Sunday 6 December 2015

Alan Gibbard, Semantic Normativity & Ontological dysphoria

Economics was once known as the 'dismal science'- its study appeared to hold little utility other than in salutarily lowering expectations and reconciling the productive classes to a miserabilist horizon.

Then, during the Second World War, some young Americans, not from established elites, got a chance to use their mathematical or analytical talents to redirect or even boost domestic production as well as to improve military tactics and strategy. What they were doing wasn't something wholly new- in the past, highly experienced men with 'expert cognition' had developed heuristics, some of which had been given a mathematical treatment. This time, however, Pure Mathematics had reached a critical mass such that exponential development in some fields- e.g. Turing's Enigma or Von Neumann's Los Alamos- completely altered the horizon in every other in a mutually intelligible manner.

Thus, after the war, though the contribution of women was still not acknowledged- indeed, historically, Economics has been the most misogynistic of the Social Sciences- the WASP establishment had to retreat in the face of some extraordinarily brilliant and prolific, often working class or refugee, Jewish scholars whose social optimism went hand in hand with a willingness to recast the foundations of their discipline on terms as rigorous as Mathematical Physics in the assurance that this would unleash human productivity in a manner more unequivocally positive than the Manhattan Project's splitting of the atom.

Rising prosperity during the Fifties and even into the Sixties, led to a completely altered public perception of the Scientific status of Economics. There followed a golden age, culminating in a Nobel prize being created for what had become, by default, the Queen of the Social Sciences, if not Philosophy's young and as yet promising Regent.

But this dizzying apotheosis was swiftly followed by a dramatic reversal of fortune- the 'Stagflation' of the Seventies- featuring a dialogue of the deaf between Keynesians and Monetarists, while Hayekians and Marxists prowled around like hyenas picking off the wounded- which brought Economics into secular disrepute but did so at precisely the time when advances in Information Technology and Dynamic Programming- as well as the increasing availability of 'big data'- had the potential to put flesh on an empirical, evolved under uncertainty, regret-minimizing, Mathematical Skeleton quite different from the Utility Maximizing Adam Kadmon of Arrow and Samuelson and Hurwicz and Baumol.

However, the ghastly interregnum between Economists pretending they were Physicists and Physicists steamrolling over Economics was not without its own landmarks such as the Gibbard-Satterthwaite and Myerson-Satterthwaite theorems which link mechanism design to evolutionary biology in an illuminating manner such that we realize Preference Revelation is not a hurdle to be surpassed but something Evolution has baked in to ensure enough hedging against uncertainty obtains. This is not to say that faith in a 'Revelation Principle'- i.e. the belief that if an outcome can be implemented by an arbitrary 'black box' mechanism, then it can also be implemented by a white box mechanism based on true preferences- might not be Muth Rational- i.e. something all agents are better off affirming.

Alan Gibbard- who first articulated this principle- is not however now an Economist but rather a glittering gem in the diadem of Academic Philosophy- a discipline which has suffered an even more radical depreciation in general esteem. He argues- following Kripke- that 'claims as to what something means are normative claims.' In other words, he refutes the notion that meaning is something objective- the solution to a coordination problem- which can be discovered by a purely alethic process- e.g. looking up words in a dictionary or working out the Schelling focal point or solving for the constrained optimum on a Social Welfare function. Rather, meaning is the instantaneous change in the action set brought about by its own acceptance. If this 'instantaneous change' accords with Gibbard's 'Revelation Principle'- so useful in incentive compatibility and mechanism design- then there is some point of view from which Gibbard's 'meaning of meaning' is also the solution to a coordination problem. If, however, people have an incentive to lie to themselves- perhaps to escape a computational cost or to baffle a parasite or predator which, otherwise, would be able to predict their behavior- then we have some Newcomb's problem or Kavka toxin type reason for believing this might not be the case.

Philosophically, one may still doubt whether 'lying to oneself' is meaningful. One method of circumventing this problem is to say that people may have a preferred ontology which is not at home in this world and so their truth-makers express ontological dysphoria, which- in Gibbard's system- could be called action states.

Consider what happens when two people meet up and first of all both agree that 'mutually beneficial trade is a good thing', and then one of them goes on to say 'the previous statement has a practical meaning- viz. you ought to trade me that pen you are not using in return for this calculator that I've no need for'.
I suppose the other could retort- 'The Myerson-Satterthwaite theorem suggests otherwise!' and that would be a 'knock down' argument.
In practice, the other party may refuse a transaction of the sort outlined above out of an obscure sense that in an uncertain future he may regret the transaction. The problem is, his interlocutor may say, or feel, that this agent is being a 'meanie'. Moreover, this interlocutor may refuse, at a later point, a univocally beneficial trade simply to 'punish' the meanie. In other words, there are hysteresis effects and reputational problems here which complicate things. 'Cheap talk' is being turned into a 'Costly signal' by an illicit mechanism.

Gibbard gives a philosophical defense against this sort of illicit mechanism but, unlike the Myserson-Satterthwaite proof (which links up with real world behavior under uncertainty in a fruitful manner such that new types of assets or exchanges can be mutually agreed) it is not intended to be categorical.
As he tells 3AM- 'The obvious approach to the meaning of meaning claims, though, is to try to define the concept of meaning in naturalistic terms, in terms that can fit into a purely empirical science. A central question for me, then, is why I reject treating meaning as a concept within a purely empirical science.
'As early reviewers of the book point out, I’m rather perfunctory on this question. I don’t come up with a knock-down argument that analyses in scientific terms won’t capture the meaning of meaning claims. In a way I don’t want to: as I say at one point in the book, I’m not convinced that treating the concept of meaning as a normative concept will be the most illuminating way to approach the theory of meaning'

Thus Gibbard on Normativity isn't like Myerson-Satterthwaite in settling an argument once and for all. Rather any hoped for similarity would lie in the fruitful avenues of further exploration which are opened up. At Gibbard says- 'A chief aim of the book is to try to show how fruitful a normative approach can be in identifying what might be at issue in questions of meaning.' 
Since most people find the math used in the Myserson-Satterthwaite proof a bit daunting- though, once grasped, it immediately suggests new and useful approaches- it would be great if there were a simple meaning to Gibbard's thesis, more especially as he tells us it would be fruitful.

Take the following dicta, as quoted by Tim Williamson-
1) 'The meanings of the words in a sentence combine to explain which inferential oughts apply to the sentence and the evidential conditions under which one ought to accept or reject the sentence. A word’s meaning what it does consists in the pattern of oughts that enters into such explanations. (p. 114)
The emphasis on explanation makes this seem a functionalist view of Semantics from a Normative perspective- not a Semantic Normativity. As such, it might encourage us to equate Game Theory with meta-ethics. However, this is not at all Gibbard's view.

2) ‘The point of normative claims is to tie in conceptually with action’ (p. 227).
 For example, ‘I can’t consistently believe I ought right now to leave my burning building and decide to stay. Naturalistic thoughts, in contrast, lack this conceptual tie to action’ (p. 224). 
As Tim Williamson points out, Gibbard's agent who thinks 'I ought to leave my burning building' has to distinguish whether this is a naturalistic thought or a normative claim. Only in the latter case does he actually have to leave the building. One way of making sense of this is to say that there is a Bayesian process going on. If you often have the thought 'gotta get out of this burning building' and you never do and nothing bad happens then, probably, there was no Gibbardian normative claim there at all. You just take too many drugs or else the fire alarm keeps going off and you have a vivid imagination or something of that sort.

3) This is Gibbard's reflection on Kripe's classic passage re. Meaning as Normativity-
Kripke  actually said 'what I intend to mean'- i.e. this was a hypothetical not a categorical imperative- but this does not alter Gibbard's point which is that normativity always has a conceptual tie to action.

 As readers of this blog may remember, I have previously argued that 'the other side of Hume's Guillotine is that belief in an 'ought' causes us to arbitrarily restrict the domain of what 'is'.
Gibbard, links the action space of an agent with Normativity. If ontological dysphoria is an action then Normativity is its high road because of the conceptual tie between thought and action Gibbard posits. Thus every admitted 'ought' leaves us less at home in the world.

Could Gibbard's first sentence, quoted above, bear the interpretation I have just put on it? Perhaps Meaning is Normative in the sense of mapping to changes in the agent's degree and type of ontological dysphoria- i.e. being a function of the mismatch between her ex post conceptual action space and that associated with the ex ante Bayesian-Nash semantic equilibrium- and this is fruitful in that we can now understand semantic processes as being like Newcombe problems with Kavka toxins releasing or damning up capacitance diversity on a high uncertainty landscape.

Alternatively, we can think of the conceptual thought-action tie peculiar to  Gibbard's system as analogous to his Revelation Principle because consistent discrimination of naturalistic from normative statements is a sequential equilibrium. However, in at least one case- viz. where regret minimization is incentive incompatible- his meaning of meaning may be gamed in an ontologically dysphoric manner such that mere Meta-Metaphoricity is its own bilateral context. Thus meta-ethics has no privileged diegesis nor intensional langue nor ideal type parole.

No comments: