This constitutes the fourth of 11 “meetings” to discuss Parfit’s new manuscript, Climbing the Mountain. First, though, a reminder announcement: all references will be to the June 7th version of the manuscript, and specifically to the PDF version of that manuscript. There are some page discrepancies, it turns out, with the Word version.
OK, now on to this week’s chapter, “Merely as a Means” (Chapter 5). In this chapter, Parfit is, to my mind, exploring whether or not to add a “Mere Means” principle to the list of defensible moral principles he began in chapter 4 with the Consent Principle. Here the idea is to see if we can come up with a relevant and plausibly defensible version of the Kantian principle that it’s wrong to treat any rational being merely as a means. As far as I understand him, the answer is no, at least when it comes to the morality of actions. I here quote from Parfit’s own abstract, adding my own comments in italics where appropriate:
We treat someone in this way [as a mere means] when we both use this person and regard her as a mere tool, whom we would treat in whatever way would best achieve our aims. On a stronger version of Kant’s claim, it is wrong to treat people merely as a means, or to come close to doing that. [This last clause is needed in order to handle the objection from Kamm that the principle is too weak, allowing, for instance, that were I to give just the slightest weight to my slaves’ well-being, I wouldn’t count as treating them as mere means. So the thought is that, in order to beef up the requirement, it must be the case that I don’t give too little weight to someone’s well-being or moral claims.]
We do not treat someone merely as a means, nor are we close to doing that, if either (1) our treatment of this person is governed in sufficiently important ways by some relevant moral belief, or (2) we do or would relevantly choose to bear some great burden for this person’s sake. [I cannot see any positive argument for this claim. If it’s supposed to be the contradictory of the gloss above on treating someone merely as a means, I fail to see how being disposed to bear some great burden for this person’s sake is the contradictory of treating this person in whatever way would best achieve our aims. I may be not disposed to bear such a burden, yet not be treating you in any way, let alone in any way that would best achieve my aims.] Even if we are treating someone merely as a means, we may not be acting wrongly. Consider some Egoist, whose only aim is to benefit himself. When this man keeps his promises, pays his debts, and saves some drowning child in the hope of getting some reward, he may be treating other people merely as a means. But these acts would not be wrong. Kant’s claim could be qualified, so that it would not mistakenly condemn such acts. On this doubly revised claim, it is wrong to treat anyone merely as a means, or to come close to doing that, if our act also harms this person. [Here Parfit is, for one thing, disambiguating two senses of “treating someone merely as a means.” On the one hand, one might regard someone as a mere means; on the other hand, one might, in acting, treat someone as a mere means. The former may not imply that we act wrongly. A gangster who regards most people as mere means, and so who is disposed, say, to injuring them whenever it would suit him, may, in purchasing coffee, regard the vendor as being no more than a vending machine. Nevertheless, he does not act wrongly when purchasing coffee from him.]
Suppose that some driverless run-away train is headed for a tunnel, in which it would kill five people. Our only way to save these people’s lives is to cause someone else, without her consent, to fall onto the track, thereby killing this person but stopping the train. It may seem that, if we acted in this way, we would be treating this person merely as a means. But that may not be true. And this person could rationally consent to being treated in this way. Though such an act may be wrong, it would not be condemned by either the Consent Principle or the Mere Means Principle. [This is the version of the Trolley Problem Parfit calls Bridge, and, consistent with my worries about last week’s claim regarding the rationality of sacrifice for the sake of strangers, I have a very hard time seeing how White could rationally consent to my dropping her, without her knowing that I was going to do it, from a bridge in order to block a train that will kill five others. Parfit’s reasoning here is that, since White could rationally consent to jumping in front of a train to save the five, she could also rationally consent to my pushing her. But there are two crucial disanalogies between the cases. First, in rationally consenting to jumping, she would be consenting to do so in full knowledge of what she’s doing. In rationally consenting to my pushing her, she’d have to be consenting to my doing so in a case where she had no idea what’s happening to her. The conditions of consent are importantly different, then, and I fail to see the rationality of consent in the latter case (further, this case may be particular vulnerable to the conditional objections Robert was running last week). Second, in the former case, she initiates the action; in the latter, she doesn’t, and this difference could be quite significant in her deliberations regarding consent. In other words, in the former case, she saves the five; in the latter case, her body is used to save the five. She could (perhaps!) rationally consent, then, to the former, but any consent she provided with respect to the latter would be (a) odd, if not irrational, and (b) based on different reasons from what might ground her consent in the former case.]
It is widely assumed that, 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. This view involves three mistakes. When we harm people as a means, we may not be treating these people as a means. [If I break your leg as a means to my escape from your attack, I don’t treat you, thereby, as a means.] Even if we are treating these people as a means, we may not be treating them merely as a means. [If I save my child’s life by using your body in such a way that one of your toes is crushed, I do not treat you merely as a means if I believe it would be wrong of me to use your body again (and cause another of your toes to be crushed) in order to save my own life.] And, even if we are treating them merely as a means, we may not be acting wrongly. [See below.] Some people give other accounts of what is involved in treating people merely as a means. These accounts seem to be either mistaken, or unhelpful. [Here there’s a nice little treatise on methodology and concepts, the upshot being that in trying to render Kant able to handle all relevant cases (and thus stretching the definition of “treating as a means” to the breaking point), theorists have done us a real disservice. In order to render Kant relevant, we do better either to simply revise his principles where they give the wrong answer or to recognize that some particular principle is a merely sufficient account of wrongness.]
If it would be wrong to impose certain harms on people as a means of achieving certain good aims, these acts would be wrong even if we were not treating these people merely as a means. [If it would be wrong of me to save the five by killing White, the wrongness of that act would not consist in my treating White merely as a means.] And, if it would not be wrong to impose certain lesser harms on people as a means of achieving these good aims, these acts would not be wrong even if we were treating these people merely as a means. [Here he has in a mind the case of a gangster who saves his child’s leg by bruising Black’s leg, without her consent.] Though it is wrong to regard anyone merely as a means, the wrongness of our acts never or hardly ever depends on whether we are treating people merely as a means.
There are two important projects in this chapter. First, Parfit provides a pellucid analysis of the “Mere Means” principle, constructing the most sympathetic possible version of it. As it turns out, though, the most plausible version includes a clause about not harming people, such that it would be wrong to treat anyone merely as a means (or come close to doing that) if our act will also be likely to harm this person. But by adding this clause about harm, we’ve rendered treating someone as a mere means insufficient for an act’s wrongness. But then the second project is to show that even the harm-appended Mere Means principle is neither a necessary nor a sufficient condition of moral wrongness in action. What we’re left with, then, is still just the Consent Principle from the previous chapter, which itself remains insufficient to account for all moral wrongness (in action).
I have just a few questions (much of the reasoning seems unassailable to me). First, is this the right interpretation of what’s going on in the chapter? There isn’t any real introduction, or an overall roadmap of any kind, to let us know how this chapter fits into the overall dialectic or what, precisely, the motivation and point is. Just one sentence to indicate all that would have been enormously helpful.
Second, Parfit is obviously trying to build up principles for what makes an action wrong (or right), and the Mere Means principle (even with the harm clause) fails to be relevant on that score, but he also makes provocative remarks about character and moral worth as he goes. At one point, for example, he claims that regarding someone as a mere means is wrong. But wrong in what sense? Rationally? Surely not. Prudentially? Again, surely not. Morally? But on what basis? Suppose I’m the gangster, regarding virtually everyone I know as a mere means, but I wind up being unfailingly polite in all my actions to them (simply because it would be too much trouble for me to be otherwise). Surely I’m a bad person, but where’s the wrongness in that? This is a question both about the scope of morality and the grounds (and coherence) of non-act wrongness on Parfit’s account.
Third, there are the questions I raised in the exposition above, in particular the following: White’s rationally consenting to jump in front of the train does not imply the rationality of White’s consenting that I push her (or, in Parfit’s case, drop her) in front of the train without her knowledge. Are there then other reasons to think of the permissibility of dropping her as justified by the Consent Principle?
Dave,
I’m not clear on some of the things you said:
(1) You write,
Fair enough. But I’m not sure I understand what your objection is. Being disposed to bear some great burden for this person’s sake is the contrary of regarding that person as a mere tool. As Parfit writes, “We treat someone in this way [as a mere means] when we both use this person and regard her as a mere tool, whom we would treat in whatever way would best achieve our aims [emphasis added].” Thus, if you are disposed to bear some great burden for this person’s sake, you can’t be treating her as a mere means, nor are you even close to treating her as a mere means since the burden that you’re willing to bear is a great one. I don’t see what the problem is.
(2) Regarding the example of the egoist who acts rightly despite regarding others as mere tools, you write,
I didn’t think that there were two senses of treating someone merely as a means, although there are two conditions for treating someone merely as a means: you have to both use the person and regard the person as a mere tool. In any case, I thought the point of this example wasn’t to disambiguate anything but just to point out, as you do, that one can treat someone as a mere means and still have acted rightly, as the Egoist does in saving the drowning child in the hopes of getting some reward.
(3) You write,
But whether one could rationally consent without knowing the relevant facts isn’t the relevant issue. I take it that the relevant issue is whether White could rationally consent to my dropping her if she knew the relevant facts. When Parfit formulated the Consent Principle, the issue was always whether the person could rationally consent if she knew the relevant facts. Why do you think that whether one could rationally consent without knowing the relevant facts is the relevant issue here?
(4) You write,
I take it that the someone is someone with value, and, in that case, in regarding that someone as a mere means one would be failing to respond to reasons that one has to take that person’s welfare into account. So it’s wrong in that one has decisive moral reasons to take that person’s welfare into account and yet one is failing to respond to those decisive moral reasons.
I hope that you’ll have time to respond. I enjoyed reading your post and did think that you did a nice job of explaining what the whole point of the chapter was, which I think that you got absolutely right.
Doug: One at at time….
Regarding (1), suppose I have a really good hammer. I use it all the time, and I really don’t want anything bad to happen to it. I realize on my way to an important meeting that I left it outside as it was about to start raining. I miss the meeting so I can rush back to the house and save the hammer from rusting.
Not only is the willingness to sacrifice greatly for the sake of X not the contradictory of regarding X as a tool, it’s not even the contrary of it.
Regarding (2), you’re probably right, although there’s some confusion given Parfit’s claims about the wrongness of merely regarding someone as a means (even if you don’t act on that disregard). Given what he says, it seems he has in mind two senses, one of regard and one of treatment in action.
Regarding (3), I was simply trying to point out a dissimilarity Parfit doesn’t acknowledge between the “I consent to jumping in front of the train” and the “I consent to your pushing me in front of the train” cases. He claims that *because* I’d rationally consent to jumping, I’d also rationally consent to being pushed. But there are relevant differences between the two cases, it seems to me. One is that in the latter case I wouldn’t know what was happening to me; this strikes me as a relevant dissimilarity, one that conceivably undermines rational consent. The second difference I pointed out was that in the former case, *I* save the five, whereas in the second, it’s merely *my body* that saves them. Again, this strikes me as a relevant dissimilarity, one that conceivably undermines rational consent in the second case, even if you could get it in the first.
Dave,
You would really do that for the hammer’s sake? How kinky.
I, however, would have thought that a hammer doesn’t have a sake in the relevant sense — the sense of having a good. I might save my hammer from rusting, but I wouldn’t do it for its sake; I would do it for my sake. Being rusted is not bad for the hammer; it’s bad for me.
Regarding (4): how is failing to take your welfare into account wrong, though? After all, that’s precisely what the gangster does in buying his coffee, but Parfit asserts outright that he *does nothing wrong* in that case. So what precisely is the wrongness of merely regarding others as mere means if that never plays a role in one’s actions? And what makes this a case of *moral* wrongness?
Regarding now your first reply, of course hammers have a good (and thus a sake), derived from their function. One doesn’t want a bad hammer now, does one?
Dave,
Regarding (3), okay, as it is, I suppose that the inference is questionable. But I think that Parfit could have just as easily argued as follows. It seems that I could now rationally consent to your pushing me off the bridge if I were ever to end up in a situation where your doing so will save five lives. There is no disanalogy between this case and Bridge – in both cases, I get pushed off pushed of the bridge by you without knowing why I’m being pushed. Therefore, I could rationally consent to your pushing me off the bridge even in the case where I didn’t give you my prior consent.
So do you think that Parfit is wrong about its being possible for White to rationally consent in Bridge, or are you merely pointing out that his argument for this claim is less than perfect?
In the hammer case, I think that you’re trading on an ambiguity. In any case, I think that it’s clear that what Parfit has in mind when he talks about being disposed to bear some great burden “for this person’s sake,” he means “for the sake of this person’s welfare.” And a hammer, as I’m sure you’ll agree, doesn’t have a welfare — at least, not in the sense that concerns moral philosophers, not in the sense in which utility is supposed to be a measure of it.
Regarding (3), I was merely suggesting that the argument was less than perfect, not that there might not be another way to reach the same conclusion. As you know, I’m not crazy about the rationality of consenting to sacrifice one’s life for the sake of five strangers, but that’s neither here nor there with respect to this particular point.
You mean the following plea from one spouse to another would be incoherent? “Honey, you’ve got to stop getting drunk and sleeping naked with your tools in the front yard during the rain. If you won’t do it for me or the kids, at least do it for the sake of the hammer!”
At any rate, even restricting the case to things with welfare, I’m still not sure I see the (contrary) relation between being disposed to sacrifice something for your sake and regarding you merely as a tool. Although perhaps I merely have cases in mind where I benefit a “tool” (causing an increase in its welfare) where the benefit wasn’t produced *for its sake*.
Dave,
Parfit’s says that “we regard this person as a mere instrument or tool” when this person is “someone whose well-being and moral claims we ignore, and whom we would treat in whatever way would best achieve our aims.” So your being disposed to sacrifice what would best achieve your aims for the sake of this person’s well-being is quite clearly the logical contrary of your regarding this person as a mere instrument or tool.
Dave,
You write,
It sounds to me like you think that the only objects open to moral assessment are acts and agents, not desires or any other kind of attitude. Is that your view? You don’t think that it is wrong to take pleasure in the suffering of innocents? You don’t think that it’s wrong to hate someone because of his or her race? You don’t think that it’s wrong to be indifferent to the welfare of one’s child? You don’t think that it’s wrong to want all people of a certain race, creed, or religion to die?
If desires and beliefs can be irrational, I don’t see why desires and other attitudes can’t be immoral. Now, you may object that such attitudes are non-voluntary, and thus not open to moral assessment. But surely aspects of our character that are non-voluntary are open to moral assessment. Besides, I think that Parfit is right when he says, “These non-voluntary responses [that of automatically believing or desiring something] are part of what makes us rational. When we are aware of facts that give us conclusive or decisive reasons to have some belief or desire, it may be impossible for us, if we are rational, not to respond to these reasons by having this belief or desire.” And it seems to me that we can say the same thing about moral: Part of what makes me moral is that I don’t want people to suffer and that I don’t regard them as mere tools.
Oh, c’mon, of course I think it’s wrong to regard others in the ways you cite. I’m just trying to figure out what the *nature* of that wrongness is *on Parfit’s account*. As I said earlier, this is a question about the scope of morality on Parfit’s view. I can’t yet figure out if he’s talking just about actions, just about a certain range of actions (those things we owe to one another), or something much broader (to include not only character but various attitudes and desires as morally assessable). And if merely regarding someone as a means is wrong, on what basis might it be so (or is regarding someone as a means just wrong insofar as it violates a principle not to so regard people?)?
As for the logical contrary stuff, d’oh! I hadn’t recalled that gloss on it. You’re right, then (damn, that hurts even to type).
You ask, “if merely regarding someone as a means is wrong, on what basis might it be so?” I gave you a suggestion in my initial comment: What makes it wrong is that you fail to properly respond to the decisive moral (or impartial) reasons that you have to regard that someone as more than a mere tool. But in your response to my comment, you seemed to treat that as not answering your question. That’s why I thought that you were questioning whether attitudes could be wrong. I apologize for not understanding your point. I’m probably still not clear on what your question is. What do you mean by the “nature of that wrongness”?
Doug: You say, “What makes it wrong is that you fail to properly respond to the decisive moral (or impartial) reasons that you have to regard that someone as more than a mere tool. But in your response to my comment, you seemed to treat that as not answering your question.”
It might be a sketch of the *structure* of a response, but it doesn’t answer my question(s) just yet. First, why think it’s a failure to respond to *moral* reasons, rather than reasons of some other kind? Parfit has been focusing solely on the moral rightness or wrongness of *actions* thus far; what reason do we have to think that attitudes are *morally* right or wrong as well (on his account)? Second, if it is a matter of moral wrongness to regard someone as a mere means, what’s the specific grounds for that wrongness? In other words, the only moral principle generated thus far has been the Consent Principle. But surely my merely regarding someone as a mere means doesn’t violate *that* principle — what reason would I have to not consent to your regarding me as a mere means, just as long as you always wound up treating me well?
So again, I’m just curious here about the scope of morality on Parfit’s account, as well as the possible grounds for the rightness/wrongness of non-action states.
I’m not sure if this is helpful, but I’ll just state what seems to be an aspect of Kant’s use of a test like this that seems left out of this discussion, and that is the connection between actions and attitudes. On my reading of Kant he assesses maxims (which I think of as statements of the point of acting in the way one is and a description of the act in the terms that the agent would recognize as her conception of what she is doing) first and then act-types by asking whether there can be permissible maxims which would motivate a person to do actions of that type. At least that seems to be how he assesses whether actions are universalizable.
It seems like this is the sort of thing he should want for the not treating as a mere means formulation as well. So my thought is that the idea Kant is after is that treating someone as a mere means is a function of one’s motive and actual conception of what one is doing in acting.
It seems to me that the nexus between the agent’s motives and way of thinking about the person and their action is missing from the discussion so far. I’m not sure how far it helps. (I’m not a great fan of Kant’s ethics so I doubt that in the end the whole idea works, but I do think that the not using as a means idea is the best element in Kant’s moral philosophy so it seems worth trying to make it work as far as it goes.) But it might help with the worry about someone who would, in some other instance, place some limits on how they would treat the person. For example, according to one sort of objection related to Kamm’s, someone would not be treating a person as a mere means if there were some limits on what one would do to a person, even if those limits don’t come into play in this case. But on my understanding of Kant’s idea, this fact would not tell us about the motive for this action here. And it would be the status of this motive which makes this action right or wrong.
Of course this is terribly abstract, and it would require some way of telling when a maxim is the sort which involves treating a person as a mere means and when it is not. And thinking about that raises a question. If you are thinking about whether the person could reasonably reject your treating them in this way and letting that determine whether you do what you do, aren’t you thereby not treating a person as a mere means? And if the answer is yes, how would this principle be and addition?
I hope at least some of this is coherent.
Couple of quick questions on Kamm. On the Kamm counterexample,
“This last clause is needed in order to handle the objection from Kamm that the principle is too weak, allowing, for instance, that were I to give just the slightest weight to my slaves’ well-being, I wouldn’t count as treating them as mere means”
But as everyone here knows, Kant always adds to “not merely a means” the phrase “but always also as a end in themselves”. I can’t see how Kamm’s objection is a counterexample to Kant’s claim unless she thinks that giving the slightest weight to her slave’s well-being amounts to treating the slave as an end-in himself. She has in mind some neo-Kantian view instead?
But putting that aside, I don’t see how putting some weight on the well-being of the slave could entail anything close to not treating the slave merely as a means. Obviously, if I give the slave enough consideration to keep him useful (feed him enough and so on) it hardly follows that I am not treating him merely as a means. Not to treat him merely as a means requires respecting the slave’s autonomy (slavish autonomy?), rather than giving some consideration to his well-being. But this is all pretty obvious, so I’m guessing I missed something.
Mike: The Kamm objection is simply to one proposal for what it *means* to treat someone merely as a means. If it simply meant regarding someone as a mere instrument, whose well-being and moral claims we ignore, and whom we would treat in whatever way would best achieve our aims, then if the Kantian view were simply “one must never treat people merely as a means,” that requirement would be too weak. But that commits her to no particular interpretation of Kant, so the worry that she overlooks the remainder of the mere means principle is moot.
Of course, Parfit goes on to suggest that, instead of revising Kant’s definition of “treating someone as a mere means,” we should just revise the principle, to read “It is wrong to treat anyone merely as a means, or to come close to doing that” (he adds a harm constraint later). Now this *does* involve leaving out the continuation of the principle in Kant, and so there must be some Kant interpretation going on here. But then it seems as if Parfit is simply taking “treating as an end-in-himself” as the contradictory of “treating merely as a means or close to it.” Is there a problem with doing that?
Mark: I’m not sure I follow everything you’ve said, but there are a couple of intriguing aspects to it. First, Parfit does discuss the role of attitudes, given that whether we are treating someone as a means depends on what we’re intentionally doing, as well as on our “attitudes or policies” (101). And “that is in part a matter of what we would have done, if the facts had been different” (101). So perhaps the counterfactual conditions you mention are dealt with after all.
Second, you ask a very interesting question: “If you are thinking about whether the person could reasonably reject your treating them in this way and letting that determine whether you do what you do, aren’t you thereby not treating a person as a mere means? And if the answer is yes, how would this principle be an addition?” If by “this principle,” you’re referring to the mere means principle, then I think you’ve actually advanced a different way of arguing for Parfit’s conclusion in this chapter, viz., that the mere means principle just *is* no interesting addition to our list of sufficient moral principles.
Thanks for the Kamm observation. I didn’t notice that she was doing that. On a different score, I don’t think Parfit’s version of “treating someone as an end-in-himself” is the contradictory of “treating someone merely as a means”. Parfit says this, right?
“We do not treat someone merely as a means, nor are we close to doing that, if either (1) our treatment of this person is governed in sufficiently important ways by some relevant moral belief”
But it can be true that (a) our treatment of someone is governed in sufficiently important ways by some relevant moral belief (and so by (1) our treatment is *Parfit-not-merely-as-means* treatment) and (b) we do not treat the person as an end-in-himself. Just take any case in which I violate your autonomy for especially good moral reasons. Here is a relevant moral belief: no one should drive whose judgment is even slightly impaired. I see you drink half a beer and, governed by my moral belief, I take your car keys.
Maybe you’d conclude that my belief is not sufficiently important. But there are lots of others that I might be used in violations of autonomy. I see that Smith is about to exercise his free speech rights in advancing his racist cause. I sabotage his alarm clock, so he wakes too late to give the speech or I sabotage the speaker system or something like this. I am guided by a principle that prohibits deeply offensive conduct. I think there is no question that I do not treat Smith as an end-in-himself, since I pretty clearly violate his autonomy. But I do so for important moral reasons, so my treatment is *Parfit-not-merely-as-means* (by (1) above). But then these are not contradictories.
David,
You caught my drift in both comments.
I don’t have anything additional to say about the second point right now, but on the point of the connection between counterfactuals and motives: I guess I think that some counter-factuals surely are relevant to motive, but that not every sort is. For example, if I trip you to amuse myself and impress my friends without regard for how you feel about it, it seems like that is the sort of thing that intuitively is using merely as a means. The fact that I would not do it if tripping you would kill you (say if you were at the edge of a cliff) and I knew that, does not seem to me to change the facts about my motives in the actual world. For I did not take into account how my actions impacted you (though I would have if there had been a salient and deadly impact). If on the other hand, I tripped you thinking that you didn’t mind (I think you like playing the clown), but I would not have had I known you minded, that does not any longer seem to me like treating merely as a means. I’m not sure how this cuts for the overall Kantian proposal, but I was hoping (in my temporary guise as a helper of Kant) it could be used to make it more restrictive about how to treat others.
Mike: Parfit’s discussion of this principle is sufficiently nuanced (fudged?) to avoid your worries, I think. What makes a moral belief *relevant*, for instance, is that the belief requires “direct concern for the well-being or moral claims of the person whom we are treating as a means” (101). What I actually think Parfit’s doing at this point is building in possible consent as a sufficient side constraint, and insofar as his consent constraint looks like your/Kant’s autonomy constraint, then *some* of the cases you’ve articulated would fail to pass the test.
There is plenty of gray here, though. As Parfit also notes, “Since relevance and importance are both matters of degree, it is often unclear whether [condition] (1) is true” (101).
Mark: This is very interesting. I wonder, though, if the relevant counterfactuals in the cases you cite aren’t simply about my knowledge of whether or not the target of my action would have *consented*?
David,
I don’t have a systematic account of which counterfactuals are relevant to determining when I am treating someone as a means. My background thought may be that if someone is sufficiently sensitive to certain facts then maybe that says something about their motives, but that is a stronger requirement than just that they would take those facts into account in certain counterfactual circumstances. My thoughts here are largely case driven, which means I may be vulnerable to being pushed around about this with cleverly constructed cases.
Whoops, I left out a ‘merely’ before “treating someone as a means,” above.
I’m joining the discussion a bit late, but I wanted say I think I agree with Mike on a problem with Parfit’s understanding of the Kantian principle in question. Parfit appears to think we might be able to understand what treating something as a mere means is without needing to understand what treating it as an end in itself is. That seems to me to be completely wrong. As I understand it, you treat something as a mere means when and only when you fail to treat it as an end in itself. And it is treating something as an end in itself that wears the pants, or is the main idea behind the Kantian principle. Here, we would need to bring in ideas such as giving the thing in question a certain sort of weight in deliberation, perhaps (ala Rawls) giving a kind of priority to it over other considerations (such as happiness or wellbeing), stuff like that. (The notion of an ‘end’ has a few facets to it, but set that aside for now.) That, at any rate, is how I understand the principle.
In any case, there’s a further connected problem I have with Parfit’s presentation and discussion of this principle. This is that he focusses on how we treat people, whereas the Kantian formula focusses, not on how we treat people, but on how we treat their ‘humanity’, which is an aspect of a person. ‘Humanity’ turns out to be something of a technical term, but it’s not too hard to see that Kant thinks it is our (supposed) capacity for directing our behavior on the basis of reasons that is the thing that must be treated as an end in itself. The point is important because it is the focus on this capacity that gives Kant’s principle any hope of being defended. In light of this, for instance, Parfit’s (B) looks like it misses the point. It looks like we should begin with the notion that how we treat people should be guided by what respect for their rational capacities would dictate. And that may be quite different from what they might, as they actually are, like us to do or tell us we should do. It might, for instance, require conforming to what they would agree to as principles governing conduct, were they better placed, rationally speaking, than they actually are.
This may just means that, in the end I will agree with Parfit that there is no extra ‘means’ principle; but that is because, properly understood, it would have to collapse back into the consent principle.
Robert — Just a heads up: Parfit’s Chapter 6 (“Respect and Value”) is very much about the issues you raise (as well as those raised by Mike), in particular trying to figure out the best interpretation of the Formula of Humanity, as well as precisely what “humanity” refers to. I won’t go any further than that now, given that the official post about Chapter 6 is coming our way tomorrow morning, but he seems well aware of your sorts of concerns.
Ah, I should have read ahead! It does make Ch. 5 seem otiose, however.
Dave is absolutely right. Parfit is clearly aware of the issues that Mike and Robert have raised and addresses them in the next chapter. As I see it, Parfit is just seeing if any Kantian inspired principle works whether it be (1) the Consent Principle: that it is wrong to act in any way to which anyone could not rationally consent, (2) the Mere Means Principle: that it’s wrong to treat any rational being merely as a means, or (3) the Formula of Humanity: that “every rational being…must always be regarded as an end…and is an object of respect.”
But tune in tomorrow (5 AM PST) for the post on chapter 6, which is all about Kant’s Formula of Humanity.