Wednesday, June 15, 2011

Dumb luck (pt. 2)

In the previous post, I posted arguments from Jennifer Lackey critical of the current views on luck. Now I will offer a positive sketch of my own view which is inspired by a contrastive account of causation by Jonathan Schaffer.

In Jonathan Schaffer's work on causation, Schaffer uses a contrastive approach which he claims to solve many of the paradoxes involved with causation. Causal relations are fundamentally quartenary composed of a quadrouple where c is a cause and C* is a non empty set of possible alternative causes, e is an effect and E* is a non empty set of alternative possible effects. Both causes and effects are "within a field" the contents of that field is determined by the causal inquiry.

There's many causes to any effect. Likewise, for any cause, there will be many effects "down the line." It's not informative to point out some of them. Saying that the plane crashed because of gravity is not normally informative though surely gravity played a crucial causal role. Context, (whether as a function of the semantics or pragmatics Schaffer is agnostic) determines what the relata are in any causal scheme. Thus, when we say that the short-circuit caused the house to burn down, we actually say something like: The short circuit (as opposed to the match, or the lightening, or the flamethrower, etc) caused the house to burn down (as opposed to remain intact). As you can see, this has the quartenary causal relation structure.

Many problematic issues common to the work on causation are solved or dissolved by this contrastive causal framework. Schaffer's solution seem convincing to me. Consider the problem of absences:

Are absences causal? To ascend semantically, can negative nominals such as ‘the gardener’s not watering my flowers’ denote causes and/or effects?

(1) (pro) Absence causation is intuitive: intuition accepts some

absences as causal.

(2) (pro) Absences play the predictive and explanatory roles of

causes and effects.CONTRASTIVE CAUSATION


(3) (pro) Absences play the moral and legal roles of causes and


(4) (con) Absences mediate causation by disconnection.

(5) (con) Absence causation is counterintuitive: intuition rejects

some absences as causal.

(6) (con) Absence causation is theoretically problematic.

(7) (con) Absence causation is metaphysically abhorrent.


Contrastivity resolves the paradox of absences, by reconciling (1)–(7). The reconciliation strategy is as follows: (i) treat negative nominals as denoting actual events, and (ii) treat absence-talk as tending to setthe associated contrast to the possible event said to be absent. Forinstance, given that the gardener napped and my flowers wilted, ‘The gardener’s not watering my flowers caused my flowers not to blossom’, is to be interpreted as: the gardener’s napping rather than watering my flowers caused my flowers to wilt rather than blossom.

Many other causal paradoxes once plagued the literature are shown resolved or solved.

Now consider luck. I claim that there is something similar going on and that luck is actually a contrastive ordered pair where s is either a fact about someone or an event that happens to her and S* is a non empty set of non actual possible alternative facts or events (what I'll call "scenarios") that may have happened to her or S* may be actual scenarios occurring to other people. s is more beneficial in some way than the scenarios in S*.

What determines the kinds of scenarios in s and S* is also a function of context and inquiry. It's a context relative matter what that criteria will be; she might be richer in s than all the scenarios in S*, she may have found a treasure in s and not have done so in S*, she may have been better looking than the possible scenarios she is in S*, etc. In the first case, the criteria of evaluation is wealth (more versus less), in the second, it is some fortunate event happening to her (as contrastied with it not happening), and the last case it is her appearance (more attractive versus less so). She might be happier, have more stuff, friends, a better spouse, etc etc, it doesn't matter so long as s is evaluated at some criteria that is the same as those in S*. What we choose to be evaluated is dependent on what the context suggest is appropriate for an evaluation of luckiness (or unluckiness as the case may be).

However, the choices for the different alternative scenarios in S* are not completely arbitrary. They should be "live options," likely or contingent alternative scenarios based on the kind of criteria we are evaluating her luck on; they are possibilities that may have occurred to her or she may be reasonably compared based on some relevant reference class in which she belongs. It should be a live option that some person might have been happy or sad within a range or rich or poor to a various range or have so many number of good friends, etc. Her relevant reference class may be her set of friends, some possible scenario other than the one she finds herself, all actual people, all possible people, etc.

But unlike what is determined to be evaluated in s and S*, it's an objective matter how she actually fares relative to the other scenarios in S*. For example, say she has more of x (money, good friends, happiness, etc) than she does in S*, she is correctly said to be lucky. I think this will solve all the problem limned in Lackey's paper.

Take a look at her example of the DEMOLITION WORKER. The context of evaluation is types of events. In the actual event, the explosion occurred because her coworker hanged his coat in the right place, at the right time, etc. Now S* may be some possible event nominal such as his hanging his coat in a slightly different place not resulting in the causal chain culminating in the explosion, or at some different time also not resulting in the causal chain culminating in the explosion, etc. All the events in S* are live options; they may easily have occurred were it not for chance. We can even specify that the scenarios in S* be more counterfactually robust than s, that is, that s is less likely than those scenarios in S*.

In the case of BURIED TREASURE, there is likewise, several ways to evaluate Vincent's luck. We may construe s as Sophie burying her treasure in the spot she did with some alternative scenario in S* in which she chose some other area of the island. We may also construe s as Sophie's choice of flowers. She in fact chose roses which resulted in her choice of area to bury her treasure. But if she had wanted to plant some other plant which resulted in another area more suitable for that plant, another area would have been chosen. In that case, we would have her choice of roses be an event s and S* maybe alternative possible choices such as s_1=her choice to plant tulips, s_2=her choice to plant daisies, s_3=her choice to plant orchids, s_4=her choice not to plant anything at all, and so on. Had she chosen to act on any of those alternative choices, Vincent would not have found her buried treasure and thus not have been so lucky. Alternatively, we may evaluate Vincent's luck by construing s as his mother favoring roses. If she had favored some other plant he would not have gone to the spot and found the treasure. Thus S* would be his mother favoring, say, daisies. The point in this schema is we hold all the other facts fixed in s while choosing alternative scenarios in S* (the field of possible alternative outcomes not as lucky) that would not have been as favorable on some chosen criteria as in s.