Parfit, Kant and Consequentialism

In my last posting on Parfit I discussed the general problems I think there are with his notion of "contractualist consequentialism" as presented in the third of his 2002 Tanner Lectures. In this posting I want to look at the concluding section of this third lecture in order to see how Parfit brings the argument there to a conclusion and to use the discussion of this conclusion to demonstrate again some reasons for thinking that Parfit stacks the deck rather when he reaches the conclusion that Kant gives the basis for a contractualist view that is substantively consequentialist.


Parfit opens the final section of this lecture with a statement that whilst Kant was not a consequentialist the argument's point is not to examine the reasons why Kant was not a consequentialist but instead to point to some reasons for taking it to be the case that we can develop a consequentialist view by appealing to some elements of Kant's positions. In making this case Parfit appeals to an important distinction. The distinction is between what he terms Kant's "moral beliefs" on the one hand and the "implications of his principles" on the other. The former are what are involved in Kant's not being a consequentialist on Parfit's view, whilst the latter, by contrast, point to consequentialist conclusions. This distinction is an important one, as we shall see in looking at the concluding phase of Parfit's argument.


Parfit's case is built by reference to the principle he extracted from Kant's Formula of Humanity as the first part of it. This principle is what Parfit terms "Kant's Consent Principle" (KCP). The "consent" involved is framed by Parfit as "rational consent" and it is connected to the act consequentialist view that rational consent can be seen as agreement that everyone be treated in ways that would make things go best. If we look at rational consent in this manner, then treating people in a way that implies adoption of an axiological standard of value is equivalent to treating them in ways to which they could rationally consent. One of the points Parfit rushes to make here is, however, meant to assuage a problem Rawls raised against utilitarianism. Rawls indicated that a problem with accepting "classical utilitarianism" was that it required some to  sacrifice themselves for the benefit of others and that such a sacrifice was one that the ones being sacrificed could accept only if they were prepared to sacrifice the very point of being in an ethical space at all (so it is connected to his problem with the "separateness of persons"). However Parfit thinks he has a reply here as if the act in question would impose great burdens on some for the sake of greater benefits for others then this act would not make things go best but instead go worse. This is despite the fact that the person in question might "rationally consent" to the act. On this basis Parfit argues that consequentialists don't have to be seen as utilitarians.


The separation of consequentialism from utilitarianism that this argument involves does, however, require, as a consequence that an optimific act is not one that can be assessed only through the prism of rational consent. Rational consent is thus not a means to determine the rightness of the act if such rightness is standardly viewed in terms of optimific understanding. In which case, whilst Parfit thinks that KCP could be accepted by act consequentialists, it is not sufficient as a standard of rightness for them. 


Parfit, however, takes the argument a different way as he now moves away from the consent principle and back to the principle he earlier termed "Kant's Contractualist Formula" (for which see this earlier posting). The contractualist formula requires that we appeal to principles that everyone could rationally choose and we could not rationally will that everyone accepts act consequentialist principles and so the act consequentialist principle is not one we should act upon. Notably, however, and remarkably, Parfit has here done something that cuts heavily against the grain. He has presented act consequentialism as sufficiently robust that it can provide a defence for the separateness of persons, such a defence that it can be distinguished from utilitarianism, and then, having done this, has used an appeal to contractualist views about universal rational acceptance, to undercut act consequentialism after all. It would, according to standard ways of presenting the moral situation, be rather more apt to say that act consequentialism, like act utilitarianism, is insufficiently robust to prevent appeal to demanding notions of self-denial and then use that to support a view of consent principles that did not view the latter in optimific terms. So Parfit's initial argumentative strategy is, to say the least, very odd.


Parfit refers now to the argument Sidgwick famously used that it would be best, from a utilitarian point of view, if not everyone was a utilitarian. This argument, which shows that utilitarianism lacks transparency, is one that Parfit endorses here against act consequentialism. In doing so he appears to appeal to the view that moral beliefs matter in the sense that if we knew it to be true that it would not make things go best if everyone acted in ways dictated by principles of maximising what would go best, we should prefer it that not everyone had the belief that they should make things go best. Indeed, not only does Parfit appear to make this move, but he also uses it to prefer rule consequentialism to act consequentialism since rule consequentialism requires us to act on principles whose acceptance would make things go best even should this be the view that not everyone believe that they should act in ways that make things go best.


Parfit now constructs an argument that is meant to move from  Kant's "contractualist formula" that we ought to act on principles whose universal acceptance everyone could rationally choose to the conclusion that rule consequentialism is correct on Kantian contractualist grounds. Now there are several stages in this argument which bear close analysis if a response is to be made. The first point to look at is how Parfit views reasons. Parfit introduces and critically interrogates the premise everyone could rationally choose whatever they would have sufficient reason to choose. This premise is interrogated as what we have sufficient reason to choose is argued by Parfit to depend on facts whilst what we rationally choose is, by contrast, based on beliefs. The difference matters since false beliefs give us reasons to choose something that it is not ultimately rational to choose. So the application of Kant's "contractualist formula" requires congruence between beliefs and facts. 


The next premise is to the effect that everyone would have sufficient reason to choose that everyone accept the principles whose acceptance would make things go best assuming the congruence we have established as a requirement for this. Now, in critically evaluating this premise, Parfit points to the "impartial reason-involving" sense of normative reason as the basis for viewing how to evaluate whether or not things have gone best. However, in assessing this claim, Parfit also refers back to his earlier distinction between "moral beliefs" and "principles" since he rejects the objection that such a principle might produce immoral action on the grounds that such an objection illegitimately introduces moral beliefs. There is a problem with this reply since it assumes that the contractualist procedure is thus far not loaded or is effectively neutral between different views. But it is not. In taking our standard in terms of an axiological conception of evaluation we have already adopted a consequentialist standard into our reasoning so it is evident such a standard will be our conclusion. If, however, our argument precisely concerns a question about how "rational consent" is to be understood, for example, then we cannot simply stipulate that it be seen in not merely an impartial sense but in terms of a view of impartial reason that is pre-set in terms of outcomes. Viewing impartial reason in terms of outcomes incorporates a "moral belief" into the "neutral" evaluation. And it does so in a way that is then used to appeal to "principles" as if they are not described in ways that are set against other "principles" where the other principles are then merely demoted to the status of "beliefs".


The reason why Parfit thinks that such a division between "principles" and "beliefs" is stable is because he sees the former as tracking "facts" and making them congruent with "true beliefs" whilst an insistence on "moral belief" simpliciter is simply an argument from what we "already accept". However the reference to "facts" here is one that involves a view of the factual that is itself normatively loaded since it takes factual support to consist in a view about what is axiologically supported and so it is not neutral between views.


Parfit argues that in pursuing this strategy he is following a standard contractualist procedure and he even appeals here to the example of Rawls to support him. Rawls, however, when constructing the original position, does so by building the position out of what he termed "formal constraints of the concept of right" and such constraints are not discussed by Parfit. Particularly important from such constraints was a solution formally argued to the "priority problem" with regard to principles. Parfit stipulates a solution by taking axiological criteria to be supreme, something that cuts precisely against his earlier recognition of the point that there is a case to be made against abstracting from the separateness of persons.


Parfit next goes on to consider this claim by balancing what he regards as the considerations of personal points of view with those of impartial reason. In doing so Parfit stresses the idea that there are non-self-interested personal reasons which, indeed, there are in terms of different conceptions of the good. However, Parfit does not consider this question in terms of different conceptions of the good but only in terms of personal reasons being seen, if they conflict with impartial ones, as self-interested. Parfit does not consider that personal reasons that oppose what he views as "impartial" reason may also be principled in their opposition. Because of this Parfit simply arrives at the conclusion that it is always optimific to follow impartial reason and that we should accept impartial reasons as personal reasons given that this is so. But this is not evident at all if there are impartial reasons that are non-optimific in form as is the case with non-consequentialist views. 


Because Parfit has construed the choice situation in the way he has he cannot avoid the problem of unreasonable burdens that, at the beginning of the discussion, he wished to argue was not a necessary element of consequentialist views. Assuming that one holds a non-optimific view of goodness by means of a conception of the priority of the right over the good one will always be told that ones view of the good is only personal and should be sacrificed optimifically which is to abstract from the separateness of persons in precisely the way Rawls complains of. This is still the consequence of Parfit's preferred rule consequentialism which is why it still incorporates the classic consequentialist move of requiring us to see impartial reasons as always optimific in form. Parfit thinks he can appeal against this since he argues that ultimately he is appealing to Kant's "contractualist formula" and basing rule consequentialism on this. However this is dishonest as the contractualist formula, in application, is always used to bolster the conception that impartial reason is framed in terms of optimific outcomes so it is only a supplementary principle meant as a support for consequentialism and is not, as Parfit pretends, foundational for ethical principles. 


So whilst Parfit concludes by saying he has not argued that we should become rule consequentialists but only that Kantian and contractualist premises support rule consequentialist conclusions this is not what his argument shows. What his argument shows is, instead, that if contractualist and consent principles are interpreted axiologically then they support consequentialism. But that is only because they have been pre-defined in such a way that consequentialism is their inevitable focus. Parfit assumes that the only response to this argument is to favour "moral beliefs" over "principles" but, as I have argued, this division is founded on an incorporation of a "belief" into his account of "principles" in the first place.

Parfit and Kant's "Contractualist Formula"

The third of Parfit's 2002 lectures concerns contractualism and involves an extended discussion of what he calls Kant's "contractualist formula". Since the formula in question is not one that Kant himself specifically states and since Parfit's derivation of this formula occurs in the concluding part of the second of the 2002 lectures it is necessary to return to this derivation before the argument of the third lecture can be considered. This is what I will do in this posting, which is devoted only to uncovering the argument by which Parfit arrives at and justifies this formula.


In the final section of the second 2002 lecture Parfit turns to the question of whether, in the course of this lecture, he can be said to have misinterpreted Kant's understanding of universal law. Whilst considering this possible objection Parfit points to the way in which Kant discusses, in the case of beneficence, the application of the law of nature formula by means of appeal not to what "everyone" could will but rather to the nature of the specific person's willing. Parfit also rejects the view that Kant's universal law test should be seen in terms of Rawlsian notions of the veil of ignorance. Parfit presents both Scanlon and Rawls as offering not interpretations of Kantian notions of universal law but as instead providing ways of revising it. This leads to Parfit presenting a complicated diagram of possible ways the universal law formula could be understood. Prominent amongst the possible understandings for Parfit is the treatment given by Scanlon which leads Scanlon to state that an act is wrong "unless everyone could rationally will that everyone acts in this way". 


However whilst Parfit is favourably impressed by Scanlon's formula he does not leave the matter there. Scanlon's formula is one that Parfit thinks is impartial but that it goes further than we have to since some acts are right without being ones that everyone could perform. This leads Parfit to revising Scanlon's formula so that it becomes, as Parfit terms it, the "formula of universally willed acts" which states: "an act is wrong unless it could be rationally willed by everyone". This formula is now understood to be equivalent to the first half of the Formula of Humanity on Parfit's construal of the latter, which he has earlier termed "the Consent Principle". On the basis of this we treat people only in ways that they could rationally consent to. The wider formula of this that has become the formula of universally willed acts now adds that an act is wrong unless everyone, if they had the choice, could rationally choose the act in question as one that would be done. 


However despite the argument having thus given us a formula that is at least loosely related to something Kant said Parfit is not finished yet. Parfit had earlier invoked the quite different standard that he termed Kant's "moral belief" formula which involved stating that an act is wrong unless we could rationally will it to be true that everyone believed it was permissible. This formula introduces a lot of complicated epistemic issues that Parfit does not really discuss but at this point all he does is argue that the so-called "moral belief" formula is one that can be converted into Scanlonian terms so that it becomes a claim to the effect that an act is wrong unless everyone could rationally will it to be true that everyone believed such acts permissible. This version of the "moral belief" formula interprets the claim about belief in a better way in a sense since it is no longer merely a claim about a given person's belief states and this is surely preferable as otherwise we seem mired in subjectivism. This version of the moral belief claim is meant to be understood as a way of addressing the problem of what general principles of action we could all will. 


Not only is the Scanlonian notion of moral belief a reply to a genuine Kantian question it also has the advantage, according to Parfit, of being suggested by Kant's formulas of autonomy and the realm of ends. The point about the formula of autonomy is that it involves a claim about rational beings giving themselves universal law through the maxims of their will. The understanding of this that Parfit wishes to champion is in terms of the Scanlonian version of "moral belief". The Scanlonian version is also termed by Parfit a "formula of universally willed moral beliefs" but it is shortened and given the name "Kant's Contractualist Formula" so that it becomes: "we ought to act on the principles whose universal acceptance everyone could rationally will". Parfit concludes the second lecture by contrasting this formula with a Rawlsian formula that involves, instead, reference to choice under conditions where we know nothing about the circumstances of others. The contrast between these formulas will motivate the structure of the third 2002 lecture. I am making no comment here on the argument Parfit has given as I wanted merely to draw it out as the basis on which the formula that Parfit consistently in the third lecture refers to as "Kant's" formula, despite not being stated anywhere by Kant, is arrived at.

Parfit and Kant on Treating Persons as Ends (II)

In a recent posting I looked at how Parfit discusses the Formula of Humanity's declarations against treating persons merely as a means in his 2002 lectures. Now I want to turn to looking, by contrast, at how Parfit's account of this topic is set out in Climbing the Mountain, the first book-length form of the work that eventually became On What Matters. The material from the 2002 lectures that I have been looking at in recent postings on Parfit is developed in Climbing the Mountain into three chapters, two of which expand on the discussions given first in the first of the 2002 lectures, and the last of which introduces topics that were not part of the first of the 2002 lectures at all. So it will require 3 postings to look in full at the way the treatment of the Formula of Humanity is presented in Climbing the Mountain.


Chapter 5 of Climbing the Mountain replicates and expands the discussion of the element of the Formula of Humanity that concerns treating people merely as a means and opens with material that is essentially the same as in the first 2002 lecture. This includes distinguishing between treating people as a means (meaning just using someone's abilities, activities or body) and treating them "merely" as a means (viewing them purely as an instrument or tool). However, an objection is mentioned from Frances Kamm who took this understanding of treating someone merely as a means to be too weak since, on this account, it would appear sufficient for a slave-holder not to be said to be treating someone merely as a means if he allowed the slaves to rest during the hottest part of the day. Parfit takes this objection seriously though it is far from clear to me why he does since, after all, treating someone as a slave is prima facie to treat them merely as a tool (as Aristotle recognised clearly). After all, if a tool is essential for a task one wants to perform then you don't act in such a way as to break it so the slave-holder in the example is merely a prudent tool-owner, not someone who is failing to treat his slaves merely as a means.


Parfit, however, affects to take Kamm's objection seriously and re-formulates the mere means principle so that it states not just that it is wrong to treat anyone merely as a means but, in his second formulation of the principle, that is wrong to treat anyone merely as a means "or to come close to that" (apparently as a way of responding to Kamm). So the slave-holder only "comes close" to treating someone merely as a means if they are sufficiently considerate to allow them to take time off work during the hottest part of the day. As already indicated I find this concession utterly unnecessary.


After beginning in this confusing way Parfit moves on to a negative claim involving what kinds of concerns would rule out the idea that our treatment of someone was not to be correctly viewed as treating them merely as a means. This restrictive negative construal indicates that we are not treating someone merely as a means if our treatment of them is either a) governed or guided in "sufficiently important ways" by a relevant moral belief or concern or b) we do or would relevantly choose to bear some great burden for this person's sake. 


The first of these ways of preventing some treatment of someone as not being subject to the charge that we are viewing them merely as a means is treated with some care by Parfit. So, for example, if a slave-holder doesn't whip his slaves merely because he is aware that this would give him a sadistic pleasure then this does not show that he is not treating the slaves merely as a means, an odd point given Parfit's apparent earlier allowance of Kamm's flawed objection. Part of the point of raising this odd case is, however, to suggest that it is not always obvious if we have a case of the first type at hand (since, apparently, if the slave-holder wasn't whipping the slaves due to some "belief" in their worth that would exculpate him!).


The second way of preventing something being viewed as treating a persons merely as a means does not only involve taking on great burdens for someone else as these burdens also have to have sufficient moral relevance to the acts being considered. The introduction of these qualifiers is meant to sharpen the way Parfit has distinguished between treating a person simply as a means and treating them merely as a means. Treating someone as a means is viewed by him as only referring to our intentions whereas, by contrast, treating someone merely as a means depends not only on this but also on underlying attitudes and policies (akin to how some have viewed the Kantian idea of a "maxim").


After reinforcing this distinction Parfit next points to an ambiguity in the notion of treating someone merely as a means since it could refer either to attitudes or to actions and he views the application of the notion to actions as something that is more difficult. In making this point Parfit introduces the example of the egoist who saves a child from drowning but only with the aim of being rewarded. The point of introducing this example is to state that whilst the attitude here is a wrong one, the action is not. However, understanding the principled basis of this distinction proves complex as becomes clear when Parfit discusses ways of incorporating reference to it in one's general account of the mere means principle. So, incorporating the distinction, in reference to the case of the egoist just considered, might lead us to introducing a third restrictive condition on evaluation of what kinds of things would not merit the charge of treating someone merely as a means.


The way that would go would be to give the formulation that we don't treat someone merely as a means if we know our acts won't harm the person in question. However, as Parfit uses an imaginary case to show, this way of framing an exemption from treating someone as a mere means is pretty problematic since it would seem to allow conduct short of actual harm even though it was motivated entirely by egoistic considerations and included no reference to benefiting someone and that appears odd. So it might well be safer, rather than trying to add this third restrictive condition on the evaluation of the action of the egoist saving the child, to instead regard this action only as one that lacks moral worth.


This leads Parfit next to a third formulation of the mere means principle so it now states that it is wrong to treat anyone merely as a means or to come close to that, if our act will also be likely to cause harm to the person. This is another puzzling feature of Parfit's argumentation, however, since stating that the act of the egoist is one that lacks moral worth is to state something quite different from saying that their action is wrong and yet this amendment is introduced as a way to indicate that treating someone merely as a means is not a distinctively wrong-making manner of treating them and this simply does not follow from his argument.


Parfit next moves to the stage of trying to give a unified treatment of the Formula of Humanity by tying together the mere means principle with the Consent Principle that he earlier located as expressed in the first part of Kant's formula. The restrictions on the application of the mere means principle included reference to conduct governed by a relevant moral belief or principle and the Consent Principle is now taken to be such a principle and hence to play the role of the first way of restricting the application of the Mere Means principle to evaluation.


Parfit next introduces the same "trolley" examples that were given in the 2002 lecture and which I discussed here. As in the 2002 lecture the point of introducing these trolley examples is to argue that the sense of "consent" in the Consent Principle is not actual consent. However, after making this argument, Parfit now adds some additional considerations that were not present in the 2002 lecture. These involve a potential objection to his argument that trades on a different understanding of the mere means principle to that which Parfit himself has given. This different understanding is expressed in what Parfit terms "the standard view" which states: "if we harm people, without their consent, as a means of achieving some aim, we thereby treat these people merely as a means, in a way that makes our act wrong".


Parfit objects to this "standard view" not least because it misidentifies what may be happening in harming people as a means since we may not be treating these particular people as a means and, even if we are, we may not be treating them merely as a means. But, most importantly for Parfit's own discussion, even if we are so treating them, this may not be sufficient for it to be said that we have acted wrongly. So you might harm someone (as in self-defence) without treating them as a means. Secondly, we might treat someone merely as a means on the standard view without evidently having done something wrong as when you cause harm to someone to save someone else (who is not yourself). 


Parfit next considers some typical claims that have been made about treating people merely as means as when Onora O'Neill and Christine Korsgaard highlight coercion and deception as treating others merely as a means. However, in response, Parfit points out that if I prevent someone from killing me by giving them a false impression of what I have done or am going to do this seems insufficient for the act in question to be viewed as wrong. Korsgaard also makes a point about free-riding pointing out that wrong actions are often such in that they only work due to the assumption that they won't form a general pattern of behaviour. However, as Parfit points out in a Bad Samaritan case I am not treating someone merely as a means if I walk on by ignoring the injured party so the wrong-making characteristic of this action has not been pulled out by application of the mere means principle (which, thus, cannot be identified simply with the generalised injunction against free-riding).


The point about the Bad Samaritan example, on Parfit's construal, is that the person responded to in the way indicated is treated not as a mere means but rather as a thing (hence not as a person at all). This may well indicate a different kind of wrongness to that of treating someone merely as a means and a much more serious moral failing thus may be involved here (although Parfit does not, having made this point, return to Korsgaard's point about free-riding). 


In the conclusion of this chapter of Climbing the Mountain Parfit returns to the distinction between regarding people merely as a means and treating them in this way. Treating someone merely as a means is viewed by Parfit in a very restrictive sense, however, since he regards a gangster who buys a cup of coffee merely because it would be too much trouble to steal it as treating the vendor merely as a means. This is a case, however, of acting in a way that lacks moral worth but it is far from obvious that it means treating someone merely as a means (it may just involve treating them simply as a means). The introduction of this flawed example is meant to pave the way to a further consideration of the third way Parfit formulated the mere means principle and includes a further treatment of harmful means in which harm is regarded as something that ought not to be caused except if it is the least harmful way to achieve an aim and, given the goodness of the aim, the harm caused is not disproportionate.


In considering this amendment to the third mere means principle Parfit points out that we have no obvious guidance for how to consider disproportionate harm. However, whilst this is a fair point, Parfit over-plays it since he takes it that the amendment proposed would rule out even mild forms of harm, something hard to square with the point about proportionality. 


Parfit's general aim in the chapter is surprisingly negative. It consists in a general claim to the effect that the mere means principle is insufficient to characterise an action as wrong (though it can define an attitude as wrong). This is in accord with the treatment of the principle that was given in his 2002 lecture but the chapter of Climbing the Mountain works harder to establish this conclusion without, however, being obviously persuasive.

Parfit and Kant On Treating Persons As Ends

In my recent postings on Parfit I have concentrated on how he looks at the first part of Kant's formula of humanity, the part that he believes involves discussion of the notion of "rational consent". The first place where Parfit formulated this view was in his 2002 Tanner Lectures and I broke off from consideration of these lectures after this posting in order to track the development of the views first expressed there in subsequent drafts preparatory to the publication of On What Matters and, finally, in the chapter length-discussion in On What Matters that was the subject of my last posting on Parfit. 


I want now to open a series of postings that focuses both on how Parfit responds to the second part of Kant's formulation of humanity and how he sets out a general reading of the overall formula of humanity. This will require, firstly, a lengthier analysis of the first of Parfit's 2002 Tanner Lectures and subsequently an account of the further preparatory drafts leading up to On What Matters before we can return to the analysis of the next chapter of On What Matters. As previously, therefore, the next element of discussion of Parfit's encounter with Kant will be lengthy and textually complicated.


The second section of the first 2002 lecture opens with a report of the general complaint that using people is wrong. As Parfit explains, however, it is far from clear that, when the complaint is so simply made, that it is right since there appears nothing wrong in principle with using a friend as a dictionary or a loved one as a pillow. This is why, in Kant's formula, it is not treating people as a means that is condemned but treating them merely as a means and not also as ends in themselves. So it is not using people that is wrong as such but just using them.


The way to avoid just using someone is to act with regard to them in such a way that one's actions can reasonably be said to be constrained by some form of consideration for them or, put otherwise, if one can be said to be acting under some moral principle that constrains one's conduct with regard to them. Parfit construes the reference to "maxims" in Kant's formulations as an expression of what he calls "underlying policies and attitudes" and considers actions that are constrained in the appropriate way as expressions of such "policies and attitudes".


However, although Parfit appears to present action as expressive of such policies and attitudes and to take the Formula of Humanity to govern the way in which attitudes and policies can be appropriately formulated such that they are constrained in the right way he nonetheless is not thereby convinced that actions that treat someone as a mere means are thereby necessarily wrong. In other words, Parfit's first move appears to be one of viewing the Formula of Humanity as providing us, in its reference to not treating others merely as means, as giving us an appropriate guideline for the attitudes that underlie actions without thereby providing us with a means of assessing the right actions to perform.


The reason for the appearance of this split in Parfit's account is that the attitude expressed by someone in a situation may be wrong without the action that follows from the expressed attitude also being wrong. Since, however, it appeared that Parfit was presenting the reference to maxims as indicative of how to understand the policy that Kant was recommending it follows from his construal that this policy is not sufficiently comprehensive to determine the attitude we should have to the conduct that is produced by the adoption of the "wrong" attitude.


To make this point clearer Parfit refers to the example of someone who treats a coffee seller as a mere means. Such a person pays the coffee seller rather than stealing from them only because in many cases it would be too much trouble to steal from them. The person in question thus has an attitude towards the coffee seller that is worthy of moral condemnation but, unless they proceed to steal from the coffee seller, their action is not appropriately one that should be taken to worthy of moral condemnation (even if it is also not one that is worthy of moral praise as follows from Kant's examples in Groundwork I).


Generally we could say of the person with this bad attitude that they are acting in accordance with duty but not from duty just as is the case with the shopkeeper who only gives everyone the right change because this is generally in accord with their best interest and not because this is the right thing to do. So Parfit now formulates a "mere means principle" as indicating that there two ways in which treating someone as a "mere means" can be said to be wrong, either by "regarding" them merely as a means or by also harming them, without their consent, and thus acting in such a way that one treats them merely as a means. 


After having arrived at this formulation of the second part of the Formula of Humanity Parfit now combines it with his view of the first half of the Formula to produce the overall account of it as stating that we do not treat someone as a mere means if we adopt the Rational Consent principle as an appropriate constraint on our means of acting with regard to them. What this entails is subsequently set out by consideration of a set of thought experiments that bear close comparison to classic "trolley" problems but I will leave consideration of these to the next posting I do on Parfit.

Parfit, Humanity and Consent (IV)

After perusing the various drafts that went into the making of Parfit's account of consent in On What Matters, most recently in the pre-publication version, I want, in this posting, to simply lay out the structure and general content of Chapter 8 of volume 1, ostensibly concerned with the topic of "possible consent". When you look at the chapter as a whole and break down the movement of its argument there are some straightforwardly surprising characteristics of it.


So, for example, the Chapter opens, as the earlier drafts did, with a description of Kant's formula of humanity and, in particular, with the way that this formula is related to the discussion of false promising. It is by means of this connection that Parfit is enabled to focus less on the sense of Kant's formula itself as on the apparent reference to treating people in ways to which they could not possibly consent though the latter reference appears only in connection with this example of false promising and not when the three other  examples to which the formula is related are discussed. 


It is by means of this account of consent in the treatment of false promising that Parfit describes and criticises views of Christine Korsgaard and Onora O'Neill who, in being treated in such a concise manner, certainly have cause for complaint of Parfit's compressed conception of their readings. The point of Parfit's first section in this chapter is, however, not primarily one that arises from exegetical concerns. It is rather to stress problems with viewing the wrongness of coercion to lie mainly in the claim that it makes consent impossible since, as he correctly points out, certain types of coercion are precisely not generally regarded as wrong despite clearly being seen as examples of coercion. Hence when we object to coercion, it would seem, there is a ground for this objection that does not rest simply on its being coercion that is being exercised.


The second section of the chapter moves to clarify this point by focusing on what it is about consent that matters to moral claims on its behalf. Here Parfit repeats the claim of his earlier drafts that the wrong-making character of appeals to consent cannot reside, as some remarks of Korsgaard and O'Neill might lead us to think, on appeal to the "Choice-giving principle" that states simply that it is wrong not to give other people the power to choose how we shall treat them. Such a naive principle is, indeed, one that it would be hoped should not arise from any kind of careful Kantian thought since it would appear that, on its ground, we should simply, for example, buy anything that someone wants to sell to us. By contrast to this principle Parfit instead emphasises what he terms the "Consent principle" which states that it is wrong to treat people in ways to which they could not "rationally consent". This emphasis on "rational consent" does have the potential problem that it appears to build quite a bit of normativity into the general notion of reason but Parfit does not appear to follow Kant in understanding this to be a reference to the "end of acts". 


Parfit presents Kant's point in the following way: "Kant must mean that, when we are choosing how we shall treat other people, we ought always to act with some aim that these people would be able to share" (181). But this "being able to share" that Parfit refers to here is not simply a way of stating the conceivability of a shared end as it rather requires the rationality of this end to be acceptable to others. However, this rationality is not understood here by reference simply to the "end of acts" because this would not in itself include discussion of the means by which ends can be achieved. That Kant's account does, however, include some constraint on this is clear enough when we remember that the Formula of Humanity involves reference to treating others as ends in themselves but this part of the Formula is not included in the chapter in question as Parfit has, for the purposes of the argument of this chapter, deliberately foreshortened the formula.


The third section of the chapter concerns reasons to give consent and here Parfit dismisses Rawls' attempt to view the "Consent Principle" in terms derived from Kant's Formula of Universal Law. Oddly enough, however, one of Parfit's reasons for dismissing this view of Rawls appears to consist in the simple assertion that Kant was inconsistent! It is usual, in interpreting philosophers, not to adopt this view unless it is absolutely necessary to make the best of their positions but here Parfit seems to assume it rather breathtakingly easily. More importantly, in this section, Parfit fills out his understanding of consent by suggesting that it is consent in the act-affecting sense that Kant means and, further, this consent has to be understood as "informed". These additions lead Parfit to refine the "Consent Principle" and he then adds some conditions on its acceptability as he takes it to be the case that it should not require us to act in ways we normally condemn and nor should it rule out too many acts we usually assume to be required.


The fourth section of the chapter then proceeds to defend the "Consent Principle" from the charge that it is superfluous. In replying to this objection Parfit argues that there are two general aims a moral principle can have, firstly, that they provide a reliable criterion of wrongness by showing that acts of a certain kind are wrong and, secondly, they can be explanatory, describing one of the reasons why wrong acts can be said to be wrong. After enumerating these points Parfit goes on to claim that the Consent Principle, if correct, offers more than a reliable criterion of wrongness (and is thus explanatory). 


The fifth section of the chapter moves on to discussing "actual" consent pointing out, by using the example of rape, that reference to it is often crucial for us in determining the wrongness of acts. This section is surprisingly long and after agreeing with the need to include reference, in some sense, to "actual" consent, Parfit moves on to look at the objection that it is possible that the Consent Principle concedes too much to actual consent. In considering this objection Parfit is, in a sense, reprising the objections to Korsgaard and O'Neill with which he began his argument in the chapter except that now he saddles them with what he terms a "Veto Principle" as opposed to the absolute objection to coercion and deception he opened by presenting them as having. Just as the "Choice-giving" principle that led to such absolute positions was earlier exposed as false so now is this "veto principle" similarly rejected and it is argued that the Consent Principle need not imply the Veto principle which would give others the automatic right to object to anything to which they either do not or would in fact refuse consent. 


It is useful, despite the repetition it involves, that Parfit does discuss the "Veto Principle". It is useful because, until this point, he has not clarified sufficiently his earlier statement to the effect that coercion is not, despite appearances to the contrary, necessarily wrong. In amplifying now Parfit discusses the notion of "irreversible" consent. It is frequently not possible to give irreversible consent but we can give such consent to things that we might later regret having consented to without it being irrational that we have nonetheless given this "irreversible" consent. So, for example, the inevitable pain that might come from certain kinds of operation might well, during the experience of its being undergone, lead us to regret having consented to the operation without this making it irrational that we in fact gave "irreversible" consent earlier.


Parfit's next move is to replace the "Veto Principle" with appeal to what he terms the "Rights Principle" which instead claims that everyone has rights not to be treated in certain ways without their actual consent. So there are certain kinds of act that would be veto-covered as the earlier principle intended but which kinds of act might well be difficult to determine given the rebuttability of many claims. However Parfit does argue that the opportunity to refuse consent does arise from the Rights Principle.  But there is a restriction on the application of this Rights Principle since the opportunity to refuse consent "must be given by people who have sufficient understanding of the relevant facts". So it does not apply to infants, the mentally ill or those under the influence of seriously distorting drugs (including being drunk, a rather problematic exclusion I think). Further Parfit also attenuates the application of the principle by stating that influences distort judgments in various degrees with the result that decisions made under some types of influence may not be entirely over-ruled but can be given less weight (a provision that would require much care).


These points also lead to Parfit bringing in temporal considerations for the first time since he argues that present consent matters more than past consent which, in its turn, matter more than retroactive endorsement. The basis of this present bias is, however, simply grounded on the reference of present beliefs to acceptance of truth since we act on the assumption at present that the beliefs we have now are true. 


The penultimate section of the chapter concerns deontic beliefs which are introduced in order to discuss the sense that wrong-making characteristics do not only arise from reference to consent. Some acts, in principle, could be wrong even though there was general rational consent to them suggests Parfit (which partly defuses the earlier sense that quite a bit of normativity was built in to his sense of "reason"). Included here, for Parfit, would be voluntary euthanasia, cruelty to animals, and, potentially, suicide (interestingly, the last of these is the first example Kant gives of application of the Formula of Humanity). So we have beliefs about wrong-making characteristics being involved with certain types of act regardless of reference to consent in relation to them. The beliefs that are so held are described by Parfit as 'deontic' reasons. Having said this, it is less than obvious that this section provides a way of dealing with such claims seriously.


The final section of the chapter looks at extreme demands that might be thought to arise from the Consent Principle which repeats the problem about the intuitive acceptability of principles that Parfit made earlier in the chapter. Here Parfit considers a revision to the Consent Principle that might be thought to be needed to prevent it from demanding too much of us which brings in reference to not requiring that we bear too great a burden (though Parfit does not specify clearly enough how to understand what "too great" would be). Finally, Parfit is aware of the fact that the Formula of Humanity has only been partially treated in this chapter and indicates the need to treat, in the following chapter, the reference to not treating others merely as a means. However, whilst this recognition is good, it is odd to have Kant named again at the end of the chapter when he has been missing from it for a considerable number of sections and when the focus on consent in general has not been systematically justified as a correct response to his general argument in terms of the discussion of humanity. It is, after all, only with regard to one example out of four when discussing the formula of humanity that Kant even refers to consent. So the suggestion that "half" the sense of the formula has been caught in the chapter is certainly peculiar.

Parfit, Humanity and Consent (III)

In my last posting I looked at the additions Parfit made to his 2002 analysis of the implied reference to "consent" in the Formula of Humanity in his first draft of On What Matters which was entitled Climbing the Mountain. In his subsequent 2008 draft of On What Matters, the last full version before the publication of the work earlier this year, Parfit returns to the topic in Chapter 8 of his manuscript.


The Chapter opens again with the usual discussion of the problem with how Korsgaard and O'Neill are said to take the Formula of Humanity. However, when he reaches the account of "rational consent" Parfit this time refers to a point not included in either of the two previous treatments. This concerns  a problem with thinking that referring to shared ends will suffice to make rational consent sufficiently inclusive. Here Parfit points out that whilst there can be an agreement on ends this does not necessarily translate into an agreement concerning the appropriate means for attaining these ends. 


Subsequently a reference to Rawls appears which cites him as interpreting the consent principle as meaning that we consent rationally to some act if and only if or "just when" we could will it to be true that the agent's maxim is a universal law. This requires invocation of the Formula of Universal Law.  The reason why Rawls refers to this is due to Kant's general claim that all the formulas mean the same thing or are "statements of the same law". Rawls assumes that this means that Kant cannot have added something to the content of the law when he states one formula rather than another. But Parfit does not accept this view and assumes, rather, that there is something in the Formula of Humanity that is not included in the Formula of Universal Law. In making this assumption Parfit is following the precedent of, for example, Allen Wood, who, likewise, assumes that the Formula of Humanity has importantly different implications than the Formula of Universal Law. This point is not a small one since the discussion of the relationship between formulas of the categorical imperative has been a major source of disputes between interpreters of Kant. Unfortunately, whilst Rawls' reason for assuming that the Formula of Humanity is not significantly different to that of Universal Law, seems rather simplistic, it is hardly helpful of Parfit to simply stipulate that he does not accept this view without arguing on both philosophical and textual grounds for taking the formulas to be different. Such an argument does require, further, some discussion of what Kant means by claiming that there is no new "content" added in any of the formulas.


Parfit also adds in this draft a claim to the effect that rational consent has to be "informed" consent, a proviso not previously made clear. The Consent Principle now becomes:

"It is wrong to treat people in any way to which they could not rationally consent in the act-affecting sense, if these people knew the relevant facts, and we gave them the power to choose how we treat them."

The notion of "treating" is also to be understood in a sufficiently broad way so as to include, for example, breaking promises to the dead. The notion that is really involved for the consent to which the principle refers is "sufficient reason" in the sense that we consent to that which we do not have sufficient reason to refuse to consent. This implies a shorter formula of the Consent Principle which Parfit also gives but which I'll leave aside here. 

There are clear constraints upon the Consent Principle since it should be both plausible in itself and have plausible implications. These constraints are clear concessions on Parfit's part to intuitive or common-sense conceptions of what morality requires and forbids and echo the concession made in Climbing the Mountain to the point that beneficence that makes oneself in need of beneficence is misdirected. These points are subsequently mobilised by Parfit to suggest that the Consent Principle cannot be integrated with either egoistic or "subjective" (desire-based) views of the good. 

The appeal to "sufficient reason" is later finessed by Parfit into a view about "facts" that pertain in situations such that they are what make the beliefs concerning the rationality of consent plausible or otherwise. However, Parfit's subsequent consideration of examples leads to the same view as in Climbing the Mountain, namely that the Consent Principle may be too demanding (and hence fail to meet the constraint of having plausible implications). Further, it is not alone sufficient to describe what it is for something to be morally right since it turns out that it is possible to rationally consent (on Parfit's view) to things that are morally wrong. And, as in Climbing the Mountain, this turns out to be the ground on which Parfit moves to the discussion in the Formula of Humanity concerning treating others in such a way that they are not "merely used as means".

Parfit, Humanity and Consent (II)

In a recent posting I looked at Parfit's account of the formula of humanity in his original 2002 lectures, the "germinal core", so to speak, of On What Matters. What is apparent from the consideration given in 2002 is that the discussion of "rational consent" is separated out from the reference to not treating persons "merely as a means" in Kant's formula. This separation of the reference to "rational consent" from the discussion of what is involved in not treating someone "merely as a means" in 2002 continues in Parfit's subsequent drafts of what eventually became On What Matters.


The second version of Parfit's work is the manuscript that goes under the name Climbing the Mountain which is available courtesy of Pea Soup. In Chapter 4 of this work we find a correlate of what was the first lecture of 2002 and it opens in a very similar way discussing, as it does, the Formula of Humanity and drawing out the same problems with the readings of Korsgaard and O'Neill that were already stated in 2002. Similarly, examples are appealed to that are meant to show problems with desire-based "subjective" views of reasons. Finally, the rape example is again used to bring out problems with thinking of consent only in terms of "possibility" and to show the need for some sense of "rational" consent.


New material begins to be discussed in Climbing the Mountain when Parfit turns to "deontic beliefs" concerning other reasons than consent for finding something to be wrong. Considering independent reasons for finding something to be wrong Parfit now builds into the "consent principle" a sense that we cannot consent to something that we have other reasons to find wrong. However, after making this point, Parfit's discussion takes an unexpected turn as the kinds of examples he goes on to consider lead in the direction of considerations touching on beneficence. This leads, in a now fairly traditional style to indications of why I should sacrifice things that would satisfy myself in order to morally help others. 


In support of what may appear here to be consequentialist question-begging, however, Parfit can cite Kant's remarks concerning how the ability to be beneficent depends itself on situations that are unjust and on Kant's claim that it is possible to "participate in the general injustice even if one does no injustice". However, in the passage from the Doctrine of Virtue where the most extensive account of beneficence is given Kant also allows that practicing this virtue at one's own expense is something that is not permissible since it would merely create a new case of someone needing beneficence, something that Parfit concedes and uses to temper his account of the duties here involved. In conceding this Parfit indicates that acts that would be morally permissible are not necessarily morally required.


What Parfit does not sufficiently explain here is how he has been led from "rational consent" to these questions concerning beneficence. The bridge appears to be that others could and would consent to my doing acts that would decisively aid them even if this required me to engage in considerable sacrifice. So the link seems to be one in which the consent of others appears as a kind of demand upon me by virtue of what it makes permissible for me. It is not so obvious however that a concern of this sort was at the heart of the false promising case that led Parfit, along with others, to assimilate the Formula of Humanity to a principle of rational consent.


Further, since some acts are wrong despite our being able to consent to them, it follows that the Formula of Humanity has not been well captured in being assimilated to the notion of "rational consent". This concludes the additional material of Climbing the Mountain on this topic.

Parfit, Humanity and Consent

The second part of the first volume of  On What Matters is derived from the earliest core of the project since the discussions here invoke the Tanner Lectures Parfit gave in 2002. As a consequence, I think it is useful to open a discussion of the specific ways in which Parfit mobilises arguments at different stages of the composition of On What Matters. In this posting I'm going to begin looking at the first of the original Tanner Lectures which is headlined "Rational Consent" and whose purpose is mirrored in Chapters 8-11 of On What Matters.


A central reason why this discussion is of particular interest to readers of this blog is that Parfit opens the first Tanner Lecture with a statement of Kant's formula of humanity and the subsequent discussion concerns different types of interpretation of it. However, what is picked out by Parfit's discussion is not something specifically stated by Kant in the formula but an implication that is drawn from it. This implication concerns an important sense in which we could be said to treat people as ends-in-themselves.This is that we should act towards them in such a way that our conduct is one that could meet their possible consent.


The immediate rationale for Parfit's passing from the formula of humanity to this question of possible consent concerns the way the formula of humanity is applied to the four examples Kant considers in the second part of the Groundwork. In ruling out lying promises Kant refers to the fact that if I make a lying promise I am treating someone merely as a means and not also as ends in themselves because "the one I want to use for my purposes by such a promise cannot possibly agree to my way of proceeding with him and thus himself contain the end of this action" (Ak. 4: 429-30). It is due to the fact that the one to whom the lying promise would be made could not consent to this promise as it would conflict with ends they had themselves that it is clearly wrong to make such a promise.


After using this example to make the transition from the formula of humanity to a question about possible consent Parfit subsequently cites some contemporary Kantian views about the rationale for Kant making the kind of claim he does about the lying promise case. The first cited is Christine Korsgaard. Korsgaard, in her paper on lying, included in Creating the Kingdom of Ends points to the strength of Kant's argument in the lying promises case since he stresses that there is something "impossible" about the one affected being able to consent with the purposes I would have in view in making the lying promise. Korsgaard turns to an evaluation of what it is in the situation that makes it "impossible" for the one affected to consent and she stresses in doing so that the one affected has "no chance" to consent. What she appears to mean by this is that the affected party has no way of consenting as they do not know what it is that they are being asked to consent to. This leads her to say: "knowledge of what is going on and some power over the proceedings are the conditions of possible assent; without these, the concept of assent does not apply".


Parfit also cites Onora O'Neill as adopting a view of a similar sort since she also says that deception rules out genuine consent though, in the citation from her Parfit gives, less is here indicated as to why this would be. Finally, Parfit derives from Korsgaard a comparison of deception with coercion as fundamental moral evils, clearly due to the suggestion she has made that deception is, in a sense, a form of coercion as it manipulates the other for one's own ends.


When the full argument is drawn out in this way, however, then it certainly appears, as Parfit indicates, as if something must have gone wrong somewhere since the mere actual absence of consent is not evidently sufficient to make an action wrong. So operating on someone without their consent (as they are too ill to give it) is not obviously wrong and, similarly, I can make a decision for someone who I am unable to contact that may well not be wrong simply because they have failed to consent to my making it. So the argument concerns not such actual consent but only some form of conceivable or possible consent.


Now, once the ground has moved to possible consent we seem to be getting to the rationale of Kant's argument but Parfit introduces now a complication by means of the thought experiment he calls "deadly knowledge". In this thought experiment I know someone is a murderer but, if I don't lie to you about this, your knowledge that they are the murderer will leave you vulnerable to also being murdered since you know this person and would not be able to conceal your knowledge of their status. Parfit suggests that this case is importantly analogous to the ones already considered where we did not take actual consent to be the key to characterising the moral quality of the action.


Not only does Parfit make this case but he indicates that the situation is importantly similar with regard to coercion as I can agree to be coerced to do something that I regard as good for me. So, for example, I might have to be tied down in order for some painful operation to be carried out as otherwise the pain would lead me to try to prevent the operation being carried out although consciously I am aware that it is better it is carried out. So, in a sense, I can adopt the end of being coerced here (can consent to it). In fact, although Parfit does not note this, arguments of similar form are important for Kant's philosophy of right.


So if the case of lying promises indicates a commitment on Kant's part to a view about possible consent it has to be understood in a way distinct from how Korsgaard and O'Neill have presented it. On their view the importance of consent turns on a possibility of being able to give it. This commits them to an act-affecting conception of consent which Parfit terms the "Choice-giving principle" and which he formulates as stating: "it is wrong to deny people the opportunity to choose how we treat them". Since this principle does not give us any way of dealing with cases where we are unable to communicate with others, however, Parfit refines it further so it becomes the "Veto principle" and refers then not merely to whether they could consent but whether they would if they had the opportunity.


When they are fully stated, however, as Parfit, points out, these principles in fact cover more ground than Korsgaard's statements might have led us to expect. This is so because they rule out more than just deception and coercion since when we don't tell people what it is we are going to do, simply don't tell them, we are not, at least not directly, deceiving them. Similarly, when we act without their consent, we need not be coercing them.


After making these points Parfit turns to his real target, which is to suggest the basis for viewing the interpretation of Korsgaard and O'Neill as faulty. With regard to the "Choice-giving principle" Parfit has an easier task since, if a publisher sends a book proposal to me for consideration and I recommend rejecting it, I act in a way to which you would not consent but this is insufficient to make the act I have performed wrong. This shows the weakness of the "choice-giving principle", however, since Kant was surely speaking of rational conceivability in quite a different way to such a case and that the "Choice-giving principle" can be undermined by it indicates well that it does not capture Kant's thought.


However the statement of the principles that arises from Korsgaard and O'Neill has another difficulty which is that it applies only to cases where there is a situation between two parties and not many. In the latter case I often cannot relate to the consent of all the parties in question and yet this is surely not sufficient to mark out any action performed as therefore wrong? 


It is at this point that Parfit indicates an alternative way of viewing Kant's argument to that which appears to arise from the construal of it given by Korsgaard and O'Neill. The point about the argument they have given is it points to the wrong kinds of reason why someone cannot be said to be involved in consent. It is not, as Korsgaard puts it, a question about having the opportunity to consent (which produces the "Choice-giving principle"). It is, rather, whether there are "decisive reasons" for refusing possible consent. That is why Parfit views Kant's claim as concerning a "rational consent principle" and indicating what is wrong is action to which there could be no rational consent. The lying promise case does appear to support the reference to opportunity to consent but it is only part of Kant's consideration in the Groundwork as, immediately after mentioning it, he refers also to attacks on the freedom and property of others. Such cases do not involve denial of opportunity to consent though they are cases of coercion.


Parfit also stresses the point that even the lying promise case refers to the one affected being able to contain in themselves the end of my action. This reference is assumed by him to indicate a need to discuss not merely conceivable consent but also rationally conceivable assent. However, whilst this rational consent view is certainly an improvement on the "Choice-giving" view it remains problematic since the question of how to understand rational consent has yet to be specified and one common way of viewing it would be that we could not rationally consent to anything that would be "bad" for us. However, again, when faced with multiple persons, it can be right to act in favour of one over another without having done anything wrong and even the one adversely affected by this cannot say that the simple application of this rule has wronged them. So they could give rational consent in principle to operating by a rule that had adversely affected them in application.


At this point, however, Parfit's discussion falls back upon the intentional view of reasons that was elaborated in the first part of On What Matters since he refers now to "facts that give us reasons". Not only is this true, however, but he also indicates that even were we to share this view that we still have difficulties since there are many kinds of "facts" in a situation and the relative weighting to give them is often difficult to determine. 


Parfit subsequently goes on to refine the rational consent proposal further to build in what he terms "unconditional" consent which would be something not subject to revision later. This is partly intended to rule out rape on the grounds that the rapist could state that the one raped could rationally consent to the act even if they don't "actually" consent to it! Parfit is indicating with the unconditionality clause that the absence of actual consent here is not something that can be hypothetically substituted for in the example in question.The overall effect of Parfit's argument is to suggest that the examination of the rightness of an action turns on its rationality not on its presumptive possibility of being vetoed.