The experience machine objection as a wrong order problem

This is the first of a planned series of posts wherein I try to place attacks on utilitarianism within a more general framework I'm developing, the wrong order problem: properties that could work to aggregate all our morally relevant concerns, like happiness or desire satisfaction, are too high an order of property to count as good-making features, and can instead only indicate goodness.

There is a long-standing and enticing tradition of considering happiness as the most valuable good that a human life can attain. It can hardly be denied that a life containing happiness seems better than one without. There have been no shortage of people who try to build an ethical viewpoint from this foundation in one way or the other. One popular and influential approach is to consider happiness to be the only good, from which everything else draws its value, and therefore what we should do is to maximise the amount of happiness in the world. This is hedonism in a straight-forward and non-perjorative sense.

Against hedonism, Robert Nozick asks us to consider the possibility of an 'experience machine', to which a person could be connected and then live through exact and entirely convincing replicas of whatever experience they like. The question is whether these ersatz experiences, which perfectly recreate all the sensations you would have had if it really happened, make somebody as well-off as the real counterparts would. They would if hedonism is true. If that is the case, the experience machine would be a key to a perfect life, seeing as how you could undergo any experience you like, believing all the time that these things are really happening to you. But, as Nozick presses us to admit, plugging up to such a machine for life would be uncomfortably close to suicide: the pleasures might be real, but the events they relate to aren't, and by plugging into the experience machine you give up engaging with the world.

The problem that the experience machine objection latches on to is that happiness stands in the wrong relationship to the facts of someone's life to be identical with well-being. A judgement of how well someone's life is going is a judgement on the state of affairs that person finds themselves in. But happiness isn't a part of that state of affairs, it's what I call a *reflective stance* towards it, and like any reflection you can have towards something, it can be misled, as the experience machine tries to show. The problem is that any reflective stance property, like 'brings happiness', is a higher order property of what is being judged, the condition of someone's life, that is, it's the wrong order of property to count as what is being judged. There is a relationship between the two, but it's not an identity relationship. This means that we should consider reflective stance properties like happiness as part of the epistemology of good-making features, rather than the ontology. That is, happiness indicates a good life, it's not what makes a life good life. This doesn't mean we should discount happiness – it can be a good indication, perhaps even a perfect one (but were that the case, there must be something wrong with the experience machine case) – but it means that we must look for the good-making features elsewhere.

The start of my European adventure

I made my last post from Hong Kong, and this comes from Frankfurt, as I keep coming ever closer to Utrecht and the start of my study in Europe. My arrival at the Erasmus Mundus Masters of Applied Ethics programme has already been delayed by almost three weeks, as I have been subject to a number of delays. I even had a last round of drama at Auckland airport as I tried to depart. But all of that draws to an end, and I am ever so close to getting down to the real business. Somebody joked that I should use my (copious) time travelling from Auckland to Utrecht to write a book about all the trials that beset me as I tried to get here, instead I wrote a technical minded-little piece on meta-ethics (the previous post) and read Bernard Williams's Morality, An Introduction to Ethics twice. This gives you some indication of what type of person I am, a request somebody else made of me and my blog.

The Euthyphro problem as a wrong order problem

Lately I have given a lot of thought to and become tremendously impressed by the Euthyphro problem, presented by Plato in the dialogue of that name, which makes life distinctly uncomfortable for a particular kind of moral view. The Euthyphro problem drives us to distinguish what makes something good from how we know that it is good. I want to suggest that this points to a general problem, to do with what type of property could count as a good-making feature. I think problems of the Euthyphro form give reasons to believe that many higher order properties of the targets of a moral theory. I have in mind what I call 'reflective stance properties', like attitudes toward something, some function you have in mind for something, and so on. I want to suggest that they can only indicate goodness, not be good-making features, and the Euthyphro problem is a paradigmatic example.

The Euthyphro problem simply is a knockout objection against divine command theory, and can be usefully introduced as such. Divine command theory claims that everything that is ordained by the divine are by that fact things we should do, setting aside all questions of what counts as 'ordained by the divine'. As Plato remarks, there are two ways this could go: either those acts ordained by the divine are good just because they are so ordained, or they are good independently of what the divine might have to say on them. But the first options is simply crazy: how could it be that anything, anything at all the divine might say goes, and not only goes but is demanded by morality? There are simply too many things a divinity might conceivably ask of us for that to be the case. There is no shortage of people who try to find ways in which 'divine sanction' is necessarily a good-making feature, but we must conclude that this is an industry of piety rather than good judgement. It must be that divine sanction indicates goodness, rather than ordains it. Not that managing 'mere indication' is a small feat: we value a compass even if the needle tracks north rather than north the needle. Believers should do as their divinity proclaims because it is so wise and compassionate and smart and so on that it will never be mistaken. But there is a lot of ground to cover between never being mistaken and the fact of the matter being at your behest. What the Euthyphro problem does, and does with great force, is make us distinguish between moral ontology and moral epistemology: divine sanction can't make good, it can only indicate the good.

The question is: what is it about the Euthyphro problem that delivers this result? Notice that divine saction of X is an attitude towards X – this is clearer in Plato's formulation, where the target of the attack is the view that the good things are those 'loved by the gods'. The fact that some party A has an attitude towards X is a higher order property than those aspects of X that invite the attitude. That is, there is some properties P, Q etc. of X which leads to A having that attitude towards X for the reason of it having P, Q, etc. This relfective stance property R, 'being the target of some attitude of A's' (for reasons of having P, Q, etc.) , is a product of properties of X (P, Q, etc.) and is thus a higher order property than P, Q, etc. To apply this schema to the Euthyphro case: For R 'is loved by the gods' there are the properties P,Q, etc, that the gods love X for, whatever they might be, and R is of a higher order than P,Q, etc. The problem is that R isn't the good-making feature, P,Q, etc. are: R indicates their presence in some way or the other, and P,Q,etc. constrain R's application – in this case, what the gods could rightly love. R is a property of the wrong order to count as a good-making feature. This shouldn't be a surprise. 'Loved by the gods' tells us more about the gods than it does what it attaches to, and this is the same for all reflective stance properties. But discovering which items in the world are moral goods isn't a project to do with us or any other audience, it is a project to do with the goods. If such unanchored properties such as attitudes we have toward something were to be the good-making features, without any restriction, then we would find such absurdities such as intractable moral disagreement about whether or not to run around trees clockwise, as Phillippa Foot points out. And if there are such constraints on what these reflective stance properties could rightly attach to, then the wrong order problem kicks in, and these reflective stance properties aren't the good-making features, but merely indicate them.

More can be said, and a lot more needs to, but I believe that this problem has some power and can tell us a great deal about how we should go about ethical reflection. A strong consequence is that the wrong order problem makes a nonsense of almost any utilitarianism, especially ones concerned with happiness, desire satisfaction, and so on. This I'm sure is the right result, and in the near future I will give a few examples where I cast famous objections to utilitarianism as cases of the wrong order problem. Because the wrong-order problem concerns moral goods, the objections I have in mind are to do with the aims that various forms of utilitarianism aim at: Nozick's 'experience machine' objection attacking hedonism and Sen's 'dominated housewife' example against desire satisfaction in particular. I also want to work up to Williams's subtle and devestating attack on utilitarianism for being unable to make sense of what it is for someone to have a project. The reader might anticipate how these might go, and this is right and good.

Brom - 'Expressive-Communicative Function of Legislation'

Frans Brom asks us to consider the expressive-communicative function of legislation, where what some law does is to give voice and legal standing to some commonly held moral value, in contrast to two other ostentive roles for legislation, as a way to codify the dominant morality of a society, and as a way to modify behaviour in certain ways the legislators deem necessary. I will respond that I'm unconvinced that this expressive-communicative function (ECF), as Brom's short piece presents it at least, is distinct from the two other functions.

It is worth quoting Brom on the ECF at a bit of length:
"The interactive paradigm of legislation shows that law can form a point of reference in moral discussions. This is especially the case when legislation is based upon a societal agreement the impact of which is not sufficiently conceptualised in moral terms. Once this agreement is restated in legislation, the law becomes an authoritative source fr the concrete normative implications of this agreement. One could even say that the societal agreement is recreated into a moral framework for the public domain."
This is all well and good, but I think Brom needs to do more to seperate the ECF from legal codification and modification. His view seems to be that the ECF of a law is to give a moral framework wherein some value is given creedence, which in turn provides normative guidance that would be lacking if we were still arguing about what that value comes down to. I say that the first part of this falls under codification and the second under modification, that is, the ECF reduces to the cases where the codification of some moral position is such to modify behaviour in the jurisdiction away from other positions, which is hardly a strange effect of codification.

If codification happens, it is always the codification of one position among many. That position needn't be one which is actually endorsed by anybody (nor is it likely to be exactly like any non-legal framework, but never mind that). The fact that such-and-such a framework is the product of a code of laws carries with it the consequence that that framework now works upon everybody in the jurisdiction. Being under the effect of some law can't help but modify behaviour in its jurisdiction to the effect of encouraging actions in line with the codified framework (leaving aside the interesting case of dead-letter laws, which don't concern us here because they also fail to authoritively express the position of anybody, because they fail to carry authority). And so much the worse for different frameworks to the extent that they differ from the one enshrined in law.

So the normative implications of ECF are covered by the classical codification and modification views as well. There still is an issue of what exactly the target of the codification is, especially in the type of cases Brom draws attention to, where there isn't widespread agreement on exactly what the endorsed value consists in, like the intrinsic value of non-human animals in Dutch animal welfare law. An important point about vague concepts like these, which David Lewis reminds us of, is that while it's unclear where their applications start and end, it isn't that anything goes. Just as the boundary between blue and green is unclear and yet there are blues that aren't at all green and vice versa, so there are many things that might be an animal's intrinsic value and things which clearly couldn't. Pointedly, the type of view where animals have ethical value only in as far as they are valuable to people simply couldn't qualify. Thus, the position being codified is actually a range of what could count under the vague term. This means that you can still have sensible disagreement in the courts about, say, whether a law protecting the inherent worth of animals prohibits conditions where they are kept in good health but in extremely restrictive settings, since it's vague whether an animal moving in its distinctive ways is part of its inherent worth. But it stops acts which egregriously disregard the well-being of the animals involved. Once again it is important to note that the view in law needn't be the view anybody had prior to the passing of legislation, so the fact that the law covers a range of positions rather than any single one isn't a problem. It can gain support from everybody across that range (though that support is likely to be qualified) in that they agree that something like the law needs to be enacted.

Thus, the classical functions of codification and modification can do everything Brom recommends the ECF for: it can posit an authoritative framework with normative implications, and can do so in a way to accomodate vague values rather than precise positions.

Verweij - 'Moral Principles and Justification in Applied Ethics'

As is usual in this introductory collection, this entry by Marcel Verweij is very compressed and yet covers a lot of ground, so I'll do what I've done mostly and offer a comment on a particular aspect of the discussion I think can do with some expansion. Verweij concerns himself with laying out some reasons to believe that general principles can't do all the work that needs to be done in applied ethics. Be that as it may, I believe that there is some work which principles of the most straightforward and forceful sort can accomplish, and that is to put constraints on what courses of action we could rightly pursue. Vermeij's presentation gives the impression that this wouldn't be the case, because principles are supposedly too abstract to give clear guidance in concrete cases. However, a thickly developed concept like 'always be honest' can do exactly that, I argue.

The main difficulty with working with principles, as Verweij rightly points out again and again, is that it isn't always straightforward to see what they would entail. However, this is a problem of vagueness rather than one concerning a principle's normative force: if you could be clear about exactly what the content of a principle is, this problem would disappear, or it would become clear that it wasn't as useful as it at first appeared. Perhaps the latter turns out to be true of every principle, but we should hesitate before we conclude as much. If we could determine what would count as fulfilling the demand of a principle, and what a clear failure would be like, that would meet the challenge to provide concrete guidance. Perhaps the principle wouldn't tell you which course of action would be uniquely right, but it would filter the options by rejecting a number of candidates which would simply be unacceptable on account of that principle: that is what I mean with principles placing constraints on what we should consider right action.

The principle of 'be honest', by which I mean 'never intentionally deceive someone so that on your account they reason from false grounds' is an excellent example of how this can be done, because it is a principle which is established in a number of independent ways. Firstly, it is a principle with enormous standing. It is the one clear success of Kant's categorical imperative, for instance: it is simply inconceivable to have a situation in which people communicate without most things they say being true, because that would only lead to a failure in communication. Doing the same but in an entirely different way, David Lewis gives an analysis and vindication of a convention of truth-telling, that is, the regularity of people telling the truth and expecting others to do so, on game-theoretic grounds. And, different again, Elizabeth Anscombe held that there simply are no situations in which it is right to lie, though not as an absolute prohibition, but as a matter of fact which drops out of the fact that each of us rely so heavily on what we tell each other that the goods we might gain from deceit could hardly be goods of moral worth. Secondly, what honesty demands of us is perfectly clear in almost every case. We know very well what it means to tell the truth, and because deceitfulness is predicated on deceitful intent, we can't help but know when we lie. That is, there are clear success conditions on being honest, and a precise way to tell when you're being deceitful. There are debates and difficult questions about how far you should pursue it: Kant's insistence that you should tell the truth even to an enquiring murderer might not impress most people, but everybody can (or damn-well should) agree that it is not good in itself to deceive other people, and that it is not bad in itself to furnish people with the truth. The degree to which a course of action demands deceit is the degree to which it is to be avoided, and the contrary: this isn't a complete decision procedure, but it is an informative and concrete constraint on action.

Perhaps it is a mistake to appeal to principles to settle the question of what you should do in such-and-such a situation, but it would be too quick to move from that to a conclusion that principles can't offer guidance in real cases. They might not tell you everything, but given the appropriate success and failure conditions, principles can tell you to some extent at least what it is you should do.

Thorseth - 'Ethical Pluralism'

May Thorseth's contribution to the collection is a piece on ethical pluralism, which she presents as a more substantive alternative to relativism. I agree that distinguishing the two positions is a very worthwhile thing to do, but here I will quickly argue that teasing pluralism apart from relativism is probably harder than Thorseth's piece makes out. Like the other pieces in the collection, this one is very compressed and covers a lot of ground quickly, and I don't imagine it's meant as the least word on the subject. Still, as presented I believe there is a form of relativism which can manage everything Thorseth gives as a reason to prefer pluralism. Then I make a suggestion which I hope can serve as a way to pin down the difference Thorseth is gesturing at.

Her case, following Charles Taylor, is that a purely formal approach such as relativism can't pick out the goods which are connected to import-attributing feelings, for which only a substantive engagement with a position can suffice. I will make the rather obvious reply that you can conceive of a relativism which is up to task, one of the form 'it is right for X to pursue those goods which promote import-attributing features'. Because this is really boring I want to instead say something about why I think it is a mistake to place too much weight on what someone might feel when you're trying to pin down concrete goods.

The idea behind import-attributing feelings is that not everything a person does is something they consider a matter of integrity: people often feel that it is of critical importance that they get to practice their religion or live a particular lifestyle, while it is decidedly odd for someone to tie how well their life is going to whether they around trees clockwise. The challenge for relativism is supposed to be that, from all the candidate goods available, a purely formal approach can't see whether some good is the target of import-attributing feelings. For instance, cultural membership is very important for many people, but there are others who attach no importance to it, so a relativism which makes cultural membership be the deciding factor in what is and isn't moral is inappropriate.

However, there is nothing that stops a relativism which determines rightness by way of which goods are the targets of import-attributing feelings. It would go something like: if X is the target of an import-attributing feeling of A's, then it is right for A to pursue X - let's call this relativism from import attribution (RIA). This is no different in form to the paradigmatically relativist position 'if members of A's culture typically pursues X, then it is right for A to pursue X'. You could complain that you can't determine what those feelings target without a substantial engagement, but nor could you see what someone's culture membership is. Furthermore, there is something right about RIA: cultural membership is a paradigmatic concern of relativism, but only because people typically consider it to be a central feature of your identity, and likewise for other candidates of the formal property that the relativism concerns itself with.

Concentrating on feelings leaves the door open for relativism, because feelings are feelings about things, that is, they are higher-order properties which can attach in various ways to more immediate goods. The problem, which RIA shares, is that such feelings can attach to almost anything, leading to Phillipa Foot's complaint that even a sophisticated relativism like Richard Hare's can lead to such absurdities as irresolvable moral disagreements about running around trees clockwise. To prevent this, there needs to be some restrictions on what could rightly be a target of such feelings, that is, propriety conditions, like which Foot's own 'natural goodness' account tries to deliver. However, doing so is very difficult, and Thorseth makes no attempt to do so, so pluralism as she presents it is in the same muddle as RIA because anything could potentially count among the variety of goods.

Nonetheless, we can establish that there must be such propriety conditions through the use of the Euthyphro problem, much like I presented in an earlier post (such uses of the Euthyphro problem are something I've written about and continue to work on). The dilemma goes as follows: either the feelings are just a way to know what goods are involved (how Thorseth, Taylor et al present it), or being the target of such a feeling is the good-making feature (as RIA has it); but it can't be a good-making feature, because then anything whatsover can be an ethical good, such as looking at rabbits under moonlight (another example of Foot's), which is just absurd; thus, the targets of these feelings are good independently of the feelings; thus, those feelings are only appropriate when they rightly target those things which are good. And now we have the task laid out clearly: to identify what the things are to which one rightly assign importance.

It's hard to keep up with the breakneck pace of academic philosophy...

As I said a few posts ago, I'm currently using this blog as a place to comment on the articles I should be discussing with my peers in Linköping, Sweden, as I start the course of study I've been invited on. I'm not in Sweden, or the Netherlands, or anywhere connected to that study, because I do not have a visa to enter Europe. I've spent the past few days wrestling with the various authorities as I try to change this regrettable and rather silly state of affairs. I'll be trying to catch up in the next few days, because I need to hit the ground running once I'm finally over there, and because I need a productive way to pass my time other than working up passive aggressive attitudes towards bureaucracy.

Futher comment re: reflective equillibrium

Thinking back on my previous post, a comment on Collste's introduction to applied ethics, I've decided my treatment suffers from the same problem I see in Collste's piece. The discussion is too short to clearly distinguish two different uses of reflective equillibrium, one where RE serves as a structure for argument, the other where it acts as justification. So let me try that again.

First, RE as a formal tool: If a suitable course of action has been picked out, or one that you believe is suitable, you can try to explain why by showing how that action sits in a reflective equillibrium. That would consist in stating the various relevant factors, drawing out the conflicts between them, staking out a suitable give-and-take between these factors and indicating how the chosen course of action fits appropriately between them. To say it in a different but equivalent matter: for relevant facor X, determine how the other relevant factors Y, Z, etc. impact on the demands of X to determine X', being the demands X makes in this case compared to Y, Z, etc, and do so for Y, Z, etc. as well; the appropriate course of action C is one which fits X',Y',Z' and so on. The fact that the appropriate course of action is so is explained by it respecting all the demands of the relevant factors as measured against the others.

Of course, not every candidate course of action can fit in such a reflective equillibrium. If it doesn't, you could try to use RE as a method to find the right course of action. Let us say the most attractive candidate course of action C meets demands X' and Y', but not Z or even an appropriately modified Z'. There could be a change to C to make a C' which might fit the bill, or if no such change is forthcoming, perhaps a further modification to the apparent demands of the various factors, maybe making a X'' which leads to changed demands on Y' which makes a Y'' and a Z'' which a C or C' could meet. Because you have made moves to change your approach to the case, and have done so based on RE, RE has done substantive work here. This is different from the former case, where RE does just formal work.

So, to sum up in conclusion: where the course of action one picks is picked independently of RE, and RE is used to explain why that is the right thing to do, there RE only does the formal work of structuring the argument; however, you might be tempted to change your approach based on what would bring about an RE, and that would be to use RE substantively, to choose to do this rather than that, and even (in what Rawls calls cases of 'wide reflective equillibrium') to modify your approach in other relevantly similar cases.

Collste – 'Applied and Professional Ethics – an Introduction'

Göran Collste contributes to the MAE handbook with a bit of an overview off the field of applied ethics. There is a point about reflective equillibrium I am a little uncertain about, which I'll discuss here.

In discussing RE, Collste remarks that it can be used either as a structure for debate, or a form of justification, but not both. As I understand it, this is because if it is used as an organisational schema, anything contributed by RE to the argument can't be informative, but can only be a different way to portray the content of the various positions. I don't think that Collste's severely compressed discussion distinguishes the two positions sufficiently, so for my purposes I'll try to do so. From what little I know of him, RE in applied ethics seems to be a particular interest of Collste's, but I have only this short article of his to work from, so that is all I'll try to engage.

Collste's example is of introducing an expert opinion (a doctor's considered judgement) and a general principle (the principle of utility) into a discussion of how to respond to a particular problem (suitable treatment for neonates). He chose an intuitionist and principlist disagreeing with eachother to bring out RE's features, he tells us. I'm a little worried about his characterisation of the intuitionist approach and what that entails, which is the point I'll work from towards my characterisation of the two different uses of RE.

Collste first introduces the doctor's judgement as an intuition, and then as a considered judgement, but there is a lot of ground to cover between the two. A considered judgement is a mere intuition (a snap judgement) that has been supplemented with careful attention to the specifics of the case and what the various options entail, and so on. A considered judgement is the type of thing you can have in hindsight, whereas a mere intuition is the product of a moment's sensation. Collste is welcome to pick out considered judgements, the least flighty type of intuition, as what he means to contrast with a deduction from a general principle, but he then proceeds to give them short shrift in what follows. He states that if the doctor were asked to back up his conclusion, there would be nothing to point back to other than their own “intuition and practice, but not much more” (p. 80). But more than intuition and practice goes into a considered judgement. 'Intuition' in the sense of 'sensation' can pick out salient dimensions of the case (the quality which intuitionists claim makes expert intuition indispensable), but what makes an intuitive judgement considered is exactly the careful attention that is given to the possibilities along those dimensions. The precedents an expert can draw upon have content, after all: in a similar case this happened when such-and-such was attempted; thus, in our case, this or something relevantly similar will happen again; and so on. If nothing else, it includes the weighing up of consequences that experience has taught one to expect. When deciding between uncertain consequences, it's hard to deny the power of an appeal to a precedent, especially one where the implications are drawn out with an expert's sensitivity. Knowing what will happen in such-and-such cases is after all exactly what we have doctors for.

On my view there is still a clear difference between an intuitionist line and a principlist one. An intuitionist one is a product of the specifics of the case, whereas a principlist one abstracts over specifics and makes proclamations of general scope. For the utilitarian of Collste's example, no other dimension other than expected happiness is relevant, whereas the doctor's intuition can draw on every aspect of every suitable precedent. When both views are given a place at the table, then, they pull in different directions: the principlist insistence that everything can be decided on some general considerations, while the intuitionist brings the intricacies of the case to bear. One can see the different sides being argued as a gauntlet that some candidate response to the problem has to run through. As I see it, if the response can satisfy every side, then RE functions as an organisational schema, and if it doesn't, it serves as a decision making procedure. In the former, the response is cast as the product of an equillibrium between the demands of ones principles and the realities of the case, and in the latter, the promise of an equillibrium motivates revisions to the candidates and the various sides to the debate until an equillibrium is reached.