also acceptances. In my discussion, I relied on a weak notion of a principled reason compromise, rather than the strong notion that May relies upon in his discussion. I agree that on the strong notion, we may never have epistemic grounds for compromise. What I argue is that a policy option may sometimes come out as in one respect better because of the fact that we disagree about pre-compromise positions. References Christensen, D. (2007). "Epistemology of Disagreement: The Good News." Philosophical Review 116(2): 187‐217. Christensen, D. (2009). "Disagreement as Evidence: The Epistemology of Controversy." Philosophy Compass 4/5(Forthcoming): 756‐ 67. Christensen, D. (2011). "Disagreement, Question‐Begging and Epistemic Self‐Criticism." Philosophers Imprint 11(6). Cohen, L. J. (1989). "Belief and Acceptance." Mind 98(391): 367‐89. Frances, B. (2013). Philosophical Renegades. The Epistemology of Disagreement. D. Christensen and J. Lackey. Oxford, Oxford University Press. Goldman, A. I. (1999). Knowledge in a social world. Oxford New York, Clarendon Press ; Oxford University Press. May, S. C. (2005). "Principled Compromise and the Abortion Controversy." Philosophy and Public Affairs 33(4): 317‐48. Wedgwood, R. (2007). The Nature of Normativity, Oxford University Press. Wedgwood, R. (2010). The Moral Evil Demons. Disagreement. R. Feldman and T. Warfield, Oxford University Press. Weinstock, D. (2013). "On the possibility of principled moral compromise." Critical Review of International Social and Political Philosophy 16(4): 537‐56. 33