Tuesday, May 28, 2013

Must we pursue good causes to have meaningful lives? (Part One)

George Bailey, It's a Wonderful Life


I’ve occasionally dipped my toe into the literature on the meaning of life and reported back my findings on this blog. I don’t know if anyone has found these reports to be interesting or worthwhile. I worry that my approach to the topic has been too piecemeal and unsystematic, gradually feeling its way into the key debates and concepts, without ever properly sketching out the broad theoretical topography. But perhaps that is the only way one can really learn about a topic like this.

This post (and its successor) are going to be more of the same I’m afraid. They are going to take a look at Aaron Smuts’s article “The Good Cause Account of the Meaning of Life”, which defends a particular kind of theory of meaning in life. Starting out, we can note three general types of theory one might have:

Subjective Theories: According to these theories one’s life is meaningful if and only if a certain set of conditions is met, where those conditions all explicitly involve subjective awareness or experience of a certain kind (i.e. the conditions for a meaningful life are subjective in nature).
Objective Theories: According to these theories one’s life is meaningful if and only if a certain set of conditions is met, where none of those conditions require subjective awareness or experience of a certain kind (i.e. the conditions for a meaningful life are objective in nature).
Hybrid Theories: According to these theories one’s life is meaningful if and only if a certain set of conditions is met, where those conditions include some objective conditions and some subjective conditions.

Each of the theories attempts to locate the necessary and sufficient conditions for meaning. They differ in terms of the nature of the conditions. Smuts argues that only objective theories are plausible. In particular, he argues for a Good Cause Account (GCA) of meaning, which runs something like this:

Good Cause Account (GCA): A person’s life is meaningful in virtue of and in proportion to the amount of objective good they are causally responsible for.

I want to summarise and comment on Smuts’s case for the GCA. This will take me a couple of posts. In the remainder of this post, I’ll look at his arguments against subjective and hybrid theories, specifically those from Richard Taylor and Susan Wolf. In the next post, I’ll look at how he articulates and defends the GCA.


1. Subjective Accounts of Meaning in Life
Richard Taylor wrote a famous paper about the meaning of life back in the 1970s (maybe earlier, I’m not sure). He cleverly titled it “The Meaning of Life”, and made use of the classic myth of Sisyphus to illustrate his key points, thereby ensuring its continuing anthologisation (I’m being facetious. Ignore me.). In the paper, Taylor makes some interesting arguments though none of them are particularly well fleshed-out. The main argument defends a subjectivist account of meaning in life.

Taylor asks us to imagine a paradigmatic case of a meaningless life. He fixes on the example of Sisyphus, doomed to roll a large boulder up a hill for eternity. Every time he reached the top, the boulder rolled back down and his task would start again. This seems to epitomise meaninglessness. Taylor then proposes an additive test for meaningfulness. In other words, he asks to add things to Sisyphus’s life until it becomes meaningful. This way we will find the sufficient conditions for meaning (and if we add and subtract things we might even find the necessary conditions too).

Adopting this process, Taylor argues that simply giving Sisyphus’s actions an objectively good or valuable end won’t do the trick. For instance, he asks us to imagine that Sisyphus’s boulders are used to construct something of great artistic value such as a cathedral. Would that make his life meaningful? Taylor argues that it wouldn’t because such productions have no enduring value: they will eventually crumble, fall and be swept under the rug of cosmic history. (This is a bad argument, but I won’t explain why here. See my earlier post on Nagel’s arguments for absurdity).

So Taylor proposes instead that we simply make Sisyphus happy to perform his task. In other words, that we give him a strong and overwhelming desire to roll boulders up a hill. This, Taylor argues, will do the trick. It will give fulfillment and meaning to Sisyphus’s life in a way that giving his actions some objectively desirable or valuable end will not. Taylor calls this the “Fulfillment” Account:

Fulfillment Account: A person’s life is meaningful in virtue of and in proportion to the extent to which their desires are fulfilled.

Smuts critiques this theory in fairly short order, borrowing in part from Erik Wielenberg’s criticisms. As Wielenberg points out, a purely subjectivist account of meaning has some seemingly absurd implications. For instance, the life of the sadistic torturer and the life of the grinning excrement-eater (Wielenberg’s example, not mine) are meaningful on this account. Indeed, they are as meaningful as, and possibly even more meaningful than, the life of the person who desires to cure cancer and succeeds.

In addition to this, Smuts’s argues that subjectivist theories cannot make room for the possibility of error when it comes to assessing the meaning of your own life. On the fulfillment theory, if you are not happy with your life, then it lacks meaning. But surely this cannot be. Smuts’s uses one of his favourite examples to illustrate this point: the life of George Bailey from the film It’s a Wonderful Life. In the film, George plans to commit suicide because he thinks his life is meaningless. But an angel comes along and shows George what Pottersville (the town in which he lives) would have been like if he had never been born. In doing so, the angel shows George how meaningful his life really is, and how many people have relied upon and benefitted from his existence. George made a mistake. His life was meaningful even though he wasn’t enjoying it. The fulfillment theory doesn’t allow for this.

For these reasons, Smuts rejects the subjectivist account of meaning. I have to say I’m not entirely convinced. The objections rely heavily on our intuitive reactions to particular cases, which can be okay, but Taylor was aiming for a somewhat counter-intuitive account of meaning in his article so I wonder if the dismissal is a little too quick.


2. Hybrid Accounts of Meaning in Life
Even if a purely subjectivist account of meaning is flawed, it seems plausible to say that some subjective conditions must be satisfied in order for life to be meaningful. A life filled with pain and misery doesn’t look like it would be worthwhile, certainly not from the perspective of the one living it. This is where hybrid theories come into play. They accept that purely subjective theories have their problems, but they want to retain some subjectivity in our accounts of meaningful lives.

The most important proponent of a hybrid theory is, probably, Susan Wolf. She defends something we shall call the fitting-fulfillment theory of meaning. This can be defined as follows:

Fitting-Fulfillment Account: A person’s life is meaningful if they are subjectively fulfilled by pursuing objectively valuable ends.

This account cleverly avoids the Wielenberg-style criticisms of the fulfillment account by adding a “fittingness” requirement. Neither the life of the sadistic torturer nor that of the grinning excrement-eater will be meaningful on this account. They may both have fulfillment, but they do not pursue objectively valuable ends. Thus, they fail to meet the “fittingness” requirement for meaning. Nevertheless, fulfillment is still an integral part of this account. One cannot have a meaningful life without it.

And therein lies the problem for Smuts. Although the fitting-fulfillment account addresses some concerns about the subjectivist account, it still does not resolve the error problem. George Bailey’s life would still be meaningless on this account because he was not fulfilled by it. But, again, Smuts submits that this is just plain wrong: his life was meaningful, he just didn’t realise it. If that’s right, then the fitting-fulfillment account must be wrong.

Once again, significant weight rests upon one’s reaction to the counterexample. One concession that Smuts makes is to say that the fitting-fulfillment account may work if our goal was to identify the conditions for an ideal life. Even the staunchest advocate of an objectivist account would concede that George Bailey’s life is better after the angel corrects his mistake. But a theory of meaning is not, according to Smuts, a theory of ideals. It is about crossing a threshold, not attaining an ideal. The error problem points out that the threshold can be crossed without a certain type of subjective awareness.

If Smuts’s criticisms are correct, the way is cleared for a purely objectivist account of meaning in life. But such an account does not simply win by default. It must be rendered plausible in its own right. In the next post, we’ll see how Smuts tries to do this.

Monday, May 13, 2013

How should we respond to moral risk? (Part Two)



(Part One)

Our moral reasoning follows well-worn pathways. An argument is presented telling us that a particular action should be deemed good/bad or right/wrong; we assess its premises, locating its strengths and weaknesses in the proces; and then, following an exhaustive process of deliberation and evaluation, we reach an all-things-considered judgment as to the merits or demerits of the particular action.

Commonplace as that style of reasoning is, it has some obvious shortcomings. The most significant of which has to do with our own cognitive limitations. When assessing ethical arguments, we are often taken into abstruse and esoteric areas of metaphysics and morality. There is a very significant and live possibility of our making some mistake about these areas of inquiry. What difference does that risk of error make for our ethical decision-making?

The literature on moral risk tries to address this question, and I’m currently trying to get a handle on it. As part of my efforts, I’m writing this series of posts on Dan Moller’s article “Abortion and Moral Risk”. In part one, I looked at Moller’s attempt to highlight how moral risks arise in the abortion debate. Now, in this part, I’ll look at Moller’s attempt to develop some principles and rules of thumb for dealing with those risks.

The discussion will take two parts. First, it will consider the general question: once you have recognised the possibility of some kind of moral risk how should you let it affect your moral decision-making, if at all? Broadly conceived, there are three possible answers, two extreme and one intermediate. Moller defends the intermediate position, according to which recognising the possibility of moral risk should make some kind of difference to your decision-making, though its not clear how much. This leads to the second part of today’s post, which looks at five different criteria Moller thinks you should keep in mind when determining how much of a difference it should make.


1. Dealing with Moral Risk: Three Approaches
When it comes to acknowledging the impact of moral risk on actual ethical decision-making, there are three possible positions one could take up. They are arranged along a spectrum, with the two ends representing the two extreme positions, and the full-length of the spectrum between those extremes representing the blurry and indistinct intermediate position. The three positions are:

1. No-difference: Even though there is a real possibility that you could be making a moral error, you should not change your behaviour as a result.
2. Some difference: The possibility of moral error should make some difference to your decision-making, though the degree of difference it makes varies in accordance with a number of factors.
3. Paralysis: The possibility of moral error should cause you to refrain from making any decision.

As might be clear from its pejorative name, the third possibility is easily dismissed. If one refrained from performing an action or making decision on the grounds that there was some moral risk associated with it, one would soon be unable to do anything. That’s because there’s some degree of moral risk associated with any decision. Risk-free decision-making is an ideal, not a reality.

So that leaves us with the other two positions. At first glance, the “no-difference”-position has some allure. Because risk-free decision-making is an ideal, we might be tempted to take a rather cavalier attitude toward risk, acknowledging its existence but ploughing ahead with our all-things-considered judgment nonetheless. This, however, would be a mistake. Or so, at least, Moller argues.

He argues this point by asking us to consider two thought experiments, both involving some level of moral risk, and asking us whether it is plausible to think we should “plough ahead” in both instances. I’ll briefly describe both thought experiments now. As you read through the descriptions keep in mind that the examples have some features that Moller thinks should alter our reaction to moral risk.These features may be pushing your intuitions in a particular direction and may cause you to question the generality of the lessons that can be learned from these examples. I certainly found this to be the case when I first read them. But I think they do make sense when you read them in light of Moller’s criteria for determining the impact of moral risk. So I would suggest being patient and re-reading them after you know what those criteria are, which you will do after reading the second section in this post.

The first thought experiment tells the story of Frank, who is dean of a medical school, and must decide whether the school should pursue important research according to Plan A or Plan B. As far as he can tell, there is little separating the plans, apart from the fact that A has less paperwork associated with it. But Frank has little ethical expertise so he waits to hear the deliberations of his five-person ethics committee. They come back to him with the following conclusions. First, they all agree that Plan B is ethically acceptable. Second, three out of five think that Plan A is acceptable, but two out of five think it comes with a real risk of doing significant ethical harm. What should Frank do?

The suggestion is that Frank shouldn’t simply discount the views of the two members of the ethics committee. They have greater ethical insight and knowledge than he does, and their belief that there is risk of significant harm should be factored in. Certainly, the fact that A has less paperwork associated with it is unlikely now to be a decisive reason in its favour. Indeed, the situation seems to be reversed in terms of the desirability of the two plans. Since B doesn’t appear to have any significant harms associated with it, it looks to be the more desirable option.

Frank’s case is what Moller refers to as a “thinly-described” example. So consider a second one, this time involving a woman named Sally who has a temporary illness that will last for a month. The illness is such that if she conceives a child within the month, the child will be born with a severe handicap. The handicap will significantly reduce their quality of life (relative to a “normal” person) but not to the point that their life is not worth living. If she waits a month, the illness will pass, and so too will the possibility of any child she conceives having this handicap (though, of course, other problems may arise). Suppose Sally adopts a person-centred theory of wrongdoing, according to which she cannot do any harm to a child she conceives within the month since the child does not yet exist. But suppose she is aware that there are impersonal theories of wrongdoing according to which conceiving the child within the month is wrong (even though it doesn’t harm the child). What should Sally do?

Again, it seems wrong to say that Sally’s risk of moral error — in this case an error having to do with personal versus impersonal wrongs — is insignificant. Indeed, it seems like conceiving the child within the month would be the wrong thing to do in this instance. She should wait until her temporary illness passes. The moral risk in this case seems sufficient to warrant the extra degree of caution.

If all this is right, then the “No difference”-position is flawed. Moral risk makes some difference to our moral decision-making. (Note: if you think it’s ironic that the defence of this view relies on the very same methods — i.e. analysis of thought experiments — that were thought to be risk-laden in part one, then you’ll be glad to know I thought that was ironic too).


2. Weighing up the Moral Risks
But how much of a difference does it make? This is a very difficult question to answer, but it is the most important for anyone taking moral risk seriously. To this point, Moller has been dealing with the low-hanging fruit in the analysis of moral risk. Establishing the existence of genuine moral risks, and ruling out the extreme positions one can take up in relation to that risk is a relatively easy thing to do; the hard part is figuring out how to deal with the blurry and indistinct line between those two extremes.

Unfortunately, Moller has only a few pointers to offer on this crucial topic. Still, they are somewhat helpful and worthy of consideration. In total, he recognises five separate factors that will affect how much of a difference moral risks make to our moral decision-making. To illustrate, assume that our choice is whether or not to perform action A (if it helps, imagine that “A” is the act of getting an abortion, since that was the case study we looked at in part one). In that case, the following five factors should be borne in mind:


  • 1. The likelihood that act A involves wrongdoing; 
  • 2. How wrong A would be if it were wrong
  • 3. The costs the agent faces if she omits A;
  • 4. The agent's level of responsibility for facing the choice of doing A;
  • 5. Whether not doing A would also involve moral risk.


Moller is clear that this is not an exhaustive list, and is at best a starting point for a complete theory of moral risk. Bearing that in mind, let’s briefly talk through each of these five factors. As we do, we’ll see how they may have influenced our judgment about the two thought experiments discussed earlier.

The first factor is relatively straightforward, conceptually, but more difficult practically. When making a decision about how much weight to place on the possibility of moral error, an obvious thing to consider is the probability that you are making that error. If the probability of error is 0.4, then it would seem to count for more than if the probability of error is 0.0001. The problem is how exactly do you come up with a measure for the likelihood of moral error? I can imagine all sorts of measures being used — e.g. the opinion of ethical experts within the relevant field — but I don’t know that any of them would be particularly good.

The second factor is also pretty straightforward conceptually, and perhaps a bit easier on a practical level. Obviously, the magnitude of the wrong done (if it is a wrong) would have a significant impact on decision-making. Abortion is a compelling example because if you do make an error (morally speaking) you might be doing something very bad (i.e. killing an innocent person). Contrast that with a case in which one of your choices comes with a risk of causing someone to feel pain equivalent to a light pin prick for less than one second. That’s morally bad, for sure, but much less so than killing an innocent person. Thus, even if there was a high likelihood of making the error in the latter case, it might be okay to run the risk.

The third factor relates largely to the prudential costs of not performing the relevant action. I’m not entirely sure why Moller considers it significant, but that’s because I’m not entirely sure whether his argument is primarily about rational decision-making or moral decision-making. Certainly, personal costs are relevant to both, but the weight attaching to them might vary depending on whether one is concerned with making rational choices or moral choices.

The fourth factor is no doubt an important one, but Moller says very little about it in the article. I think the basic idea behind it is that the more responsibility one has for the relevant decision, the more important the possibility of moral error becomes. But that intuition would need to be worked out in greater detail.

The fifth factor is given pride of place in Moller’s discussion as he thinks it to be particularly important. The gist of it is that moral risk has its biggest impact on decision-making when the risks involved are asymmetrical. In other words, when the errors (both in terms of probability and magnitude) fall predominantly on one side. This is probably why Frank and Sally’s cases were compelling. In Sally’s case, for instance, there doesn’t seem to be any moral risk associated with waiting a month to conceive a child (except for maybe the risk posed by anti-natalist theories); all the risk attaches to conceiving within the month. When there are roughly equal risks on both sides, there will be little impact on decision-making. In between, there are any number of problematic cases.


3. Conclusion
So there we have it, a brief overview of Moller’s paper “Abortion and Moral Risk”. To quickly recap, in part one we looked at the types of moral risk that might arise in the abortion debate. As we saw, moral risks could arise from getting the theory of personhood wrong, or weighting one’s reaction to different thought experiments incorrectly. In part two, we considered what we should do about such moral risks. Dismissing the extreme positions, we saw how Moller argued that moral risk should make some difference to our decision-making, though maybe not a decisive difference (this was called the “some difference-position”). We then looked at five factors that determine how significant moral risks really are. They varied from the likelihood of the risk to the asymmetrical nature of the risks involved in the relevant decision. It was suggested that the latter was particular important insofar as many of the most compelling cases for the “some difference”-position involve highly asymmetrical risks.

Wednesday, May 8, 2013

How should we respond to moral risk? (Part One)



Philosophy can be an exhausting and debilitating business. As someone once said (I’m paraphrasing) “the unexamined life is not worth living, and the examined life isn’t all that great either”. One of the reasons is the unavoidable desire (and tendency) that philosophers have to “go meta”; to jump to a higher and more abstract level of debate. Just when you think you are beginning to get a grip on the arguments and reasons offered at one level of analysis, someone comes along and tells you that you need to start arguing about those arguments, reasoning about those reasons, analysing that level of analysis.

One example of this tendency to “go meta” can be found in the (burgeoning) literature on moral risk. As you are all probably aware, many ethical debates rely on obscure and hotly contested concepts and principles. What’s more, there is lot riding on whether we get those concepts and principles right. After all, if we don’t, we could be committing some grave ethical sin.

The literature on moral risk considers the implications of all this. It highlights how ethical reasoning (should) be seen to involve two distinct levels. The first level — which we can call the ground level — involves reasoning about the ethical principles and concepts that are relevant to a particular domain of ethical concern (e.g. principles of fair distribution). The second level — which we can call the meta-level — involves reasoning about the possibility that we get things wrong at the ground level. In other words, it involves working out how we should react to moral risk.

As much as the tendency to go meta wearies me, I have to admit that the possibility of getting things wrong (morally speaking) worries me quite a bit. Thus, I think there is merit to the philosophical analysis of moral risk, and I want to learn more about it. That’s why over the next two posts I’ll be taking a look at Dan Moller’s paper “Abortion and Moral Risk”, which is a useful entry point into the debate.

Moller’s paper gives a nice overview of moral risk, and shows how thinking about moral risk might affect our practical ethical decisions. He chooses the always-controversial test case of abortion to illustrate his approach. His analysis breaks down into two distinct phases. First, he tries to convince us that there is a very real risk of getting the ethics of abortion wrong. This is important because unless we acknowledge this risk the practical question of what to do about it — which is the primary question in the field of moral risk — will seem pointless. And second, having provided the requisite motivation, he looks at how this risk should affect our ethical decision making.

In the remainder of this post, I look at the first part of Moller’s analysis, leaving the second part to a subsequent post.

(Just a warning at the outset: I have no desire to make or defend any particular claims about abortion in this post or the next. The claims about moral risk are all that interest me.)


1. The Deprivation Argument
As I just said in the intro, ethical decision-making is high stakes and subject to much heated debated. Consequentialists and deontologists lock horns on a range of issues, and set up well-fortified ideological camps for arguing the pros and cons of many practical questions. But this disagreement, in and of itself, is not the basis for moral risk. No; the risk emanates from the esoteric and obscure principles and claims that support certain moral position.

Many examples could be chosen to illustrate this point — Moller explicitly mentions vegetarianism as a good one — but the one discussed in his article is abortion. This is a good example for a variety of reasons. It is contentious, likely to raise hackles, and the risks involved are asymmetrical (which is particularly important when come the second-level of analysis as we shall see in part two).
The only problem with using abortion as an example is that the literature on abortion is huge, encompassing many different arguments and theories. To make the example more manageable (and more useful) we’ll have to hone-in on a very small sub-section of it, specifically a section dealing with the Deprivation Argument against abortion. In its simplest form, this argument runs as follows:


  • (1) It is wrong to deprive a person of future positive experiences.
  • (2) Abortion deprives a person of future positive experiences.
  • (3) Therefore, abortion is wrong.


If the argument is right, then abortion is a serious moral wrong (equivalent in wrongness to killing). If it is wrong, then abortion might be permissible (I say “might” because there are other arguments to consider here too). But how can we tell whether it is right or wrong?

The standard approach is to consider the reasons that could be offered for and against its respective premises. But suppose you do that, and at the end of the process, you decide that the argument fails: It provides no good reason to think abortion is wrong. Is that all you need to do in order to make a morally informed decision? Moller says “no”. You also need to pay close attention to the kinds of moral error you could be making in your assessment.


2. Two Examples of Moral Error in the Deprivation Argument
There are many errors you could be making, but let’s just focus on two for the time being. One relates to the second premise, and one relates to the first premise. We’ll start with the second premise. This premise contains the all-important phrase “deprives a person”. In doing so, it assumes that there is some identity relation between the foetus and the child or adult it may eventually become such that they are one and the same person. But is this true?

To answer that, we need some theory of personal identity. There are two leading theories out there, the Neo-Lockean theory and animalism. According to the Neo-Lockean theory — named in honour of John Locke — a person is a continuing subject of conscious experience. In other words, a person is an entity which is sentient, aware of its own sentience, and aware of its continuing sentience over time (alternatively, sentience is not critical here but rather some other overlapping set of psychological traits). Under this theory, the foetus is not a person who can be deprived of future positive experiences. The foetus (presuming the abortion is performed early enough) would not have the sufficient mental apparatus to develop the requisite psychological traits. So if we adopt a Neo-Lockean theory, we would conclude that the Deprivation Argument is false.

The picture is rather different under animalism. According to this theory, neither a continuing self-conception nor an overlapping set of psychological traits is essential for personal identity. Instead, personal identity is held to be an essentially biological characteristic. A person is just a particular set of dynamic, homeostatic biological processes. So what matters for identity over time is whether the foetus and the future child or adult share the same underlying set of biological processes. Arguably, they do. So if we adopt an animalistic theory, we would conclude that the Deprivation Argument is true.

The thing is, it’s very difficult to say which theory is right. There are a variety of arguments and considerations that could be marshalled in favour of one or the other, but the subject matter is so arcane, and the weighting of the arguments so difficult, that we could very easily make an error in favouring one over the other. That gives us our first possible moral error:

First Possible Moral Error: We could have the wrong theory of personal identity. Under the Neo-Lockean theory, the Deprivation Argument would be false; under Animalism, it would be true. But we could very easily make an error in favouring one theory over another.

This brings us back to the first premise, and the possible errors we could be making there. One obvious criticism of that premise is that it is too absolutist in nature. There are many times when it seems like depriving another person of future positive experiences is permissible. For example, killing in self-defence or assisting suicide (not that this isn’t without its controversies too). One way of exploring the bounds of permissible killing is to consider a range of thought experiments. One of the most famous of these thought experiments is Judith Thomson’s violinist case, from her 1971 article “A Defense of Abortion”. In this thought experiment, you are asked to imagine waking up one morning to find yourself hooked up to a famous violinist who needs to share your kidneys for nine months (otherwise they will die). You are then asked whether it is permissible to disconnect from him/her?

The suggestion from Thomson, and many others, is that it is. Hence, assuming this case is roughly equivalent to that of pregnancy, it follows that premise (1) is wrong and that abortion is permissible. But, of course, Thomson’s thought experiment is subject to all manner of objections. One of the most interesting comes from the work of David Hershenov, who suggests that the intuitions pumped by Thomson’s thought experiment might depend on distorting factors. Specifically, he argues that it is easy to change one’s reaction to these kinds of scenarios by grouping/linking the people involved in different ways, or by changing some of the variables so that the situation is classed as an emergency. I can’t do justice to his arguments here, so I advise reading his paper to get a feel for the different ways in which he tweaks the variables in the thought experiments.

Now, it may well be that Thomson is right, but how confident should we be of that? Knowing which moral intuitions from which thought experiments to credit is a complex task, one that we could easily get wrong. This gives us our second possible moral error:

Second Possible Moral Error: We could be wrong about the possible scenarios under which killing is permissible. Specifically, we could be wrong about which variables to credit when it comes to weighting different thought experiments.

3. Moving forward…
So when it comes to assessing the merits of this simple argument against abortion, there is very real chance of making a moral error. The principles, concepts and considerations involved are complex and arcane, and their successful evaluation relies on considerations and weightings which we are not well-positioned to evaluate.

But that’s only the first step. Thus far we have been reasoning at the ground level of ethics. Next, we need to jump to the meta-level, and figure out what we should do about the possibility of moral error. Should we change our behaviour? Or stick to our guns? We’ll find out in part two.

Sunday, May 5, 2013

Raz and the Argument from Authority (Part Two)

Joseph Raz


(Part One)

Joseph Raz is known for his werewolf-like mane of hair, and his complex conceptual analyses of morality, law and authority. Although I have little interest in cultivating the former, I have some interest in the latter and, fortunately for me, Raz combines all three areas of inquiry in his famous article “Authority, Law and Morality”. That article provides the inspiration for the current series of posts.

In it, Raz presents his unique conception of authority, and uses it defend the Sources Thesis of law. According to this thesis, all law is based on social facts, not moral ones. As a thesis about the nature of law, it is to be contrasted with the Incorporation Thesis — which says law includes social facts and whatever is entailed by those social facts (which may include moral truths) — and the Coherence Thesis — which holds that law includes source-based law along with the best moral justification of that source-based law.

As we saw at the end of the last post, Raz’s argument for the Sources Thesis follows this pattern of reasoning:


  • (1) If law is to claim authority, then it must have certain properties P1…Pn
  • (2) Law claims authority.
  • (3) Therefore, law has properties P1…Pn.
  • (4) The Sources Thesis is compatible with law having properties P1…Pn; the Incorporation Thesis and the Coherence Thesis are not. 
  • (5) Therefore, the Sources Thesis is (probably) the correct account of the nature of law.


I accepted premise (2) the last day. This frees me up to consider the more important premises, (1) and (4), in this post. The remainder of this post looks at both, starting with premise (1).


1. The Two Conditions for Claiming Authority
In part one, I sketched the basic elements of Raz’s Service Conception of authority. According to this conception, one of the key functions of an authority — any authority — is to mediate between the subjects of the authority and their reasons for action. Two people may be in some dispute as to whether or not a certain action (call it “A”) should be performed. There are reasons weighing for and against it and they can’t decide which set of reasons is weightier. An authority can help them by performing the necessary deliberations and issuing a directive about A. This directive then replaces (in part) the reasons that the two people had for doing or forbearing from A. In this manner, the authority steps between the two people and the set of reasons; it mediates between them and their reasons.

An authority can perform this mediating function in more or less successful ways. To be fully successful, the directives issued by the authority would need to be fully justified. That is to say, they would need to give the subjects of the authority decisive reason to follow the directive. The mechanics of justification are complex, but if you’re interested, Raz’s preferred model of justification was discussed in part one. I ignore it here because what matters now is not success, but intelligibility. We do not assume that law successfully claims authority, only that it does so intelligibly. How does it do so?

To unsuccessfully, but intelligibly, perform the mediating function, two conditions must be met:

The Agency Condition: The directive issues must represent (or be taken to represent) some agent’s (or agents’) view of how the subjects of the authority ought to behave. In other words, it must represent somebody’s judgment about what the set of reasons we have for doing or forbearing from some action requires.
The Identification Condition: It must be possible to identify the directive without resorting to or relying upon the reasons or considerations on which the directive purports to adjudicate.

These two conditions are the properties that the law must exhibit if it is to claim authority. Thus, we can fill in the variables in opening argument as follows:


  • (1*) If law is to claim authority, then it must meet the Agency Condition and the Identification Condition.


Why is this? I take it that the argument for endorsing the Agency Condition is relatively straightforward. If an authority is to perform its mediating function, then its directives must represent some judgment about the set of reasons, otherwise there would be no “stepping between” the subjects and the set of reasons. It does not need to be a single agent’s view. It could be a collective body of agents, and the view of the collective may not be simply reducible to the views of its individual members. Nevertheless, it must be someone’s judgment about the set of reasons.

The argument for the Identification Condition is more subtle, but it makes sense. What Raz is saying is that the directive issued by the authority must not simply replicate, or redirect the subjects to, the set of reasons for doing or forbearing from the relevant act. If that was all that the authority did, it could not perform the mediating function. It is easiest to understand this by considering an example. Imagine two farmers who wish to divide a plot of conjointly owned land between themselves. They agree that the relevant set of reasons for resolving their dispute consists in the principles of fair and equitable distribution. However, they cannot decide which principle (or principles) should determine the outcome. They submit the dispute to an arbitrator, who gets back to them with a directive: “You should divide the land between yourselves in a fair and equitable manner”.

Clearly, the arbitrator fails to perform the mediating function in this example. The arbitrator simply redirects the farmers to the set of reasons they already knew to be governing their dispute. The reasons are not replaced. If the arbitrator is to perform the mediating function, they need to provide guidance that can be identified and understood by the farmers apart from the original set of reasons. Thus, they need to tell the farmers something like “You should divide your land exactly midway between the boundary fences” or “Farmer X should get 2/3 of the land and Farmer Y should get 1/3”. This way the farmers can know what is being demanded of them, without being forced to deliberate about fair and equitable distributions themselves.

So these are the two conditions that Raz thinks the law must meet if it is to claim authority. One problem with this is that it relies heavily on the Service Conception of authority. What if that conception of authority is wrong? In his article, Raz suggests that a weaker conception of authority results in much the same conclusion. Basically, he suggests that authorities must act for reasons, and those reasons must make some difference to their subjects’ reasons. But in order to do this something very similar to the agency and identification theses must hold true. I won’t spell out that argument in detail here. As it happens, I think the Service Conception of authority is pretty cogent (though that doesn’t mean I think it morally ideal).


2. The Case for the Sources Thesis
That leaves us with premise (4), which, if the preceding discussion is correct, must read like this:


  • (4*) The Sources Thesis is compatible with law meeting the Agency and Identification conditions; the Incorporation Thesis and the Coherence Thesis are not.


Supporting this premise are three distinct arguments. First, there is the argument in favour of the Sources Thesis. Second, there is the argument against the Incorporation Thesis, and third there is the argument against the Coherence Thesis. I’ll briefly consider each.

According to the Sources Thesis, law is identified by reference to its sources, and its sources are certain social facts. Thus, the typical sources of law are legislation, case law and custom, each of which constitutes a social fact. Raz argues that each of these sources meets the agency and identification conditions. In the first instance, they each represent some person or persons’ views about what ought to be done. In the second instance, identifying the content of the directives requires no more than a knowledge of history and/or language. For example, to know what the tax statute demands, you simply go to the text and work out its meaning.

But that’s actually not always that easy. To figure out what a text demands, you need to interpret it, and Raz’s views on interpretation are undertheorised in this particular article (this is something he corrects for in other articles, see for instance his recent collection of essays Between Authority and Interpretation). The problem is that legal texts may often simply direct subjects to sets of reasons that they must deliberate on. For example, a prohibition on “cruel and unusual punishment” or a commitment to trial in accordance “due process of the law” is not particularly informative. If they mediate between subjects and sets of reasons at all, then they do so in a very minimalistic way. They do not really represent a considered view about what reason demands in the relevant contexts.

Of course, it could be that these sorts of directives simply fail to be “law” in the proper sense of the term, but I don’t think that’s Raz’s view. That much becomes apparent when you consider his critique of the Incorporation Thesis. According to this thesis, law consists of sources plus whatever is entailed by those sources. This leads to a violation of the identification and agency conditions. Law cannot perform the requisite mediation if it consists of whatever is stated or expressed in its sources and whatever is entailed by those expressions. Raz explains this by means of an analogy:

Suppose that a brilliant mathematician, Andrew, proves that the Goldbach hypothesis…is true if and only if the solution to a certain equation is positive. Neither he nor anyone else knows the solution of the equation. Fifty years later that equation is solved by another mathematician and the truth of the Goldbach hypothesis is established. Clearly we would not say that Andrew proved the hypothesis…” 
(Raz, 1985, p. 312)

For similar reasons, Raz argues that law, qua authority, cannot be said to direct you to do or forbear from doing everything that is entailed by its directives. Including every possible entailment would prevent law from claiming authority because it would include things not reasonably attributed to the authority.

But, of course, this creates huge headaches when it comes to understanding legal interpretation, which Raz fully appreciates. For it seems that in interpreting law, judges do add all manner of entailments to its original contents. Are we saying that these entailments do or do not form part of the law? Do we restrict the contents of law to whatever the original drafters of the law believed to be entailed by their directives? Or do we allow for more creativity and innovation in legal interpretation? In the article, Raz clearly endorses the creative view of interpretation, but that creates tensions with his primary argument about what law must do if it is to claim authority. The tension is not satisfactorily addressed in this particular article. As a result, I think the case for the Sources Thesis is less than fully persuasive.

Given this, a discussion of the argument against the Coherence Thesis seems somewhat pointless, but I can summarise Raz’s thinking pretty quickly. The Coherence Thesis represents Dworkin’s view of law. According to this, law includes sources and the moral principles that best justify those sources. Raz argues that by including the “best justification” of the sources, the agency and identification conditions are violated.

You can probably imagine how his argument goes. First, he says that the “best justification” of law does not represent any agent’s considered view of what reason requires, it merely represents a placeholder for what could be a considered view about what reason requires. Second, including the “best justification” of the sources of law within the law redirects attention to the set of reasons that the law was supposed to pass judgment on. Thus, the Coherence Thesis does not allow the law to perform the requisite mediating function between subjects and reasons.


3. Conclusion
To conclude, Raz’s argument from authority says several interesting things about the nature of authority and law. It says that authorities play an important mediating function between their subjects and the sets of reasons those subjects have for doing or forbearing from some action. It says that law necessarily claims authority. And it says that in order for law to claim authority it must satisfy two conditions: the Agency Condition and the Identification Condition.

Raz then uses these propositions about law and authority to defend the Sources Thesis of law. According to this thesis, law consists entirely of social facts (sources). Raz argues that only the Sources Thesis allows for the Agency and Identification conditions to be met. However, in the preceding discussion I suggested that Raz’s defence of the Sources Thesis isn’t entirely persuasive. To become fully persuasive, a more complete theory of legal interpretation is needed, one that explains how the creative development of the law is compatible with the underlying claims about authority. Raz tries to develop this theory elsewhere, but in this particular article his views are undertheorised and so the end result is somewhat unsatisfactory.

Raz and the Argument from Authority (Part One)

Joseph Raz -- definitely one of the coolest looking philosophers out there.

Joseph Raz is one the more estimable modern moral, political and legal philosophers. His work covers a range of topics, from the relationship between law and morality, to conceptions of authority, to theories of practical reason and morality. I must confess, I find a lot of his work dull in the extreme, but it’s certainly well-regarded and people have been telling me to read more Raz for ages.

So I finally did, starting with one of his more famous articles, “Authority, Law, and Morality”. In this article, he presents his famous argument for legal-political positivism. The argument is based on a particular conception of authority, and a particular view about the relationship between legal directives and that kind of authority. It is one of the most widely-discussed and important arguments in political and legal philosophy. In this post, and the next, I want to briefly exposit and comment on it. I haven't seen this done well anywhere else on the web so it might prove useful to somebody other than myself.

The argument defends what Raz calls the “Sources Thesis”. This is one of three competing theses about the relationship between law and morality that Raz mentions three at the outset of his article:

Sources Thesis: All law is source-based, i.e. its existence and content can be determined by reference to social facts alone.
Incorporation Thesis: All law is either source-based or entailed by source-based law (e.g. if the law says “There shall be no unfair discrimination” then the content of that prescription can be determined solely by reference to social facts or whatever is entailed by those social facts).
Coherence Thesis: The law consists of source-based law together with the morally soundest justification of source-based law (Ronald Dworkin’s view of law).

These descriptions might look a little odd to those who are not familiar with contemporary views in jurisprudence (and, indeed, they may look a little odd to those who are familiar with such views since Raz’s labels are somewhat idiosyncratic). Their significance for the law/morality debate is as follows: if the Sources Thesis is true, then law is a matter of non-moral social fact; if the Incorporation Thesis is true, then law might be a matter of social fact but might also include some moral facts (if they are entailed by the source-based law); and if the Coherence Thesis is true, then morality plays a very significant role in determining the content of the law.

Raz argues that if law claims authority, then the Sources Thesis must be true. He develops this argument in three distinct phases. First, by sketching a general conception of authority. Second, by outlining a way of arguing for the Sources Thesis based on this conception. And third, by defending the key premises of that argument. I’ll to follow these three phases in my discussion below, though I’ll break the discussion down into more parts. In this post, I’ll sketch Raz’s account of authority and the basic structure of his argument. In the next post, I look at how Raz defends the key premises of his argument.

A caveat at the outset: Raz’s article is long, complex and not always perspicuously expressed. I pass over much of the complexity and detail in these two blog posts and try to isolate the core argument. This means I ignore certain discussions that might be of interest to the reader. Obviously, you should never just assume that my presentation of the argument covers all the important bits. I try, but I am both fallible and lazy, so bear that in mind.

With those confidence-inspiring words out of the way, let’s get started.


1. Raz on Authority
My neighbour, Paul, and I are having a dispute. I think he plays music too loud in the evenings, and he thinks I should “mellow out”. I want him to stop; he does not wish to stop. We both have reasons for our views. I need peace and quiet in the evenings so that I can concentrate on my work (which I often do late at night); Paul says he needs to listen to music so he can unwind after a long hard day at work.

Our dispute escalates and things start to get unpleasant. Acknowledging a need to do something to resolve the situation, we voluntarily agree to submit the dispute to a third party (an arbitrator) for resolution. The arbitrator will weigh and assess our reasons, and reach a conclusion about what should be done. We will follow her decision, since we voluntarily agreed to be bound by it. In this example, the arbitrator acts as a legitimate authority, and by paying close attention to exactly what the arbitrator does, we can learn something important about the nature of authority.

So what does the arbitrator do here? Roughly, the arbitrator takes a set of reasons for doing or forbearing from some action (in this case playing loud music in the evening). These reasons are already “live” for the two people submitting themselves for arbitration (myself and Paul). The arbitrator then assesses those reasons (and possibly some others we may not have thought of) and issues an authoritative directive. This authoritative directive then replaces the reasons myself and Paul previously had. In other words, it becomes a new preemptive reason for us to act upon.

This description forms the basis of what Raz calls the “Service Conception” of authority. According to this conception, the key role of an authority is to mediate between actors and the reasons those actors have for doing or forbearing from some activity. In other words, instead of us doing all the weighing and deliberating of reasons, the authority does the hard work for us. We just have to do as they say.

The Service Conception of authority is made up of two core theses:

Dependence Thesis: The directives issued by the authority should be based (at least in part) on reasons which apply to the subjects of those directives in the circumstances covered by the directive (Raz calls these reasons “dependent reasons”, hence the name of the thesis).
Preemption Thesis: The directive issued by the authority is a reason for doing or forbearing from some relevant act which replaces some of the preexisting reasons for actions (this might also be called the replacement thesis).

Raz is clear that by themselves these two theses do not imply that authorities supply us with decisive reasons for action. Indeed, it is central to his whole argument that authorities do not necessarily issue directives which their subjects are bound (morally speaking) to follow. For that, additional conditions must be met. In the example given, the fact that we voluntarily submitted ourselves for arbitration would supply the missing ingredients. But when voluntary submission to an authority is absent, Raz suggests that the “Normal Justification” Thesis applies. Unfortunately, his specification of this thesis in the article is an absolute trainwreck; a real paean to confusing sentence structure. My best attempt to ungarble it is as follows:

Normal Justification Thesis: The normal way to establish that a person or institution should be acknowledged to have authority over another person (the “subject”) is to show that the subject is likely to be better off (morally speaking) by following the authority’s directives than by weighing, assessing and deciding upon the available set of reasons themselves.

This is an interesting claim, and one could have many long and fruitful discussions about whether it successfully justifies authorities. But such discussions would be a distraction here. I included the thesis merely for the sake of completeness. Raz’s account of authority is highly influential and worth knowing, but to understand his argument about the nature of law, only the first two theses are needed. Bear them in mind as we proceed.


2. The Structure of the Argument from Authority
Raz’s defence of the Sources Thesis can be boiled down to this: The law claims authority. In order to intelligibly claim authority, the law must have certain properties that are typical of authorities. Those features are only compatible with the Sources Thesis. Therefore, the Sources Thesis must be true. To express that in standard form:


  • (1) If law is to claim authority, then it must have certain properties P1…Pn
  • (2) Law claims authority. (3) Therefore, law has properties P1…Pn
  • (4) The Sources Thesis is compatible with law having properties P1…Pn; the Incorporation Thesis and the Coherence Thesis are not. 
  • (5) Therefore, the Sources Thesis is (probably) the correct account of the nature of law.


Obviously, the key to this argument lies in the specification of the properties in premise (1). This is because, in addition to determining the plausibility of premise (1), those properties determine the plausibility of premise (4), and premise (4) is where the major claims about the nature of law are defended.


Still, before we spell out those properties, a word or two must be said about premise (2). Is it really true that law claims authority? Raz argues about this at length, suggesting that it is an essential property of law to claim authority. He also talks about what it means to say that law claims authority. He notes that one could say that anything claims authority if one wished, but that would be silly in many instances. The key thing about law is that, because it issues action-guiding directives, it makes sense to say that it claims authority. Things would be different if law were to consist entirely of propositions about volcanoes (Raz’s example), but it doesn’t. It consists of action-guiding directives. That’s not to say that the law never fails to successfully or legitimately claim authority — that’s a different claim — but it is to say that law can be intelligibly characterised as claiming authority.

I don’t really have problem with any of that. I’m perfectly happy to accept that law claims authority, provided we don’t assume that it is morally successful in claiming authority. Since Raz grants that exception, I’m willing to move on to the more important parts of the argument. This is something I take up in part two.