Wednesday, 6 June 2012

Parfit on Contractualist Consequentialism

In my previous posting I objected to the account Parfit gives of the "veil of ignorance" and in the posting prior to this I advanced an account of deliberative rationality that challenged the view articulated by Parfit. Both of these postings were responses to part of the argument presented in the third of Parfit's 2002 Tanner Lectures. However the arguments discussed in the previous two postings were essentially negative in import for Parfit, ways of clearing the way for the view of contractualism that he wishes to present within this third lecture. The argument that is presented here is for something that has not previously been identified as a possible contender in the area of moral theory, a notion of contractualism that is consequentialist. In this posting I aim to do two things: both present the considerations that enable Parfit to arrive at the understanding that it is possible to reconcile consequentialism with contractualism and also to show why the combined view that thereby arises is unsustainable. So my purpose here is both reconstructive of Parfit's argument and critical of its import. This ensures that this posting will be more than usually involved.

After replying to Rawls in the first two parts of this third lecture Parfit returns to his original account of the basic notion of contractualism in his view, the notion of 'rational agreement', a notion that requires unanimity. In doing so he contrasts the idea of such agreement as understood on the minimal view of David Gauthier with what he terms "Kant's Contractualist Formula", the derivation of which I treated previously. Now, in contrasting these two views at this point, Parfit points a methodological contrast between them based on the fact that the Gauthier type formula requires adoption of principles that everyone would choose whilst, on Parfit's construal "Kant's Contractualist Formula" (or KCF) only requires adopting principles that each reasoner could see as required themselves. (Notice that the contrast between them is thus structurally similar to that between the two versions of the veil of ignorance treated in my previous posting.) So rather than conducting one thought-experiment in which we try to find principles for all of us there are many thought-experiments, basically one each in which we use reference to an understanding of moral belief in order to reach the principles that everyone would accept if rational.

Now in justifying KCF over the Gauthier type formula Parfit appeals to certain safeguards that contractualists are concerned should be in place. One of these is to eliminate inequalities of bargaining power (something Gautheir's formula does not do for reasons that Rawls provides). There is no requirement for each of us to choose principles just on the grounds that others will choose them as we first ask if they are reasonable to us. A second concern that is meant to be met is that we should avoid the outcome where some are viewed as capable of bearing burdens for the sake of others (as often arises from utilitarian conceptions). This second concern is what Parfit terms the contractualists' "protective aim". Again Parfit argues that a Gauthier type view cannot achieve this as the aim of reaching unanimous agreement gives those who possess unequal power an advantage (though, notably, only if we accept the "equal chances" view of the veil of ignorance). Parfit again misunderstands Rawls' view here since he takes Rawls to adopt a self-interest conception of rationality although he had previously indicated he knew this to be false.

KCF neither requires unanimous agreement nor has a veil of ignorance and these characteristics of it are meant to show that it achieves the "protective aim" of contractualism better than previous alternatives considered. Further, its failure to guarantee unanimity means that should it achieve it then this would be all the more impressive. However KCF does imply a particular view of reasons which Parfit elaborates as a "wide value-based" conception as opposed to desire-based theories. The value-based conception is, however, one in which "we have strong reasons to care about our own well-being, and in a temporally neutral way". However, least one voice the view that this implies an egoistic conception, Parfit is quick to point out that it is part of KCF to assume that we can rationally care as much about some other things as well as this, including here the well-being of others and "justice" (although he takes no time here to define this latter notion). So when we adopt KCF we take with it this view of reasons. I'll return subsequently to the point that if we see KCF as involving this view of reasons then we will have grasped KCF in such a way that it precisely is not, despite Parfit's arguments to the contrary, really a contractualist conception at all.

However before we move on to the reasons why I take this to the case it is important to next follow the claims that Parfit makes that enables him to arrive at the conclusion that KCF is both contractualist and consequentialist. The first point to be introduced is the criteria that Parfit gives for KCF succeeding in meeting the problem that contractualism aims to address. This is that it must meet a 'uniqueness condition', such that, there is some relevant principle, and only one principle, that everyone could rationally choose. In illustrating a way that a principle could meet such a condition Parfit gives an example in which there exists a quantity of unowned goods that no one has a special claim on and the distribution of which would have the same total sum of benefits however it was done. In such a case, says Parfit, we can all see that a basic principle of equal shares applies. It is notable that the example of application is a distributive one of some set of resources and not, for example, of a criteria that would define rights or liberties. Such an example already implies a definite conception of what type of thing we should be concerned with when aiming for a notion of unanimity.

A wrinkle that is worth observation in terms of how Parfit justifies the appeal to equal shares is that there is a difference, on his view, between consenting to acts and choosing principles. So we might consent to an unequal share, taking less for ourselves in order to aid others. However this is quite different to adopting a principle that some (whether ourselves or others) should as a matter of course have unequal shares since that principle is not one we could rationally accept. (On the basis of the wide value-based conception of reasons.) This argument is used also to show that the type of principle we should end up with is not act utilitarianism since act utilitarianism could state that it is permissible to give some no benefits at all since the net effects would not thereby be affected.

Having reached this point in his argument Parfit turns next to a further connection between moral beliefs and beliefs about reasons suggesting that our moral beliefs may partly depend on our beliefs about reasons. There are two ways that this connection is perceived by him to go. On the one hand we might hold a view, such as contractualism, which determines the understanding of the rightness of acts by reference to the relationship of such acts to principles that everyone could rationally accept. On the other hand, we might find that moral beliefs depend in some way on our non-moral beliefs about the kinds of things that are good or bad. Now the second sort of relationship between moral and non-moral beliefs turns out to be the one that Parfit exploits and he does so at the expense of the first sort of relation. An example of this second sort of relation concerns our general understanding of hedonic states. The point that Parfit makes about these states is that our reaction to them involves us in accepting that there are certain actions that are good or bad to perform and that this understanding reflects what he terms a "reason-involving" sense of goodness and badness. 

In claiming that we have a "reason-involving" response to hedonic states Parfit takes himself to be in disagreement with Kant given that Kant does not take inclinations to be themselves reason-giving. Parfit views this as a claim to the effect that whilst hedonic states may not be the ground of a moral claim they still could be the ground of a reason-giving claim and that Kant ignores the latter. This, however, is an odd reading of Kant since Kant claims not that there may not be reasons to relate to hedonic states in certain ways but that these reasons are part of the formulation of maxims and it is the endorsement of the maxims that provides the reasons in question, not simple experience of the hedonic states alone. Putting this in Parfit's terms it is not that the hedonic states are themselves "reason-involving" in the sense that the mere experience of them produces a reason to respond in a given way. It is rather that we take the states, in conjunction with maxims in which they feature, to provide grounds for appropriate actions. And this is quite different from the way Parfit puts the case as the way he puts it suggests that it is the experience of the states simpliciter that is reason-involving and that is what Kant denies.

Parfit takes hedonic states to be "reason-involving" as he assumes that there are person-relative reasons always to avoid painful states. I have denied this claim even though it seems true intuitively. The reason I deny it is simply that the painful state is not itself motivationally sufficient for action as it has to be related to other considerations that are prevailing at the time it is experienced. Parfit's claim could be reformulated as saying that pain is something we avoid cateris paribus but the conditions involved remain important and indicate that the response to pain is not one we normally assume to be simply permissive of actions following a certain pattern (say, of avoidance). Further Parfit moves swiftly from the claim that the response to hedonic states has this person-relative sense to the more general claim that painful states are impersonally bad. It is important to note that there are counter-intuitive aspects to Parfit's claim here just as I have indicated there are to the way I denied his claim concerning person-relative reasons. The counter-intuitive sense of his claim becomes apparent in the following sentence: "If more people suffer, that would be worse, even though it may not be worse for any of these people". 

In a claim like the one just cited it becomes apparent that Parfit conceives of hedonic states in an aggregative fashion so that the more they point one way or another en masse the worse or better the general situation is. This is to view hedonic states, however, in complete abstraction from the experience had of them as Parfit confesses when he agrees that the "worse" outcome on his view may not be worse for anyone. This indicates that the "non-moral" sense of badness that he has is radically impersonal since its assessment of badness is not one that refers to criteria of what is bad for persons. This "non-moral" claim is, however, one that it is not obvious should be accepted as a serious criteria of goodness and badness by persons. Parfit sloughs over this point passing from the acceptance that the "non-moral" sense of goodness and badness he is advocating is impersonal to the quite different claim that it is agent-neutral. It is not at all evident that a seriously agent-neutral claim is one that has to be impersonal. After all if we think of the "mere means" criteria of the Formula of Humanity we refer to an agent-neutral claim as it specifies a way of relating to persons generally regardless of who they are but it is not thereby "impersonal".

Parfit's conception of agent-neutrality is of a form of reason that is a "reason for everyone" and this applies to the application of the Formula of Humanity but it does not apply to his kind of "non-moral" form of goodness since the latter is in fact a reason for no one which is precisely implied by the notion of its impersonality. Each of us has, Parfit implies, reasons to prevent or relieve pain, whether of oneself or others. But this is untrue not merely because it is not always possible to do so but also because the pain experienced may not be, in the given case, a "worse" state for the person in question. This is directly admitted by Parfit and since it is so it follows that the response to pain does not have the over-riding "non-moral" value Parfit claims it has. Not only is this so but if we adopted a different standard, from the "mere means" criteria of the Formula of Humanity, for example, we would have an agent-neutral way of responding to others that was not impersonal and which would be a general guide to conduct.

Now if the first point of difficulty with Parfit's view concerns the way in which he slides from 'agent-neutral' reasons to "impersonal" ones the second problem arises from the conception of how to weight the reasons that his "impersonal" view has led us to consider. Parfit states that this should be in terms of outcomes so that the best "impartial" reason for preferring something is if the features it has give everyone "the strongest impartial outcome-given reasons to prefer this outcome". At this point it appears we have simply specified a consequentialist view and so one may wonder what happened to the apparent contractualist procedure. Parfit is sensitive to this point and agrees that the impartial outcome-given reasons should not be assumed to be the only impartial reasons. Further Parfit also accepts that some other types of impartial reasons might outweigh on given occasions impartial outcome-given reasons. However whilst Parfit refers here to rights as other sources of impartial reasons he gives no general account of rights and nor does he provide any lexical procedure for describing the relationship between rights and impartial outcome-given reasons. 

The view thus far is described by Parfit as "semi-consequentialist" since it gives some weight to impartial outcome-given reasons but not necessarily over-riding weight as yet. Consequentialism generally does give over-riding weight to such reasons as Parfit agrees in specifying act-consequentialism as the view that acts are right just in case they make things go best in terms of impartial outcome-given reasons. Parfit is clearly no more willing to endorse this view than he was act utilitarianism but this leaves him open to the attractions of rule consequentialism in which acts will be right if they are permitted by one of the principles whose acceptance would make things go best in an impartial reason-giving sense. Since this conception is related to the "wide value-based" view of reasons it is a form of consequentialism that Parfit feels able to term a reason-consequentialism. 

At this point the basic problem emerges with some clarity. Contractualism involves the conception that we aim to develop a view that understands that as right which we can all rationally accept. However Parfit understands this claim by means of the consequentialist idea that was already expressed in his value-based conception of reasons to the effect that we have strong reasons to care about considerations of well-being and subsequently moves from this to the understanding of impartial reasons in terms of impersonal outcomes that no one in particular ever has some reason to want. This ensures that his conception is one that presents as a reason we should all rationally accept a consideration that no one of us has any reason to endorse. And that hardly meets the contractualist criteria.

Parfit's claim that contractualists and consequentialists are "climbing the same mountain on different sides" is essentially built on the claim that the consequentialist appeal to impartial outcome-based reasons will coincide with contractualist appeals to what everyone has a rational basis for choosing but in doing so it overlooks that the former presents for the latter reasons that no one has any particular reason to endorse. Not only is this so but there is a further problem involved in the relationship between these theories. This problem was expressed by Rawls in section 5 of A Theory of Justice where Rawls contrasted two methods of moral reasoning. On one of them "the good is defined independently from the right, and then the right is defined as that which maximises the good". This is the means by which consequentialism works. It has an independent conception of the good, as Parfit claims here when he speaks of hedonic criteria as providing a "non-moral" standard of goodness. This methodology contrasts with the other one that Rawls speaks of and which he follows Kant in preferring. This is outlined by Kant as indicating that it is only by means of and through the moral law that we arrive at a view of the good. So the good is not then independently defined but viewed rather through the prism of the right.

This second type of relationship is implied in contractualism which views the good not as an independent variable but as something that is arrived at by means of right reasoning. On the contractualist conception the right is that which we would all rationally want and the good emerges as that which is constrained by appeal to this notion of the right and thus becomes manifest to us by means of it. This methodology is quite distinct from that of consequentialism and is directly opposed to it. Not only is this so but correctly seeing the right as prior to the good in the way that Rawls and Kant direct us to implies that there is always a basis of lexical priority between the "moral" and the "non-moral" goods since the nature of what can be permissibly classed under the latter is defined by the former. 

Parfit cites Rawls' account of the way that consequentialist theories involve giving the good priority over the right and appears to agree with Rawls that this indicates a problem with such theories. However the reason Parfit gives for rejecting the claimed priority of the good is not Rawls' reason and it is also not a good reason. Parfit states that if consequentialism requires us to take the good as prior to the right then this implies that the right is trivially defined so that we end up with tautological claims. This is not the way Rawls understands the point. Instead Rawls' point is that taking the good to be prior to the right ensures that the independently defined good is one that is given over-riding priority in considerations that inevitably require abstraction from the nature of reference to persons. This is precisely what we have seen follows also from Parfit's view. Parfit attempts to avoid this by stating that whether the right or the good is made prior you end up with trivial claims but this is simply false. On the one hand you produce a claim that there is an impartial reason-involving good which is grasped in an outcome-based way (as in his view) whilst on the other hand you arrive at a conception of law and right that defines and describes what can be taken to be good (as on Rawls and Kant's view). And the difference between these two is anything but trivial.

The development of Parfit's own view is one that he sets up as involving a criteria that he thinks does mitigate its general consequentialist thrust when he introduces criteria of distribution of benefits and burdens that is meant to favour an egalitarian conception. So the goodness of outcomes is then taken not in a bluntly aggregative sense but in a way that relates this aggregation to the way the benefits and burdens are distributed. Rawls considered such a view in section 5 of Theory and determined that it would no longer be "teleological" in a classical sense as it would now take distribution to be an independent variable. Taking it to be so implies a notion of right (through distribution) that is meant to apply to the impartial outcome-based reasons and to provide a way of assessing different outcomes. Notably it does not affect the understanding of agent-neutral goods as impersonal goods however so it remains in its basic orientation a theory that is still based on the priority of the good over the right. Simply accepting that not all distribution is equal so that there is some form of distribution that is morally preferable to others and thus endorsing an egalitarian principle is not in itself sufficient to dilute the general consequentialist thrust of the theory. The nature of what is good and the form of right in relation to this good remain broadly consequentialist with only the balancing of distribution taken to be one that has a basis for preference that is not blunt. But the rationale for the preference for equality is not evident and the goods distributed are still essentially understood in a way that abstracts from consideration of the good of persons.

In the next posting I am going to conclude assessment of the third of Parfit's 2002 lectures by coming back to how the conclusion of his argument relies on his assessment of KCF and what this tells us about the kind of reading of Kant that Parfit has undertaken in his work. 

No comments: