Monday, April 15, 2013

Did Davidson slip? A quick one on reasons and causes

Finally tracked down my copy of _Essays on Actions & Events_.  There are places where Helen Steward describes a certain view about the relationship between singular and sentential causal claims as 'Davidsonian', but I wasn't entirely clear where Davidson defends the view.  Found it. In 'Causal Relations', he discusses the view that causes correspond to sentences rather than singular terms for events.  On such a view, the logical form of (1) is given by (2):

(1) The short circuit caused the fire.
(2) Because there was a short circuit, there was a fire.

He argues persuasively that these differ in logical form and it seems that one of the take away lessons of that we should think of causal relations as holding between events and causal explanatory relations as holding between something else entirely. (He says sentences, but I'd prefer propositions or facts. Let's just call everything in this lot 'dicta'.)  This leaves us with a question, which is how singular causal claims are related to sentential causal claims like (2).  Davidson suggests on pp. 155 that (1) entails (2), but (2) doesn't entail (1).

I'm sort of surprised to see Davidson say this.  If swallowing the Burgundy just is swallowing the poison, then wouldn't Davidson have to agree that both of these are true if one of them is?

(3) The swallowing of the Burgundy caused the death.
(4) The swallowing of the poison caused the death.

He wouldn't hold, however, that these are both true:

(5) Because there was a swallowing of Burgundy, there was a death.
(6) Because there was a swallowing of poison, there was a death.

I don't see how (5) and (6) can be entailed by (3) and (4) if both (3) and (4) is true but (5) is false.  The truth of the singular causal claims doesn't turn on how the event is picked out. Sentential claims like (5) and (6) seem to provide us with information about causally relevant features that isn't provided by the singular causal claims that Davidson suggests entails them.  So, he must be wrong, right?

I've been working through Davidson because I've been struggling to understand why he might have thought that reasons were causes.  He says that they are in 'Actions, Reasons, and Causes', but I don't see anywhere in there any reason to think that reasons are causes as opposed to dicta.  If he thought that sentences like (1) entailed (2) because (2) was really just some sort of generalization of (1) [a view that seems just completely unmotivated, so far as I can tell], then maybe he thought it didn't matter much whether we thought of reasons as causes or dicta. If there's no logical relationship between (1) and (2), however, maybe the question is a bit more pressing.

Here's a pitch for identifying reasons with dicta rather than causes.  First, let's assume that Davidson is right and nothing can be both a cause and something that corresponds with or is the explanans.  Second, let's note that we can identify the cause of an event and be utterly in the dark as to why something came to pass.  It seems that questions about relevance often arise after we've identified a cause.  It seems that these questions have all been put to rest, however, once an explanation is in place.  We should identify reasons with dicta rather than causes for just this reason.  When you have the reasons before you and they figure in a correct explanation, questions of relevance have all been settled. The singular causal claims don't settle these questions.  So, singular causal claims don't identify reasons.


praymont said...

Davidson's (2) reads "The fact that there was a short circuit caused it to be the case that there was a fire." (p. 151) He argues that, despite appearances, the relation indicated by '...caused it to be the case...' obtains between events, not dicta. As I read him, he says that (1) and (2) differ in form only because (1) is singular while (2) is general (existentially quantified). I think he'd say the same about your (2), taking it to be an extensional context (where the 'because' indicates a causal relation between events, not an explanatory one between dicta). So, I'd take your (3) to entail your (5), where the latter just means, "Some swallowing of Burgundy caused some death." I see dicta as entering the scene (for Davidson) only with his (8) (p. 153), which he doesn't take to be entailed by a statement of (1)'s form. (His (8) concerns a different example, but sticking with the short circuit example would make (8) read as follows: That there was a short circuit explains the fact that there was a fire.)

praymont said...

Oops, I should change "taking it to be an extensional context (where the 'because' indicates a causal relation between events, not an explanatory one between dicta)" to "taking it to be a case where the 'because' indicates a causal relation between events, not an explanatory one between dicta."

Dregs said...

You say that you've "been working through Davidson because [you've] been struggling to understand why he might have thought that reasons were causes."

I'm struggling to understand why someone might think that reasons were not causes.

If the reason for my paying the phone bill is that I desire to make phone calls, then my desire to make phone calls is the cause for my paying the phone bill.

The reason is the cause, moreover it's the explanans for my action in virtue of being its cause.

This last is evident from the consideration that no normal person would accept a reason as an explanation if that reason were not causally related to what it was meant to explain.

For example, if I said my reason for paying my phone bill was that I thought Jupiter was the loveliest planet in the solar system, most wouldn't accept that as an explanation. The problem with it is that there is no causal relation between my action and my stated reason for taking that action.

I take all of this to be in line with Davidson in 'Actions, Reasons, and Causes'.

Clayton Littlejohn said...

Hi Dregs,

You wrote, "I'm struggling to understand why someone might think that reasons were not causes."

It's a struggle because I think these claims are distinct:

*Causalism: Rationalizing explanations are causal explanations.
* R = C: Explanatory reasons are the cases of one's actions.

If, as Davidson argues, the explanantes of causal explanations aren't themselves ever causes, it seems that one might consistently maintain that causalism is true and reject R=C in favor of:
*R=E: Explanatory reasons are the explanantes of rationalizing explanations.

I don't see anything in Davidson's work that justifies rejecting R=E in favor of R=C, but both claims are consistent with causalism. So, the view I was considering wasn't one on which there's no causal story to be told about an action.

If I thought, as you seem to, that cases are themselves exlanans, than I guess I wouldn't think there was an issue here. I was working under the assumption that Davidson's conclusions in 'Causal Relations' was correct and that there's nothing that could be both a cause and an explanans. Do you think that his arguments in that paper don't succeed?

Dregs said...

Hmm, it's true that I don't find all of Davidson's points in 'Causal Relations' compelling.

That said, I don't see where he denies that a cause can be an explanans.

Are you looking at his distinction between causes and causal explanations near the bottom of the first paragraph, last page?

Perhaps my difficulty in following this is that I take Davidson to argue in 'Actions, Reasons, and Causes' that:

(1) Reasons for actions are causal explanations, and (2) the primary reason for an action is the cause of that action.

It seems to follow from this that some causes are also the explanantes of causal explanations.

Clayton Littlejohn said...

Hi Dregs,

My guess is that Davidson's views in ARC were refined over time and the more refined view is in CR. I guess I find his arguments in CR convincing. The relata of explanatory relations are fine-grained and the relata of causal relations are coarse-grained. (Or, that's how it strikes me.)

Dregs said...

Let's say that I put a kettle of water on a stove, turn on the burner, and the water subsequently boils.

Let's also say that I want to explain why the water in the kettle boiled.

I might say, "Heat from the burner was conducted to the water, causing it to boil."

This would be a causal explanation.

The cause of the boiling was the heat from the burner, which also features in the explanans of the above causal explanation.

This seems to be a clear-cut instance of the explanans of a causal explanation being a cause.

I still don't understand why you'd deny that (or what in the CR excludes it, but that's a different question).

Clayton Littlejohn said...

Hi Dregs,

I'd agree that the transfer of heat caused the water to boil (although you might have meant something a bit different in talking about 'the heat' as the cause). So far, so good. I'd also agree that in saying this, you've offered an explanation:

(i) Heat from the burner was conducted to the water, causing it to boil.

I suppose the question I'd want to ask is this. The line Davidson seemed to be taking in CR is that the explanatory relation relates things that are individuated in a way that's fine-grained but that's not the case with the causal relation. To test whether this is so, we'd want to test whether apparent cases of event-citing causal explanations allow for a kind of truth-preserving substitution in which the event designator 'e' is replaced by another event designator 'e2' which happens to designate the same event as 'e'. My guess is this. In the apparent cases of event-citing explanations (as opposed to event-citing causal claims) we find that they seem explanatory because we take it for granted that the explanatorily relevant features of the event are mentioned in the causal claim but the event-citing causal claim isn't _really_ an explanation because it's logically equivalent to claims that aren't themselves explanatory and would be false if translated into more explicitly explanatory talk. (e.g., we shift from 'cause' to 'because', 'for the reason that', etc.)

Take an example from Steward's discussion. There's a device that's supposed to light some matches remotely. While testing it, Adam was holding a match and tried to get Beth's attention to get her to activate the device. To do this, he made a motion with his arm that involved a striking of the match which alerted Beth so she pressed the button. She wants this to be a case in which the following are true:
(ii) The match was struck, causing it to light.
(iii) The match didn't light because it was struck but because Beth activated the device.

I take cases like this to illustrate that the truth of event-citing causal claims doesn't turn on whether the relevant events are picked out by means of terms that mention their causally relevant features. If, however, the truth of an explanatory claim requires that the causally relevant features figure in the claim, then there must be more information contained in an explanatory claim than the event-citing causal claims.