Wednesday 26 August 2015

Metanormativism is throwing up in the sink instead of doing the washing up.

This is from William MacAskill's Doctoral dissertation, titled 'Normative Uncertainty'. My comments are in bold.

'Normative uncertainty is a fact of life.
'Suppose that I have £20 to spend. With that money, I could eat out at a delightful Indian restaurant. Or I could pay for four long-lasting insecticide-treated bednets that would protect eight children against malaria. In comparing these two options, let us suppose that I know all the morally relevant facts about what that £20 could do. Even so, I still don’t know whether I’m obligated to donate that money or whether it’s permissible for me to pay for the meal out, because I just don’t know how strong my moral obligations to distant strangers are. So I don’t ultimately know what I ought to do.'

This is a good reason to hold that normative uncertainty can never be a fact of a truly ethical life but merely a fallacy that a self-publicist may strategically cultivate. Why? Well, a person whose life is truly ethical can never have disposable income for indulgence in a luxury while some people lack necessities. Thus, an ethical person never has £20 to spend on a 'delightful' Indian meal at a restaurant because she is already eating at a langar- or Community Soup kitchen-  and handing over her entire earnings to those in need.

'For an example of normative uncertainty on a larger scale, suppose that the members of a government are making a decision about whether to tax carbon emissions. They know, let us suppose, all the relevant facts about what would happen as a result of the tax: it would make presently existing people worse off, as they would consume less oil and coal, and therefore be less economically productive; but it would slow the onset of climate change, thereby increasing the welfare of people living in the future. But the members of the government don’t know how to weigh the interest of future people against the interests of presently existing people. So, again, those in this government don't ultimately know what they ought to do.'

Members of a Government are not principals, thus their own normative preferences are irrelevant, they are agents simply. If they know 'all the relevant facts about what would happen as a result of a tax', their duty is to inform their principal- viz. the citizens on behalf of whom they exercise authority. It is up to the citizens to decide how to allocate resources between generations. Once again, normative uncertainty can't arise unless members of a Government are violating their duty to act as agents, not principals, and thus are not living an ethical life.

'In both of these cases, the uncertainty in question is not uncertainty about what will happen, but rather is fundamental normative uncertainty. Recently, some philosophers have suggested that there are norms that govern how one ought to act that take into account one’s fundamental normative uncertainty. I call this suggestion metanormativism. '
Actually, in both these cases, people who are not living an ethical life are simply pretending that the reason for this is because they haven't yet made up their mind as to what type of ethical life they ought to adopt. Thus 'metanormativism' isn't normative, it is pathological. Indeed MacAskill himself writes 'Metanormativism isn’t about normativity, in the way that meta-ethics is about ethics, or that a meta-language is about a language. Rather, ‘meta’ is used in the sense of ‘over’ or ‘beyond’: that is, in the sense used in the word ‘metacarpal’, where, the metacarpal bones in the hand are located beyond the carpal bones. Regarding metanormativism, there is a clear analogy with the debate about the subjective or objective ought in moral theory (that is, whether moral norms are evidence-relative or belief-relative in some way). However, using the term ‘normative subjectivism’ instead of ‘metanormativism’ would have had misleading associations with subjectivism in meta-ethics. So I went with ‘metanormativism’ – with the caveat that this shouldn’t be confused with the study of normativity'
If you do the cooking, it is normative that I do the washing up.  Meta-normativity is like my claiming I'm actually doing 'meta-washing-up' by getting drunk and vomiting in the sink in which you piled up the dishes.
 If you are intelligent, you will say to me 'fuck off. Meta-normativity' is meaningless cognitivist shite. I'm going to beat you till you sober up and clean that sink.' 
However, if you are stupid- for example if you subscribe to comuptational cognitivism- then you are obliged to take my claim seriously. Following MacAskill's 'Maximal Expected Choiceworthiness' framework, you will be distressed to find that I am ethically superior to you because I have caused you to devote scarce resources to 'Philosophical research' which stupid people like you (i.e. computational cognitivists) consider a very good thing even though sensible people condemn it for 'crowding out' socially beneficial actions.

'There are two main motivations for metanormativism. The first is simply an appeal to intuitions about cases. Consider the following example:
Moral Dominance
'Jane is at dinner, and she can either choose foie gras, or the vegetarian risotto. Let’s suppose that, according to the true moral theory, both of these options are equally choice-worthy: animal welfare is not of moral value so there is no moral reason for choosing one meal over another, and Jane would find either meal  equally tasty, and so she has no prudential reason for preferring one over the other. Let’s suppose that Jane has high credence in that view. But she also finds plausible the view that animal welfare is of moral value, according to which the risotto is the more choice-worthy option. In this situation, choosing the risotto over the foie gras is more choice-worthy according to some moral views in which she has credence, and less choice-worthy according to none. In the language of decision-theory, the risotto dominates the foie gras. So it seems very clear that, in some sense of ‘ought’, Jane ought to choose the risotto, and ought not to buy the foie gras. But, if so, then there must be a sense of ‘ought’ that takes into account Jane’s first-order normative uncertainty.
Jane finds 2 options, which cost the same, equally good. Should she starve, like Buridan's ass or should she makes a choice based on an irrelevant alternative? Obviously, she should make a choice, finish her meal quickly, and get back to work. In this case, choosing the risotto represents compliance with a deontics that isn't 'true' because it includes supererogatory prohibitions of no ethical worth but which may have some signalling or strategic function.
There is no first order normative uncertainty here because we are told that an accessible 'true moral theory' obtains.
Decision theory is irrelevant. It doesn't matter what she eats. What matters is that she finish her meal quickly and get back to work.

'A second motivation for metanormativism is based on the idea of action-guidingness. There has been a debate concerning whether there is a sense of ‘ought’ that is relative to the decision-maker’s beliefs or credences (a ‘subjective’ sense of ought), in addition to a sense of ‘ought’ that is not relative to the decision-maker’s beliefs or credences (an ‘objective’ sense of ought). The principal argument for thinking that there must be a subjective sense of ‘ought’ is because the objective sense of ‘ought’ is not sufficiently action-guiding.
Once again, we find that the claimed motivation for metanormativism arises from the refusal to grant that some actions have no ethical or deontic status- they are 'supererogatory'. This is a good thing if Knightian Uncertainty obtains because the more 'free' choices each agent can make, the faster and more thoroughly the fitness landscape can be investigated. Suppose Knightian Uncertainty is small whereas the risk of a catastrophe is known to be high- e.g. 90 per cent. In this case, it might make sense to require that subjectivity be conditioned to show a preference for 'metanormativity' iff
1) there is always a null option- i.e. a choice which has neglibible effect
2) no scarce resources are used up as a result
In other words, metanormativism is not empty or pathological provided the people to whom it is touted can do no good but, at the margin, might do some harm. In this case, it makes sense to baffle them with bullshit.
However, there is a superior alternative. Tell them they are shite and they have a duty to resign from any responsible office or position of power or authority.

Consider the following case
 Susan, and the Medicine -
Susan is a doctor, who has a sick patient, Greg. Susan is unsure whether Greg has condition X or condition Y: she thinks each possibility is equally likely. And it is impossible for her to gain any evidence that will help her improve her state of knowledge any further. She has a choice of three drugs that she can give Greg: drugs A, B, and C. If she gives him drug A, and he has condition X, then he will be completely cured; but if she gives him drug A, and he has condition Y, then he will die. If she gives him drug C, and he has condition Y, then he will be completely cured; but if she gives him drug C, and he has condition X, then he will die. If she gives him drug B, then he will be almost completely cured, whichever condition he has, but not completely cured. Her decision can be represented in the following table, using numbers to represent how good each outcome would be: Greg has condition X – 50% Greg has condition Y – 50% A 100 0 B 99 99 C 0 100 Finally, suppose that, as a matter of fact, Greg has condition Y. So giving Greg drug C would completely cure him.

What should Susan do? Obviously, she should give him drug B. It's called 'regret minimization' or 'hedging your bets'. But, since you are a Professor of Ethics or some such shite, you aren't gonna say 'D'uh! The answer is B.' because the way you guys get tenure is by staying the stupidest possible thing.  In some sense, it seems that Susan ought to give Greg drug C: doing so is what will actually cure Greg. But given that she doesn’t know that Greg has condition Y, it seems that it would be reckless for Susan to administer drug C. As far as she knows, in doing  so she’d be taking a 50% risk of Greg’s death. And so it also seems that there’s a sense of ‘ought’ according to which she ought to administer drug B. In this case, the objective consequentialist’s recommendation — “do what actually has the best consequences” — is not useful advice for Susan. It is not a piece of advice that the she can act on, because she does not know, and is not able to come to know, what action actually has the best consequences. So one might worry that the objective consequentialist’s recommendation is not sufficiently action-guiding: it’s very rare that a decision-maker will be in a position to know what she ought to do. In contrast, so the argument goes, if there is a subjective sense of ‘ought’ then the decision-maker will very often know what she ought to do. So the thought that there should be at least some sense of ‘ought’ that is sufficiently action-guiding motivates the idea that there is a subjective sense of ‘ought’. Similar considerations motivate metanormativism. Just as one is very often not in a position to know what the consequences of one’s actions are, one is very often not in a position to know which moral norms are true; in which case a sufficiently actionguiding sense of ‘ought’ must take into account normative uncertainty as well.

A Doctor is an agent, not a Principal. The Doctor only gains salience in a Decision situation if there is a 'skill' or information asymmetry- in which case there is a dilemma re. operationalizing informed consent. . In this case, however, nothing of the sort obtains. Since Susan is posited as someone for whom advice from an Ethicist could be 'useful', it must be the case that she is as stupid as shit and thus a shite Doctor. She should resign. Why? Because it is 'impossible for her to gain any evidence that will help her improve her state of knowledge any further.' In other words, she will learn nothing from a failure. Consequently, in obedience to the Hippocratic oath, she has a duty to give the guardian of the Patient all the information quoted above, return any fees she received, and quit the role of Doctor. There is no 'normative uncertainty' here, unless she is living an unethical life and is happy to continue doing so.
Metanormativism, MacAskill tells us, is motivated by wanting to continue acting in an ethical capacity even when one knows one ought not to so act by reason of ignorance or stupidity or lack of competence. But such metanormativism isn't part of Normative Decision making any more than my throwing up in the sink is part of my duty of doing the washing up. However, as a matter of fact, not theory, if you invite me to dinner and I promise to do the washing up, what actually happens is I get drunk and vomit all over the plates you have piled up in the sink. You manage to get me into a taxi and hope you've seen the last of me. I send you an Email the next day showing, using MacAskills' 'Maximal Expected Choice Worthiness' decision framework, how my actions at your dinner party were actually highly commendable from the Ethical p.o.v. After all, you could have left the dishes in the sink for a couple of days without being greatly inconvenienced- in other words, the duty of doing the washing up at the soonest possible time was supererogatory to some degree. By throwing up in your sink, I made the action of cleaning it and the dishes that much more urgent. This tackled a lacuna in MacAskill's theory which neglects supererogatory duties. Another lacuna in his theory arises from the neglect of culpa levis in concreto type implicit delegation of duties such that a required action is better or more thoroughly or more predictably performed. Clearly, my duty of doing the washing up can be delegated to you if I am incapacitated. By throwing up in the sink, the duty of cleaning the dishes and the sink have become more urgent- you had to perform it right away. Furthermore, you are better at cleaning sinks whereas I'm good at making them dirty. Thus, my actions at your dinner party did not result in the dishes not getting washed. They were washed, probably more thoroughly than would otherwise have been the case. However, it remains the case that you may think there was a Normative failure on my part. This is quite untrue. You are actually suffering from Normative Uncertainty. You don't understand that though Metanormativism has nothing to do with Normative Behavior, nevertheless, if MacAskill aint talking utter bollocks, by causing you to devote more resources to a purely philosophical argument- viz. my claim that my behavior at your dinner party was super ethical- I am advancing the cause of Ethical Altruism which is a true Moral Theory.
As MacAskill says 'Moral philosophy provides a bargain in terms of gaining new information: doing just a bit of philosophical study or research can radically alter the value of one’s options. So individuals, philanthropists, and governments should all spend a lot more resources on researching and studying ethics than they currently do.'
By throwing up in your sink, and then sending you this email, I have caused you to devote more resources to 'philosophical study' and thus made you an immeasurably better man. Thus getting drunk at dinner parties and throwing up in the sink instead of doing the dishes is prescriptive for Effective Altruists provided Normative Uncertainty is ubiquitous or computational cognitivism aint shite.

8 comments:

Anonymous said...

MacAskill's dissertation is about what norms arise if first order normative uncertainty obtains- for example where x has a duty to y iff y is a member of Z but membership of Z is not known with certainty. Notice, x may still have a duty- a different one- if y is not in Z.
Clearly, such situations arise all the time. Suppose you work for a Chain Store with a 'no questions asked' return policy for its customers. A member of the public comes to you to return an item but lacks a receipt. You are in doubt as to whether that person really was a customer. They may have shop-lifted the item or stolen it from someone else. If so, your giving them a 'no questions asked' refund helps them do something immoral. On the other hand, refusal to give the refund may cause them suffering if they are poor or ill.
In practice, you may decide to be strict and to call the police and charge the customer with attempted fraud, or you may be magnanimous and pay the person in question.
Your beliefs and intentions and 'cognitive biases' are all relevant data, as are the consequences of your choice of action. MacAskill provides a decision theoretic framework for the underlying calculus which, by clarifying and thus quantifying aspects of your option, enables you to make a better informed decision.
At the individual level, no doubt, it may be too costly actually do the computation. However, at the aggregate level it would be cost-effective to do so. Thus, the Marketing Dept. of the Chain Store would find it worthwhile to commission research to find out the true costs and benefits of the 'no questions asked, Returns' policy.
MacAskill isn't looking at trivial situations- e.g. doing the washing up if someone else cooked the meal- but major life project decisions- e.g. which career to pursue in order to maximize benefit to others.

windwheel said...

There is no first order normative uncertainty here. Either you are an agent- in which case normative uncertainty can be resolved by consulting your Principal or making an 'expert cognition' type decision not to do so (because such consultation is known to you to be counter-productive because the result is that innocuous or supererogatory actions are turned into binding duties because of your query) and thus setting uncertainty to zero by your own volition. Alternatively, you are a principal. In this case, it is up to you to set your own norms. If you believe you are bound by a deontics, that deontics must be well enough specified to leave no room for normative uncertainty, otherwise you are not in fact bound by any deontics at all. Metanormativism in this context is bad faith simply. You are pretending to be bound by a deontics when this is not in fact the case. Your soi disant 'ethical life' is just mauvaise foi posturing.

BTW throwing up in the sink after having promised to do the dishes is too a life project- that of yours truly. MacAskill's dissertation proves this life project of mine to be prescriptive for Effective Altruists subject to my resource constraints.

Anonymous said...

Thank you for your quick reply. I'm afraid you miss the point. Whether as an agent or as a principal, real world subjects generally have only a fuzzy notion of the correct 'first order normative theory' binding upon them. Instead, MacAskill says, they have a Credence function over a set of such theories each of which has a 'choice worthiness' function relating to outcomes.
Since no rational person has any credence in your own first order normative theory, which assigns a high value to vomiting in the sink, it can be discounted.

windwheel said...

If agents have a non-fuzzy credence function for a 'first order normative theory' how can they have a fuzzy understanding of it?

Anonymous said...

Having looked at the dissertation, I see you are right. MacAskill is assuming cognivitism is correct. Still, consider the following situation- you are employed by an organization which requires employees to behave in conformity with a range of first order normative theories for which they are to choose their own credence function and be judged accordingly.
Something like that surely does happen in the real world.
In any case no first order normative theory would gain credence such that vomiting in the sink is prescriptive.

windwheel said...

How do we know there isn't a first order Normative theory which prescribes whatever we ultimately to do? In positing a credence function in a bunch of theories, how do we know there isn't another theory out there which 'sublates' them and in which we have 100 percent credence though we don't know what it is?
First order normative theories can rule out vomiting in the sink. But MacAskill isn't giving us a first order theory. He is pretending that 'Metanormativism'- such that doing a bad thing like vomiting in the sink, can turn out to be a very good thing provided it motivates some supposedly highly beneficial 'Philosophical Research' into his own silly theory.

Anonymous said...

I think you are confused about what 'fist order Normative theory means'. Think of it as a list of rules, with the first having to be fulfilled before the second can be implemented. In the real world, we may have a doubt as to which rule to start with and the credence function captures that information. Ultimately, our choice will be explainable, or captured by an algorithm, by a string of second order rules. E.g. start with Utilitarianism's rule set till you come to a situation where someone might be caused mortal harm. Then switch to the Kantian rule-set.
In other words, what you end up with is couched in 'second order' terms- propositions about the appropriateness of first order theories.

windwheel said...

The author states- 'An option is a proposition, understood as a set of centred possible worlds,that the decision-maker has the power to make true at a time. A centred possible world is a triple of a world, an agent in that world, and a time in the history of that world: an
option therefore can include the action available to the decision-maker, as well as the decision-maker’s intention, motive, the outcome of the action, and everything else that could be normatively relevant to the decision-maker’s decision'
Clearly 'regret' is something that comes under the rubric.
The author states' A first-order normative theory is a function from decision situations to an ordering of the options in the set in terms of their choice-worthiness,where the function’s domain is all possible decision situations. The theories form a partition. I take choice-worthiness to be defined as the ordering that first-order normative theories produce.'

Clearly a regret minimizing function exists and is first order- because of the way option has been defined- even if the way we calculate it involves- as you suggested- seemingly 'second order' references to first order theories.

I will write a separate post showing why ignoring 'regret minimization' utterly vitiates the author's argument.