Thursday, March 01, 2007

Logical Distinctness. An interpretaion of Templing

Robert Templing gave a great talk on causation last night best summed up in his own words: “What the heck is logical distinctness anyway?”. Rob was looking into what seems to be a fairly standard assumption in the philosophy of causation which is that the cause must be logically distinct from the effect. Two questions are: 1. Does the cause have to be logically distinct from the effect? What motivates this rule? 2. What is it for two facts, events, objects (whatever causes and effects are) to be logically distinct?
I will give my answer to the first question without further ado. Cause is a linguistic concept and language is governed at least partially by pragmatics. If one can deduce B from A anyway, one doesn’t need the concept of cause to explain the fact that B followed A. Where there is the relation of logical entailment between 2 events a causal explanation of the relation is redundant. So intuitively speaking “Frodo puts on the ring” is not the cause of “Frodo puts on the ring” because the counterfactual dependence is already explained logically. This explains the intuition as far as I am concerned. By way of argument, all I offer is that often people do make these kinds of causal claims, and their uselessness is the best argument against them. For example: “Eating something poisonous causes one to be poisoned.” The intuition I have is not “False!” but “Well duh! No shit Sherlock.”
The second question is the more interesting and wide ranging., especially now when the new logical inquisitors are asserting as fact that “necessary” means “logically necessary” and “possible” means “logically possible” and “identity” means “logical identity”. Logic is to do with true and false, and true and false only attach to representations, maybe even only linguistic representations. The universe as a whole is not true. It is not a conjunction of true propositions. The universe just is. The flash point in Rob’s talk as far as I was concerned was a challenge to logical distinctness from Davidson. Davidson views cause to be a relation between events. Logical distinctness seems to be a relation between descriptions. So the logical distinctness condition is false since you can always describe a cause and an effect in such a way that the cause and the effect are not logically distinct. Moreover, it doesn’t really mean anything to say that two events are logically distinct. Davidson’s argument is that
Suppose A causes B. This means the description “the cause of B” truly describes A. So we have this true causal statement “The cause of B causes B.” This is a tautology. Therefore under at least one description every cause is not logically distinct from its effect.
But this is rubbish. If “the cause of B” rigidly designates A, then, given that some causal relations are logically contingent, it is true that it is logically possible that the cause of B does not cause B. If on the other hand “the cause of B” only refers to whatever is in fact the cause of B, then “the cause of B” no longer refers to A necessarily, but merely contingently. Either way the cause of B and B are logically distinct.

To try and make this a little less abstract. Chapman by the action of shooting murdered John Lennon. Therefore Chapman’s action was the cause of John Lennon’s death. Now we have a description of an action “the cause of John Lennon’s death”. This refers to Chapman’s action. But suppose Chapman’s action was not the cause of John Lennon’s death. Now we have: “Suppose the cause of John Lennon’s death was not the cause of John Lennon’s death.” Are our high Priests of Modal logic going to deny us this hypothesis on the grounds that it is impossible? Of course they aren’t, they are reasonable people. So it is Davidson who has made the mistake. The cause of B is logically distinct from B, even so described.
Take a non causal example. Jack is married to his wife. Is this a logical tautology? Not really. It makes perfect sense to me that Jack actually never got married to his wife. I mean in this world, indicative possibility, epistemic possibility. They might never have had a ceremony.

Another example of this pervasive error is the fact that everything I know is true. Williamson and others take the appearance of tautology in this statement to ridiculous extremes. The fact that I know that p does not entail that p in any interesting way. I now hereby claim to know that Robert Templing gave a talk last night. These two events are distinct. Robert could have given his talk without me knowing about it. I could have claimed to know it without him having given it. It is true in a sense that if he hadn’t given his talk, my knowledge wouldn’t be knowledge. It is also true that if I hadn’t known about it his talk may have been a slightly different event. But this is not a logical relation. In my private world I refer to this representation of Roberts talk as “knowledge”. This act of describing does not warrant the inference that it is logically impossible that Robert failed to give a talk. It just entitles me to assume with certainty that he gave a talk. It is logically possible that I am a brain in a vat. Or that the world consists of two identical iron balls. The only logically impossibility is that logicians have got it wrong. This puts logicians up with Freud and the Pope.

Labels:

4 Comments:

Anonymous Lee Walters said...

Why do you think cause is a linguistic concept? Laws are linguistic but is not cause a metaphysical notion?

One motivation for some sort of distinctness requirement is, as Kim notes, that counterfactual dependence seems to be a wider phenomena than causal dependence. The instances of counterfactual dependence that appear not to be instances of causation seem to involve analytic/logic dependencies or part/whole dependencies.

Two examples from Kim: that I write "Larry" is counterfactually dependent on me writing "rr" but we may be reluctant to say the latter caused the former.

Secondly, my sister's giving birth did not cause me to be an uncle although the latter counterfactually depends on the former.

I'm inclined to think that it is part-whole distinctness that is required - this is what explains your Frodo example. If we don't rely on analytic/logical distinctness then we don't fall ionto the problems you mention.

1:43 PM  
Anonymous Jonny said...

Yes, Robert talked about the same examples. When I say the causation is a linguistic concept this was loose talk and wrong probably. I just mean that if we test linguist intuitions to try and find problems with counterfactual analysis we may find ourselves thinking "that ain't causation", but the intuitions may stem from pragmatic considerations. With the examples, it is not so clear that you can't think of other examples where the intuitions go the other way. Mike Gabbay came up with "Mike's winning the race caused it to be the case that John didn't win." Even the birth ofthe neice case is not so clearly not a case of cause. As for the LARRY case: If the question was "what caused you to write double r on that type writer when you know it makes the keys jam?" then my writing Larry does seem a good cause of my writing double r. Intuitions go wither way, and it depends what use you want to put the concept of causation. Different things can be called the cause of the same event according to differently driven enquiry. "Poison caused him to die" "The hatred he attracted to himself caused him to die" This statements could both be true, or at least communicate something useful, of the same event. Yet neither is it a case of over causation or a contradiction.
I think the intractability of the analysis of causation is because there really are differing concepts of causation. We could call them "Aristotelian" Which includes teleological, constitutive as well, And "Humean" Where it is a constant conjunction of contiguous yet logically independent facts. In language we use both, and the counterfactual analysis is wider than Humes, but narrower than Aristotles.
Just to try yourself out.
1 Does the fact that a liquid is H2O cause it to freeze at 0 degrees celcius? (yes Aristotle, no Hume)
2. Does the need to reach tall leaves cause the Giraffe's long neck. (yes Aristotle, no Hume)

10:07 PM  
Anonymous Lee Walters said...

OK, I agree with most of what you say - we often stress one thing rather than another as *the* cause. That is something I am not interested in.

I also agree that it is not clear that some of the analytic-ish counterexamples are counterexamples. that's why I emphasise part-whole distinnctness. (I avoided the anlytic-ish type cases as didn't want to get into event-individuation).

As for the Larry case: I take it you agree writing "rr" is not a cause of writing "larry". The question you raise is whether writing "larry" is a cause or writing "rr". If so this would threaten my thesis of part-whole independence although I could easily revise it to make it one way - parts don't caus their wholes.

However, I'm inclined to say writing "larry" is not a cause of writing "rr". If it were we would have backwards causation without counterfactual dependence. That in itself is not an argument against it, although such instances of backward causation would be ubiquitous - eating my breakfast caused me to eat that slice of toast, having a shower caused me to wash my balding pate etc. Rather I think we should say that whatever caused us to write "larry" caused us to write "rr". So my intention to write the former was a cause of writing the latter.

In your case of multiple causes without overdetermination - the poison and self-loathing - presumably the response is that these are both part of the same causal chain - the self loathing caused him to buy and ingest the poison which caused him to die.

If you change the example to say the lit match and the presence of flammable material both caused the fire, we don't get overdetermination either, since neither sufficient alone, but both sufficient in the circumstances, which include the other. The fact that we fix on the match is I think psychological / reflects the typicality of the causes.

8:54 AM  
Anonymous Jonny said...

The one way part whole thing is something i brought up in the talk. It seemed to me at the time that a part can be the cause of a whole whereas the whole can't be a cause of the part. It is the arsenic in apple pips that cause them to be poisonous. The fire in pudding lane caused the great fire of London. But reading your comment makes me realize that with the "Larry" case my intuitions go the other way. Writing "rr" can only be at most a necessary condition of writing "Larry", whereas writing Larry seems to be a sufficient cause of writing rr.
Ruling out backward causation is a conceptual decision. If you do rule it out it becomes hard to talk of function and goal oriented behaviour. I see no a priori reason to rule it out. There does seem to be something about ordering that is essential to causation though. This is why I am not so dismissive of psychological considerations as you seem to be. It is not the direction of time that is important, but the direction of explanation. In essence I think that the cause of an event is the feature of the world the preventing of which would prevent the effect. In the match versus oxygen in the air case, it is clearly not an option to prevent the fire by removing all the oxygen. In the writing "Larry" case, you could prevent someone writing Larry by preventing them from writing rr, but not the other way round.
Here it is. When asking what caused B, we are asking either "how could we have prevented B?" or "If we want things like B to happen, how can we go about it" or, "under what conditions should we expect A".
My metaphysical view is that the universe is a causal interdependent whole and not one atom could be different. In this sense everything is causally dependent on everything else in a metaphysical way, so there is not much use in a metaphysical concept of causation other than for purposes of wonder and gratitude of a Panglossian type. (nothing wrong with that of course)

9:33 AM  

Post a Comment

<< Home