Here's a point that came up recently following a talk by Arché visitor Dean Zimmerman on Molinism.
It can't be sufficient, to count as having control over someone else's actions, that their actions vary counterfactually with yours - that were you to F they would F, and were you to G they would G, etc.. For if this were the case we could have symmetric control. It might also be that were they to F you would F too, and so on. (Suppose that, necessarily, you F iff they do.) But you can't each be controlling the actions of the other - control is an asymmetric relation.
So what do we need in order to explicate the notion of control? It doesn't help to specify that exactly one of them is aware of the counterfactual relationship between his actions and the other guy's, and say the controller is the one who's aware. For one can perfectly well be aware that one's actions are being controlled by someone else (who is not aware of the fact).
Unsurprisingly, I think what we need to appeal to here is probably explanation. What matters is whose actions explain those of the other. You are in control just in case your actions explain the other guy's. Explanation brings with it the right kind of asymmetry, as well as making sense of the feeling that the controllee's actions depend upon the controller's.
Subscribe to:
Post Comments (Atom)
8 comments:
Surely there has to be a deeper account. It would be baffling if explanation was conceptually and metaphysically free-floating, in the sense that X's actions could explain Y's, but Y's not explain X's, without there being any other relevant conceptual and metaphysical differences between X's actions and Y's. If there are such differences, why not cite them in the account of control? If we are confident that there are such differences, but are at a loss to say what they are, why not just accept that we can't yet explicate the concept of control over action? It seems to me that the appeal to explanation can only be a step towards an explication, not the missing piece of the puzzle.
Is the idea that
(1) X controls Y's actions
entails
(1*) Y acts in virtue of X (or X's will, etc)
where 'in virtue of' is more exclusive than 'causally in virtue of'?
If so, why does this relation hold between some objects/events/facts but not others? What's the epistemic route that allows us to know when it holds? (In the symmetric action case, is it unknowable? Or is the idea that there are other facts about the situation that incline us one way rather than another?)
Maybe you're going to say that the holding of the relation is response-dependent. But in e.g. the moral case, we demand consistency of response (same subvening (natural) facts, same response) on pain of conceptual incompetence. Looks like explanation will share this kind of feature. And just as in the moral case, won't the demand be to say why this should be so?
Hi Andrew,
"If there are such differences, why not cite them in the account of control?"
Because I suspect they'll be different in different cases. That is, I expect that all cases of control have the explanatory assymmetry in common, but doubt whether there's anything 'deeper' that they'll all have in common.
The 'in virtue of' locution feels right, but I think that might be just because we can use 'in virtue of' to indicate explanatory dependence.
Your other questions (why does it hold between some things and not others, what is the epistemic route etc.) aren't ones I know the answers to at the moment. If I was going to write a paper on this topic I would probably try to work out what to think about them, but unfortunately I far too much other stuff to do!
(I do know, however, that I'd want to steer well clear of response-dependence!)
Say we come upon a new case. In pretty average circumstances, we unhesitatingly and independently converge on taking it to be a case of asymmetric explanation, even though (i) it doesn't have anything significant that links it with the prior cases of such explanation that we have considered and (ii) the mere fact that we respond to it in that way in certain circumstances is not sufficient for it to merit being counted as a case of such explanation.
wouldn't that strike you as a bit remarkable?
Yes. But I'm not sure I see why that matters. If you like, think of control as a role property, best explicated in terms of explanation, with an associated set of (familiar) realizer properties. New cases that we recognize as possessing the role property will have one of the familiar realizer properties too. Still, control is the role property, not any (or all) of the reazlier properties.
Hi Carrie,
Maybe this is obviously problematic, but why isn't causality what you're looking for? What matters is whose actions cause the others.
That would avoid one apparent difficulty for the explanatory story. Suppose a Jackson-style epiphenomenalism about mental states. I take it we don't want to say that mental states *control* bodily behaviour (do we?). On the Jackson picture, mental states explain but don't cause bodily behaviour.
Here's one difficulty, though, for both causal and explanatory stories. One sort of case that raises difficulties for the counterfactual proposal is one where y's actions counterfactually depend on x's, because y is voluntarily imitating x. But it also seems in the imitation case, that x's actions cause and explain y's (of course, there's another causal and explanatory factor - y's decision to imitate x - but that doesn't obviously undermine the case).
The point seems to be (to put in generally) that there are lots of ways of y's actions depending on x's that don't amount to x's controlling y. Something like "intentional dependence" seems needed: y's actions depend on x's, and x intends to bring about y's F-ing by his F-ing.
My explanation of the Jackson thing is a little cryptic. What I meant was that *under the hypothesis of epiphenomenalism*, we wouldn't want to say that mental states control bodily behaviour. Yet they can still explain behaviour.
Hi Robbie,
I'm tempted to think that when you decide to imitate someone you're letting their actions control yours.
But suppose the general point is right that there are ways for y's actions to depend on x's without x controlling y (can you give another example?). I don't think your fix would be quite what we'd want; it might provide a sufficient condition, but I don't think it would be necessary, because I think there can be unintentional control.
I just don't see how this kind of thing helps. Control, on the suggested account, is a matter of there being a X-based realizer that explains why Y acted as she did. Now, is the fact that an all-knowing God deliberately placed Y in a situation where she would freely choose to act as she did a case where God's act counts as a realizer? How does it help just to bring in the notion of explanation?
I'm also puzzled about (i) what the epistemology is which originally classifies the 'familiar' realizer properties and (ii) what the argument is for identifying control with the role property, rather than the moderately disjunctive realizer property?
Post a Comment