Sen on Hare's universal prescriptivism.

HUME'S LAW AND HARE'S RULE

AMARTYA K. SEN

THIS note claims that contrary to his denial,' Mr Hare's 'adherence' to 'Hume's Law' conflicts with his adherence to 'universal prescriptivism'.

You can say 'no ought gives rise to an is' while also saying 'there is a cat which is a dog'. That's what Hare did. He said ' whoever makes a moral judgment is committed to the same judgment in any situation where the same relevant facts pertain.' He didn't use the word ought. He said is. He was wrong. A cat isn't a dog and a moral judgment is not a 'universal prescription'. 

Or, to put it differently if value judgments are indeed universalisable, then there is at least one value judgment that follows from exclusively factual premises.

All value judgments could be said to follow from an exclusively factual premise- viz. that God exists and is the only efficient cause and implants value judgments in agents of various sorts. But this is like the factual claim that cats are actually dogs who cunningly disguise themselves for some fell purpose. This is a factual claim but a deeply silly one which we think is wrong.  

To keep the argument simple, we shall take a case of a nonmoral value judgment discussed by Hare, where the pattern of universalisability is obvious. Consider the following two statements. A purely descriptive one, to be called C: 'This motor-car is exactly like the next motor-car'.

 If a person thinks 'x is like y' we say this is a subjective reaction on her part. It is not factual. I may say 'cats are like dogs. As far as I am concerned a cat can do for me anything that a dog can do.' You may say 'cats are not like dogs at all. You can't play fetch with a cat in the same way you can play fetch with a dog.' 

The second, to be called V, is a value judgment: 'The next car is as good a motor-car as this one'.

What a person thinks is good is subjective. I may say that broken down jalopy is just as good to me as a brand new Rolls Royce because I'm not materialistic and don't covet what is more expensive or that which has superior functionality'.  

Examine now the following statement of Hare: ' I cannot say "This is a good motor-car, but the one next to it, though exactly like it in all other respects is not good".' 

This may be a factual claim- Hare is simply incapable of making this statement. It may be a value judgment- Hare thinks no one ought to make this statement.  

Does this violate Hume's Law?

In the latter case, yes.  

Not directly, for what is being claimed here is that if I do make a value judgment, which we may call V* (that this is a good motor-car), and accept a factual statement, which we may call C (that the next one is exactly like it in all descriptive features),

this is not a factual statement. Like is like good.  If 'descriptive features' are identical, then by identity of indiscernibles, there is identity. However, value judgments don't have to obey Liebniz's law. 

then I cannot make another value judgment, V** (that the next motor-car is not good).

You are welcome to make a factual statement 'people always obey Liebniz's law when making imperative statements'- which others can show is false.  

We can look at the argument slightly differently. If the first car is good and the next car cannot be called 'not good', what is being shown here is that I cannot deny that the next car is as good a motor-car as this one, i.e. I cannot deny V, as defined above.

And yet this is easy enough to do. I'm welcome to say I like or think good only such and such car even though it is indiscernibly identical to other cars rolling off the assembly line. You may say people ought not to say such things because it offends your ideas about how imperative logic should work.  

So it is being claimed that I cannot say, 'V*, C, and not V'. Would Hare permit us to drop out V* from this, and say 'C and not V'? Can we say, for example: 'This is a medium quality car. The one next to it, though exactly like this one in all other respects is not as good as this one'? Clearly, according to Hare we cannot say this either, because good being a 'supervenient epithet', two objects cannot differ only in 'goodness' being exactly alike (L.M., pp. 80-81, 130-1). So V* is really inconsequential in this context. What, according to Hare, we cannot say is simply: 'C and not V'.

If it is 'cannot' then it is a wrong factual statement. If it is 'ought not' then it is a value judgment. Either way, there is no conflict with Hume's law.  

In truth-functional logic,

either existence is a requirement for truth- in which case there is objective verification or falsifiability for valid propositions- or else truth is imperative and unverifiable. In this case we have a deontic logic which is wholly subjective. 

to deny 'C and not V' is to assert 'C implies V'.

Only if C and V are well-formed propositions. But, if they are, then there can be no violation of Hume's law.  

But this inference cannot be immediately applied to the logic of value statements,

any inference at all can be applied to what is wholly subjective. You are welcome to say 'If I think x is good I should infer x is bad'. This is because values are 'intensional'. They don't have a well defined 'extension' and thus don't obey Liebniz's law of identity.

as it might be Hare's intention to claim that we cannot say 'C and not V' without committing us to saying 'if C then V'. He might mean to suggest that if we say C, we cannot say 'not V', but we need not affirm V either. However, it is easily confirmed that this is not what Hare is suggesting; he is saying that we are committed to affirming V in such a situation. 'And if I call a thing a good X, I am committed to calling any X like it good.' (F.R., p. 15.)

Presumably commitment is about what one 'ought' to do. But this does not arise from anything factual. It arises from the notion that one ought to think in a manner consistent with a particular logic.  

 

This settles one problem but once again this looks like an argument with a value premise (this is a good X) and a factual premise (another X is exactly like it) with a value conclusion (the other X is also a good X). But it is easy to show (just as in the last para- graph) that what is being asserted is 'if C then V' (substituting X for motor-cars), i.e. if I agree that A is like B as an X,

 I am making a value judgment.  Cats are like dogs as pets because I like to play with cats in the same way I do with dogs. But, I understand that you value playing fetch with your dog and would reject a cat as a substitute. This is a matter of taste. 

then I am committed to saying that A is as good an X as B.

only if I make the value judgment that my imperative logic must have Liebnizian rather than dialectical rules.  

We arrive, therefore, at the position that according to Hare if we do say C, not only can we not say 'not V', but we are committed to saying V.

only in the manner I've outlined. Sen is simply wrong.

But this means that there is a value judgment that can be derived from an exclusively factual premise.

No there isn't. Two motor cars aren't identical for deontic logic. One is owned by me which is why I pay to wash and service it. The other is owned by you. I don't give a shit what happens to it.  

This is not transparent because superficially the argument has the form: 'if V * and C, then V * *', when V * stands for the value judgment 'A is a good X', C for the fact that A and B are exactly alike,

if Sen thinks this is a fact then he would wash and service every car of the same model and year as his own 

and V** another value judgment 'B is a good X'. To see how this violates Hume's Law, we can go in two steps. First we convert the argument into affirming: 'if C, then V* implies V**', i.e. 'if A and B are descriptively alike, then A being good implies B being good'. This itself is a violation of Hume's Law,

No. It is a violation of the rules of logic. Only if A and B are indiscernibly identical does the proposition follow. If you sleep with your wife's identical twin sister you are not considered to have done a good thing. 

but since the conclusion is in the hypothetical form, this might look like a trivial violation of it. (In fact, it is not a trivial violation; contrast the much-discussed trivial case: 'if x then y' implies 'if y ought to be avoided than x ought to be avoided'.

This is silly. Farting at the dinner table should be avoided. Yet eating food implies I will toot a lot. This does not mean I should avoid eating. It is better that I toot then that I starve. True, I could go fart in the toilet but fuck that- right?  

) We can, however, replace 'good' by 'medium quality', 'poor', 'high class', or any other value expression; and

still have only a value expression

consistent with 'universalisability' of all these value judgments, Hare will confirm the statement in each case. So that the conclusion is really more general than 'A being good implies B being good'. It is of the form 'A is as good as B', a relative evaluation of the two irrespective of the absolute value of A's goodness. Thus the argument is indeed one, as claimed earlier, represented by 'if C then V'.

But C is a value judgment and V is a value judgment so there is no violation of Hume's law. Sen's mistake is to think 'mediocre' isn't just as much a value judgment as 'good'. 

The source of this problem is this. Hare defines 'good' and 'better' in terms of 'ought' (L.M., section 12.3), and 'ought' in terms of an implied imperative (L.M., pp. 168-9); this fits in with 'prescriptivism'. But because of 'universalism', the notion of 'goodness' as studied by Hare cannot be independent of descriptive features.

It is independent of OBJECTIVE descriptive features. If they are subjective then there are no facts, there are merely opinions or matters of personal taste. Paranoid maniacs can find descriptive features in their hallucinations. But those descriptive features are not factual. They may however relate to the values or spiritual or other such beliefs of the person suffering from them. 

Hare states this quite clearly himself, but points out that there is no unique relation between descriptive features and their 'goodness', and criticises the naturalists for tying value judgments 'analytically to a certain content' .

Nothing wrong in that. It is obvious that some people may have values which others consider 'unnatural'. I see two men kissing. I think this is good because it shows that homosexuals are finding love and companionship. You think it is unnatural or even diabolical.  

But thanks to universalism he has to accept that if two objects have the same descriptive features, they cannot differ in goodness;

only for someone whose subjectivity is constituted in a particular way 

and even this modest claim violates Hume's Law.

No it doesn't. It is merely a claim about particular subjectivities. I may not believe angels exist but I may think that if you think St. Michael is just as good as St. Peter, then you should also consider votaries of the former to be as pious as votaries of the latter. But you would be right to point out that my views in this respect are arbitrary and lack any factual or logical basis because Liebniz's law can't apply to the relevant intensions. 

Hare seems to overlook this because he poses the problem in the model of 'if V* and C, then V* *', as discussed earlier. But thanks to universalisability of all value judgments, as shown above, this amounts to claiming 'if C, then V', when C is a factual statement and V a nontrivial value judgment.

C isn't factual. 'Like' and 'Good' are imperative not alethic.  Of course, one can have the subjective view that subjective preferences should be transitive or 'universalizable' or pleasing to God or whatever. 

Hare devotes a lot of time

time utterly wasted 

to distinguishing his position from that of the naturalists. In this context, he points out that 'for a naturalist the inference from a non-moral description of something to a moral conclusion about it is an inference whose validity is due solely to the meaning of the words in it'. (F.R., p. 21.)

but the meaning of words is whatever the hearer of those words decides. The thing is subjective or arbitrary. 

Accepting this description of a naturalist (and bearing in mind that the point at issue here is not the distinction between moral and other kinds of judgments), we can define a naturalist as one who claims that the inference from a factual statement to a value judgment is 'due solely to the meaning of the words in it'.

Only because we can also define a naturalist as one who shoves his own head up his own arse when nobody is looking.  

We can distinguish between two kinds of naturalists: 'existential naturalists', those who claim that at least one value inference can be made like this, and 'universal naturalists', viz. those who claim that all value inferences have this property. What is being claimed here is that Hare's position is an existential naturalist one.

as opposed to an existential naturalist one with its head up its arse 

The violation of Hume's Law follows from 'the way in which the word "good" functions' (L.M., p. 130).

to people who are violating the laws regulating the degree to which the spinal cord can bend by shoving their heads up their arse. In other words, no fucking law is being broken. This is just loose talk is all.  

The point can be put a little more precisely using elementary concepts of set theory.

No it can't. A set has to be well defined- i.e. there can be no ambiguity as to whether or not an object belongs to it. Sen & Co continually commits the 'intensional fallacy' and treats things which are not sets as if they are sets.

Consider the three following sets, S. T, and U. Each object to be compared is given a different number to identify, and the set of these numbers, each standing for one object, is S.

This is true only if there is a set of 'objects to be compared'.  

Each object has certain descriptive features.

'certain descriptive features' is vague. There is ambiguity as to what qualifies. Moreover, 'descriptive' is epistemic. It changes as the knowledge base changes. Thus there is no

set of all possible combinations of descriptive features, we call T. Finally, 'goodness' is represented by a set of numbers U,

this can only be done arbitrarily. Even so, the thing is epistemic or otherwise suffers from the intensional fallacy. This is not a set. 

such that a higher number represents 'better'. (These numbers need not have cardinal properties.)

So- there is a ranking or partial order- or would be, if there was a set or class or unique pre-order. But, in that case the Szpilrajn extension theorem applies and so cardinality comes in anyway.  

In terms of 'ought' statements, this corresponds to saying that if one has to choose between two objects with two different 'goodness numbers',

one is living in 'Imaginationland' and also has one's head up one's fucking arse. I can say 'if Sen has to choose whether to breathe in or breathe out he has to choose between devouring the turds of a Dalmatian and an Alsatian. This a wholly arbitrary statement. I can give some cockamamie reason for it- e.g. appealing to Yoneda lemma- but that doesn't change that it is merely ipse dixit bullshit

and one cannot have both, then one ought to choose the object with the higher number.

just as, if Sen chooses to breathe in, he ought to choose to devour Dalmatian turds not Alsatian turds.  

The universal naturalist will claim that by virtue of the meaning of the words, there is a unique 'transformation' of T into U.

While the Socioproctologist will stick to his claim re. these nutters eating only dog turds.  

If someone does not agree that one collection a of descriptive features (i.e. one element in T) must be related to one particular value P of goodness (i.e. one element in U) given by this unique transformation, then he is revealing a cognitive defect.

These guys are as stupid as shit. They don't even realize that they are eating nothing but dog turds.  

Hare does not claim this and does not, therefore, advocate universal naturalism. But his 'universal prescriptivism' does assert that while each of us can entertain a different transformation from T to U, the relation between the set T and the set U must be, for each of us, one of transformation (in the strict sense) from T to U.

It can't be any such thing because neither T nor U are sets. Both represent 'intensions' of an epistemic kind. The extension is unstable and ambiguous.  

That is for each element in T there is (for each of us) one and only one element in U (though for each element in U there might be more than one element in T). This means that if we take two objects, i.e. two numbers from the set S, but if they correspond to the same element in T, i.e. have the same descriptive features, then they must correspond to the same element in U, i.e. be 'equally good', or one 'as good as' the other.

This is why I can claim that Amartya Sen eats only dog shit. I arbitrarily assign the same number in some set of my own invention to dog turds as any given item in his diet.  

To illustrate, in the old example, the first car and the one next to it represented two different elements in the set S, but having the same features, corresponded to the same element in T, and thus, thanks to universalism, had the same goodness number, or were equally good.

They may have had the same price or resale value. That is an objective and factual matter. However, different people would consider their own car- or that owned by the priest or Doctor- to have more goodness than that which is being used by gangsters to do drive-by shootings. We may say 'use drone strikes to destroy the bad cars.'  

Thus, while universal naturalism requires that there be a unique transformation of T into U, Hare's position requires only that each relation between T and U must be one of a transformation of T into U, though not necessarily a unique one shared by everybody.

Because the thing is subjective and imperative, not objective or alethic. 

But this violates Hume's Law, because two elements in S (two motor-cars), because they correspond to the same element in T (have the same descriptive features),

They do so subjectively for some possible person. There is no violation. One can make value judgments about things which exist just as much as one can do so about things which don't exist- e.g. unicorns and vampires.  

must be judged to correspond to the same element in U (be regarded as good as each other).

by some person assumed to behave in a particular way. I assume that Sen eats only dog turds. I then stipulate that any item in his diet is less preferred by him than a steaming piece of puppy poo. I infer that he quietly slips away from High Table and quickly devours dog poo. His fellow diners don't notice. But Socioproctological Induction uncovers Sen's dirty little secret.  

And it is in this sense that Hare's position is an existential naturalist one.

It isn't. It is some silly attempt to be a bit Kantian while also sucking up to the Logical Positivists.  

We should now briefly refer to the background question whether two separate objects (or situations) can be exactly alike.

Indiscernible identity is sufficient for Leibniz's law to function.  

For a variety of reasons this problem is not a fundamental one for our purpose. Firstly, for Hare's universalisability, they need not actually be exactly alike;

which is why it is stupid shite. It amounts to saying, if you wipe your own bum you ought to wipe everybody's bum. It ignores 'uncorrelated asymmetries'. Still, I suppose, Sen was finding that his wife was alike enough to the wife of his best friend- which is why he'd run off with her.  

if they are thought to be so, that is sufficient. Secondly, Hare also considers cases where two objects differ in some respects that are irrelevant to the choice (F.R., pp. 140-41). Provided they are exactly alike in other respects, universalisability is applied. The notion of relevance brings in a possible ambiguity, in case there is disagreement about what is relevant.

Ambiguity means there are no well-defined sets. 

But this does not make Hume's Law any more valid.

Yes it is does. Anybody can arbitrarily assert that the Law of Gravity is not valid. You levitate when nobody is looking. But there is no proof of this assertion. Where is the counter-example? That's the problem here as well. 

Note the following argument to which Hare is committed : 'A and B have the same descriptive features except in respect R' implies 'If respect R is irrelevant to the choice, then A and B are equally good'.

This is a 'sequent calculus' of an imperative type. Its partial tautologies don't hold if tautologies themselves don't hold because of intensionality. 'Relevance' depends on the knowledge base. It changes all the time. All that we have here is a provisional or contingent decision which may have been good enough for a particular purpose- till it was discovered not to be. 

The conclusion is a conditional value judgment depending on another value judgment implied in the notion of relevance, but the condition is such a weak one that it does not make the conclusion trivial in any sense.

It means that the 'conclusion' is just an educated guess, nothing more.  

(It is like concluding in the motor-car argument: if the relative situation of this car and the next one on the shop floor is irrelevant to their relative goodness, and they are exactly alike otherwise, then they are equally good.)

This is a best guess. Still, we know one might turn out to be a 'lemon' which is why there is a market for extended warranties etc.  

The force of this conditional judgment is hardly less than that of V in our exact-likeness case.

The thing is just an educated guess.  

Lastly, we need not really spend sleepless nights on whether A and B can be relevantly alike. If they cannot, then Hare's principle of universalisability is empty of content.

No. It's just a simple ethical idea about how it's nice to be nice. If you like living happily you should understand other beings too want to live happily.  

If they can, then more than one element in S can correspond to the same element in T, and the problem discussed here arises.

What problem? Sen's ignorant assertion that 'like' is objective and 'good' isn't? The fact is, only if you have a structural causal model in which there is perfect substitutability or else 'identity of indiscernibles' (e.g your physical theory can't distinguish individual elementary particles) can you have Liebnizian identity and thus analytic or tautological statements. Otherwise, there's always a chance that the law of identity is violated. 

A more formal statement of the proposition being proved here is: Either Hare's principle of universalisability is empty of content,

it isn't. It's just saying 'its nice to be nice' or 'do unto others'. 

or it conflicts with Hume's Law.

It doesn't. There can be 'mixed' imperative and alethic propositions that are informative. They just aren't part of either an alethic or an imperative sequent calculus. 

The plain fact is, what we thing good or good enough or relevant depends on our preferences or values or interests.  

I should emphasise that I am not arguing here against Hare's 'universal prescriptivism' which is certainly among the most fruitful approaches to ethical discussions.

This guy would go nuts if he read the Sermon on the Mount.  

I am only objecting to combining this approach with a claim of strict adherence to Hume's celebrated 'law'. It does not worry me unduly to think that Hare's universal prescriptivism implies an existential naturalist' position,

it may imply anything it likes to a fucking cretin 

but it worries me to think that this implication is denied.

what should have kept Sen up at night back then was that his folk were at high risk of famine and genocide.  Instead he was writing ignorant nonsense and seducing his best friend's wife. No wonder he is considered the Mother Theresa of Economics.