Monday, 10 April 2017

B.K Matilal & the impossibility of an 'action guide' dilemma

B.K Matilal wrote-

So, Matilal thinks an 'action guide' dilemma arises if there is what in Economics is called a 'choice situation', such that an action has an 'opportunity cost' in terms of the next best alternative that has to be foregone. Scarcity means that actions have an opportunity cost. Our time is scarce. If I devote this minute to typing this, I can't also use this minute to go and get myself a biscuit. Even thinking about something involves an opportunity cost. If I spend this minute thinking about that biscuit which is piteously calling to me from the cupboard, I can't spend this same minute thinking about how best to murli Manohar Joshi.

It is not moral or religious to waste time thinking obviously stupid thoughts. To imagine anyone, strictly motivated by Religion or Morality, can have an 'action guide' dilemma is to have wasted time thinking an obviously stupid thought. That's why non-stupid Religious and Philosophical authorities say 'action guide' dilemmas don't exist within their systems of ethics. What may exist is mental derangement of some sort which causes a person to feel that they are damned if they do and damned if they don't. The Sanskrit word 'Vishada' denotes this type of derangement. Arjuna suffers from Vishada which is why he says things which indicate he has an 'action guide' dilemma. Krishna cures him of this Vishada. That's what happens in the Gita. A sick man is cured. It turns out there was no 'action guide' dilemma at all.

Why is it obviously stupid to think a Religious or Moral 'action guide' dilemma can exist? The answer is that, if such a dilemma arises, then either
1) the agent has already done something which puts him beyond the pale of that Religion or Morality- viz. whatever actions led to him being put in the dilemma. Take Hamlet's quandary. It only arose because he had commerce with an unclean spirit. Neither Religion nor Morality says you have to listen to ghosts- more especially when they tell you to kill your Uncle. No one thinks of Hamlet as featuring an 'action guide' dilemma of a particularly Christian or Moral sort. It is a tragedy. A mind keen enough to grasp Agrippa's trilemma has been caught in a self spun snare. This is hamartia- a tragic flaw in a Prince of great intellect and noble character.
2) that Religion or Morality is not truly action guiding at all and hence of no use to its votary for whom some such crutch seems necessary. Drop the crutch. Economics is the discipline whose subject is opportunity cost. Any given Religion or Morality, if it satisfies a plausible condition I will describe in my next paragraph, can be shown to have a 'complete deontic code' such that there is always a prescribed action under every state of the world and thus no choice has to be made and hence no opportunity cost is incurred. It is a different matter that a hermeneutic or epistemological or 'signal extraction' problem might arise in decoding what that prescribed action is. But, qua that action, Consequences are irrelevant. A mass murderer may be let free on a point of law even if the consequences will be very bad. That doesn't matter. What matters is that the law is correctly interpreted according to its own system of 'artificial reason'.

Let us call a Religion or Moral Scheme's deontic code complete if it prescribes an action at every moment in time for its votary.  This means that the votary never incurs an opportunity cost- there was literally nothing else he could conscionably do- and hence experiences 'zero regret' with respect to his actions if he is a true believer.

If the moral code is not known to be complete but a 'zero-regret' trajectory is feasible for a particular votary, then, by the Szpilrayn extension theorem, we know the moral code is complete for that agent.

Can this moral code be complete for any arbitrary agent? Yes, if there is a choice sequence through Stalnaker-Lewis closest possible worlds such that the ideal agent's world deforms continuously into that of the arbitrary agent so that the feasible zero regret trajectory of the former maps onto the decision space of the latter

Alternatively, we could define a coordination game whose correlated equilibrium expresses the same thing. This enables us to use information contained in the 'zero regret' stipulation regarding the ideal agent's trajectory.
Imagine two players in a coordination game who are both privately informed by an omniscient being about how to act. One is told to behave like the ideal agent and given signals so as to make decisions she won't regret. The other is told to behave like the arbitrary agent and given signals to make decisions which the other player would not regret were she in the shoes of the arbitrary agent.
The omniscient being can now construct a pay-off mechanism such that no signals need be given. Hart & Mas-Collel have shown that knowledge of this pay-off matrix is sufficient for the correlated equilibrium to be achieved. Thus, we don't have to have an omniscient being at all. We have shown there must be some game where the Hannan consistent strategy gets us to the result we need.

Thus we know a complete deontic code for everybody must exist if there is a feasible 'zero-regret' trajectory for even just one votary of it.

Let us look at this another way. Suppose I believe Mother Theresa could have lived a life in obedience to a deontic code which I myself believe in and that if she had done so she would have had no regrets nor faced any dilemmas. Is this sufficient information for everybody to, in good-faith, agree that there must be a complete moral code for me as well- even though I am a man, not a woman, and live in a different world with different opportunities and different threats?
The answer is yes.
Suppose every rational, non antagonomic, person who ever had or will have a view on this topic were endowed with infinitely long lives and infinite information and computing power and that they were all put in a room together and asked to decide on this question.  Then, sooner or later there would be something like 'Aumann agreement' between them as to what Mother Theresa would do in my shoes. Of course, this consensus might only be achieved very very long after the Universe has been destroyed. Still, this suffices as an 'existence proof' that so long as Mother Theresa had a feasible zero regret trajectory and I believe I should do what she would do in my shoes, then there is a complete moral code for me which is 'objective' but which might be beyond my ken.

What if we, quite sensibly, don't want to have anything to do with mathematical theorems or thought experiments involving infinite computing power?

Then, it is still the case that if we think there is at least one person who, on his deathbed, can truthfully say 'if I'd followed such and such deontic code, I'd never have had anything to regret' then we know of a moral code that is complete. Either there is some 'work on oneself' we need to do so as to be fit to implement that moral code, or else we're just shit out of 'moral luck' and damned by that deontic code for a reason which Religion or Morality can treat as a mystery it would be impious to inquire further into.

One way out of this gloomy conclusion is to reflect that any regret minimising strategy could represent a moral code over its duration.  By backward induction, choosing that moral code renders it zero-regret. Stitching together durations to make a life history, composed of 'stations of life' regulated by different deontic codes, is then sufficient to redeem our lives. The Hindu 'Varnashrama dharma' is sometimes viewed in this light. This is fine if the lens we are looking through is something like  'Hanann consistent adaptive learning'. However, as a justification for a 'casteist' status quo it is worthless shite because it cashes out as nothing but moral dilemma piled on dilemma without any action guiding actually occurring.

To conclude, we know there is a complete moral code, such that 'normative reasons' exist at every moment of our lives, and, what's more, it's what we'd have chosen anyway given our preferences, endowments and Bayesian priors. However, the code may not be effectively computable or else may have exponential Kolgorov complexity. That's a good thing because Newcombe problems exist and also because an easily hackable code renders us vulnerable to parasites and predators.

A moral code that insists that it be effable at every point- i.e. justify its actions in rational terms- must be different from the globally regret minimising moral code if computation, that is cognition, uses up scarce resources or else if strategic problems arise. The effable code may be superior for an agent who devotes himself to a 'second order' good- e.g. proselytising for deontics. However, in that case the associated complete moral code considers those who do first order good to be 'damned'.

Does this give rise to a dilemma? Nope. People who think they are doing 'second order good' are too deluded to understand that they need to stop talking shite and do some 'first order good'. Suppose you get paid more, or gain more prestige by doing 'second order good'- it remains the case that you are crowding out the first order good you can do. It may appear that the second order good creates much more of the first order good than you could possibly produce yourself. This is a delusive appearance. It is just bad Economics not some fancy-shmancy moral dilemma you can vapour over. Get over yourself. Admit the truth. Moral dilemmas are the chrematistics of a worthless class.

No comments: