Amartya Sen's first Dewey lecture

 In his 1984 Dewey lecture, Amartya Sen stated that

An agent is a being with the capacity to act, and ‘agency’ denotes the exercise or manifestation of this capacity. The capacity to act is biological and economic. It is a potential to do certain things. As 'well-being' rises, such potential rises. A healthy person with money can do more than a sick person who has no money. However, both would not want to do things which someone else can do better even if they are themselves the beneficiary. True, a child may say 'I want to get dressed by myself.' This is because the child wants to develop a particular capacity. There are things we may want to do on our own even if, once we have mastered the skill, we are content to let someone else do it. This follows from the theory of comparative advantage. It is one thing to develop a capacity 'just in case' we have to depend on ourselves. It is quite another to do something for which we have a higher opportunity cost than someone else. Thus I may want to learn how to cook a particular dish. But, supposing I am a busy business executive, I might prefer to have it prepared for me. I would say 'part of increasing my well-being is my learning to do things which I find interesting. But, because of the high opportunity cost of my time, I do not want to exercise that particular type of agency.' 

Speaking generally, agency does not matter. It does not matter how a thing is done unless there is some utility in doing it- e.g. a guy who likes cooking gaining pleasure from cooking a meal- but disutility- i.e. the opportunity cost of doing something- may be much higher which is why the busy business executive doesn't actually do any cooking though it is something he enjoys. The reason for this is that by not cooking, he gets higher utility by working in his own field of specialization.

Once one understands that disutility is opportunity cost, we need not bother with 'agency'. It is irrelevant. The principal has agency even if he employs an agent to get something done. Only utility matters and that is what 'well being' is about. Freedoms are Hohfeldian immunities concerning the disposal of specific things- e.g. your time, property you own, casting the vote the law assigns to you, standing for offices for which you are eligible etc. This freedom exists even if it is delegated in a lawful fashion. What matters is whether there is a legal remedy for violations of a Hohfeldian immunity. That is a justiciable matter. Courts will not say 'x is no longer free because x has voluntarily entered a Rehab clinic with onerous rules.' They will say 'x suffers from an addiction. To be cured of this she has voluntarily entered a clinic which, in some respects, is like a prison. But x is the principal and the clinic is the agent she has hired for a specific purpose. She is free to leave. The complaint, made by her drug dealer, that she has been falsely imprisoned does not hold water. She is free to either overcome her addiction by herself or by employing an agent.' On the other hand, suppose this lady has entered a cult, been subjected to brain-washing, and thus no longer has free will. In these circumstances, if her kin can prove she suffers harm and is no longer able to exercise her free will, the Court may order her removal from the premises of the cult and appropriate medical or psychiatric treatment. These examples show that only utility matters when it comes to freedom. 

I suppose one might say 'a Nanny State which does everything for you denies you agency. You are not truly free'. But, if you can emigrate, this is not really the case. It is just that one jurisdiction may have fewer Hohfeldian immunities than another. In Texas, if you hear a prowler in your house, you can shoot the fellow under 'stand your ground'. In New York, there is a 'duty to retreat'. You call the police and they may open fire. 

The second thing which Sen gets wrong is the notion of 'informational constraints'. This relates to the cost of acquiring additional information. If it is too high, decisions must be made on the basis of information known to be incomplete. One consequence is that we treat dissimilar things as though they were similar because, for our purpose, this is a good enough solution. Consider Syria and Afghanistan. We know there are differences between the two countries. Yet, for our purpose, we may predict that the new regime in Syria will go the way of the Taliban in Kabul. In other words, the new regime in Damascus will drop the pretence of having given upIslamist ideology in favour of pragmatism. We can imagine a person of Syrian origin discussing whether she should return to Syria with a friend from Afghanistan. 'Don't do it' the friend says. 'I know your country is different from mine but the people who have gained power there are similar to the Taliban.' This may be 'rough and ready' advise, but it is fit for purpose, given our existing information set. 

Sen thinks the information constraint has to do with 'universalizability'- i.e. the moral principle that states that actions are only morally acceptable if they can be applied to everyone in similar circumstances. However, this means there is a criteria of eligibility. The burden of proof falls on the person disobeying the law to show why he is not eligible to be judged under it. Generally speaking, information asymmetry means he has low cost for providing such proof. 

But we know that additional information will show that things currently treated as alike- e.g. Syria and Afghanistan- aren't really alike. Still, if the outcome will be the same in both countries for a particular group- e.g. women- then, for a narrow purpose, we may well treat them as alike. What Sen is talking about is Leibniz's 'identity of indiscernibles'. But this has no application to epistemic objects- i.e. things which change as more is known about them. A person with great expertise on Syria might say 'such and such district in Syria will be perfectly safe for such and such group.' Still, broadly speaking, they may agree that Syrian refugees need to be a bit cautious about returning home.

Sen quotes Sedgwick who observed that if something is right or wrong for me but not for you this must be because of some difference in the two cases rather than a difference between persons. However, what is different is that two people may have different 'uncorrelated asymmetries'. Thus, in law, Queen Victoria could do no wrong. This was because she was the Sovereign. Sedgwick could do wrong. He wasn't the sovereign. 

So, either this is a tautology- viz. 'irrelevant stuff is irrelevant' or it is the 'intensional fallacy' writ large. The fact is, when it comes to 'epistemic' (i.e. knowledge based) objects- e.g. preferences, or expectations- then Leibniz's law of identity does not hold. All information may be relevant even if it is not obvious why this is the case. Consider the Monty Hall problem- Suppose you're on a game show, and you're given the choice of three doors: Behind one door is a car; behind the others, goats. You pick a door, say No. 1, and the host, who knows what's behind the doors, opens another door, say No. 3, which has a goat. He then says to you, "Do you want to pick door No. 2?" Is it to your advantage to switch your choice? The answer is yes. Your expected probability distribution has changed because of new information. After all, the TV host would not have opened the door with the car behind it. That would have spoiled the game. 

It must be said, for practical purposes, we may assume 'independence of irrelevant alternatives'. Thus suppose I have an ice cream van. As a rule, I decide to go out on sunny days not rainy ones. However, if there is a pop concert or football match, I might make an exception to this rule since some extra information has become available to me. 

Actions are never identical. Motivations and consequences and uncorrelated asymmetries (e.g. whether you have the legal right to dispose of a particular thing because you have legal ownership of it) play a part. Horizontal equity says that like cases should be judged the same way. But vertical equity says unlike case should be judged differently. Additional information can show that like cases are actually unlike. This is what keeps lawyers in business. 

Sen has made the claim that 'Hume's law' (no 'ought' can be derived from an 'is) conflicts with Hare's 'universal prescriptivism. As I have pointed out, there can be no conflict. Hare is welcome to say there is a cat which is a dog. We are welcome to say he is wrong. It may be that science will come up with a genetically engineered creature which can reconfigure itself as a dog though most of the time it is a cat and displays feline characteristics even when it turns into a dog. But, currently, we would probably be right to insist that no cat is a dog. 

Moral approaches have one advantage. They are predictable. Moreover, they seek to be 'robust'- i.e. change little as information changes.  That is why it is worth acquiring a reputation for morality. Of course, there may be hypocrisy or else people may be reluctant to spend money doing what they claim is the right thing to do. 

Sen thinks otherwise. He does not understand that morality says if you get married you must not run off with your best friend's wife just because you got new information which inclines you to think you would be happier with her. 

This is nonsense. Utilitarianism can focus on 'disutility' (i.e. opportunity cost) and 'transferable utility' (money). The Government can do things which raise productivity and thus increase tax revenue. If it does not, it may go off a fiscal cliff. In the Thirties it was important to get rid of 'interpersonal comparisons' because this could lead to the physical liquidation of entire classes or races of people. Equally, economists needed to be able to predict market behaviour. They needed to focus on 'revealed preferences' and thus estimate the elasticity of demand and supply as an empirical matter. 

Moral approaches had the virtue of being predictable and robust no matter what new information became available. A moral man won't ditch his wife just because someone more attractive came along. This was good for the family unit and thus the wider community. 

I suppose one could say 'the hedonist has his own moral code. If the wife knew she was marrying a hedonist, she has only herself to blame if he ditched her for a strumpet'. Here, what is important is predictability. The wife may say 'I've married this hedonist because, apart from strumpets, he also likes money. He dare not ditch me because this is a 'community property' state. I would take him to the cleaners in a divorce court'. 

Is utilitarianism a 'moral' theory? As intended by Bentham, yes. Moreover, it was 'progressive'. Diminishing marginal utility of money meant you could have progressive taxes on Income and Wealth. Many considered this highly moral at a period when the great mass of the people led miserable lives. 

Can 'morality' be factorized? Yes. One can say certain moral principles have precedence over others. In technical terms, a particular morality has an expression in deontic logic. This is likely to have a built in ramified type theory. However, the use of this deontic logic may be arbitrarily 'buck stopped'. In fact, this is what happens in Courts of Law. At some point, there is ''Res judicata'. The case has been decided an no further appeal is allowed. But this is a rough and ready, pragmatic solution which may fall short of 'harmonious construction'. 

Act Utilitarianism is the moral theory that assesses the morality of an action by considering its consequences and how it affects others. This can be made justiciable by the terms of a contract or the law. Alternatively, there could be a 'Cost Benefit Analysis'. Information about consequences may be elicited from concerned parties and a calculation may be made. There is no 'factorization' here. True some consequences may be remote while others are directly received. But all consequences of which the Court or Commission of Inquiry are notified can be weighed up. True, there may be legal or political issues- e.g. lack of jurisdiction or 'doctrine of political question'- which prevent a judgment being given effect. That is a separate matter. 

This is gibberish. 'Welfarism' is the same as 'Act consequentialism'. The 'goodness of information' is not a factor. Either the information is reliable or it must be rejected. Adding things up does not depend on the 'goodness' of information. Outcome utilitarianism is the same thing as Act Utilitarianism. Not doing the right thing may be just as culpable in law and morality as doing the wrong thing.

Sen thinks 'Welfarism' is indifferent between states with the same total utility. This is not the case. Consider a person who says, with good reason, 'My life is horrible. Frankly, I'd be just as well off dead.' Welfarism would say his living, as opposed to dying, is intrinsically worthwhile. While alive, even if his well-being is zero, still being alive is intrinsically valuable. The difficulty here is that welfarism could turn into an unthinking type of pro-natalism. 

Pluralism is like dialethia- i.e. allowing for contradictory things to be equally true. It can be useful but it can also be an excuse for endless jibber jabber. 

Still, one may say most people are moral pluralists. At work, they may be focused on economic goals. At home, their behaviour is motivated by love and affection. On Sundays, when they go to Church, it is spirituality which they hold as the highest good. There is no great difficulty in 'compartmentalizing' our lives. We apply different principles when doing different things. Spiritual goods may appear incompatible with material prosperity. But, if you earn more, you can give more to the Church so that Spiritual goods become more widely available. 

Marx said 'to each according to his contribution' till scarcity is abolished. This was actually quite useful at a time when some thought highly skilled workers should agree to take home less money so that the less skilled, or those with bigger families to support, could have a decent income. Berlin's point was different. At the time when he was writing, many believed Kruschev's boast that the Soviet people would soon have a higher material standard of living than the Americans. Even if this were true, there might be some reason for preferring American freedoms. 

Can a deontic logic be 'complete'? The answer is no, because of impredicativity- i.e. it can't include its own 'meta-language'. One may be tempted to say, this means there will be unresolved 'conflicts' within a deontic logic (i.e. the representation of a system of morality). However, this is not actually the case. There are conflicts between 'interpretations' or 'models' of the deontic logic in question. It is likely that there is no non-arbitrary way to decide between them. But, the fact that a procedure is arbitrary in some respect does not alter its morality or utility. The buck has to stop somewhere.

Nobody wants or needs a 'complete ranking'. I don't care whether Mrs. Jones in Birmingham rather than Mr. Smith in Manchester gets the last jar of Marmite. Still, if get to hear that Mrs. Jones bashed Mr. Smith and grabbed the last jar of Marmite out of his hands, I would feel she acted wrongly. To have a moral position is not to claim omniscience. As for 'interpersonal comparisons', they happen all the time. The fact that you are happy to sell me your car at a given price means that I get more utility from it than you do. This may be because, as you mention, you are moving to a place where public transport is excellent but parking is very difficult. I, on the other hand, need a car for my new job as a travelling salesman. 

Sen seeks to distinguish between 'open incompleteness' where a partial order can be extended and 'closed incompleteness'. The problem here is that utility is epistemic. It is based on our knowledge base. Suppose I discover that my new job comes with a company car, then my utility from the car I bought from you greatly diminishes. Similarly, if you find that your new girl friend only wants sex when in a car, you may regret selling your car. More generally there can be no relational algebra or mathematical orderings- save of a rough and ready sort for a particular purpose- when it comes to anything 'epistemic' by reason of the 'intensional fallacy'. 

Nothing wrong in that. It would be foolish to rank asteroids on the basis of their moral character even if we believe Ceres has a rapey vibe and shouldn't be let anywhere near Uranus. 

We all evaluate actions of any importance to us. Such evaluations are not dependent on any theory we might have about what is good or bad. Once a person's 'evaluation' has been revealed- e.g. by offering to sell or buy a thing at a particular price- we can speculate on what motivated it. We see on Social Media that a man who is selling his car, has recently become an Environmentalist. We guess his motivation is ideological. Another man says 'I am selling my car because I'm moving close to work'. His motivation is utilitarian. However, no 'informational restriction' is placed on our interpretation of his action. After all, if new information becomes available that used cars are likely to rise in price because of Trump's tariffs, then both may chose not to sell at this time. 

Welfarism filters out non-Welfare considerations but nothing prevents a person being both a Welfarist and an Aesthete. He may say 'From the welfare point of view, Trump will be better than Kamala. But Kamala is prettier. As an Aesthete, I want Kamala because Welfare is less important for me.' What Sen has opened the door to is putting value on putting value on putting value- i.e. he has opened up an infinite regress. But, at the limit, the value of the thing remains the same because the value of putting value on putting value may be negative. Don't do it unless you have nothing better to do. 

This becomes apparent when we consider Brutus, a soldier patriot of Rome, who stabbed Caesar because he felt it was his duty to do so. As a soldier, in his view, that duty would remain the same for any soldier patriot. Judges or Accountants or other professionals have a non position-relative notion of duty. On the other hand, propriety may cause them to recuse themselves if failure to do so might appear to give rise to 'conflict of interest'. 

Appian lived long after Brutus and thrived under the successors of Augustus. Naturally he would take a dim view of Brutus. The question Brutus wanted the Roman people to ask themselves was whether Caesar was a threat to the Republic. Was killing him the patriotic thing to do? If so, any and every soldier patriot had a duty to do what Brutus and his colleagues did. Since the consequences of the killing of Caesar were bad- it led to Civil War- we think Brutus was wrong though perhaps his intentions were noble. 

This is not to say that Brutus denies 'uncorrelated asymmetries'. If a foreigner, or a disgruntled plebian had killed Caesar, he would have considered the action to be wrong and worthy of the most draconian punishment. He and his colleagues were patricians with a hereditary duty to protect the Republic. But, that Republic was doomed. 

Sen concludes his first Dewey lecture thus- 

Sen, like other Economists of his generation, had been involved in project appraisal. But it is he who pays the piper who calls the tune though, no doubt, he may pay a little money for a positive evaluation of the tune. However, if you want to get paid for appraising appraisal or appraising the appraising of appraisal, you are shit out of luck. Information is costly to acquire and the means to acquire it won't be made available for those who make distinctions without a difference regarding the evaluation of evaluation. There can be no substantive moral discussion with people obsessed with making a case for the making of a case for the making of cases.