A sketch of an argument to the conclusion that agglomerated intentions needn't be jointly realizable
Suppose that you, like Michael Bratman, are prepared to identify intentions (and plans, but then you might even be Bratman) with the functional roles they play in the psychic economy, and to derive various strictures to which they're subject on the basis of their being the sort of thing to play those roles. It seems to me that you are then obligated, when confronted with a question about whether it is possible to intend some A, to ask whether it is possible that something could successfully carry out the roles characteristic of an intention to A, and not to ask whether the intention to A meets the strictures one has already developed. The roles are primary, and the strictures are developed out of the analysis of the clear-cut cases of intending with which one started; there is therefore the risk that, because those cases are clear-cut, the formulation of those strictures may have been overly broad. I think something like this is the case with the purported pressure on intentions that they agglomerate (that is, that the intention to A and the intention to B be combinable into a jointly realizable intention to A and B), on pain of irrationality in the agent.
The case I'm thinking of is this: suppose you're applying to two law schools, Stanford's and Yale's, that coordinate admissions. No one will be admitted to both. (As far as I know they don't actually do this.) It seems, given the formulation of agglomeration above, that it's impossible to intend to be admitted to both, since it's impossible in fact to be admitted to both.
But this ignores an important aspect of the case, namely, that the impossibility of being admitted to both has absolutely nothing to do with me or my actions at all. Suppose I were only interested in Yale, and applied only there; my intention to be admitted to Yale lead me to do things like round up recommenders, write essays, keep track of whether I've sent in my information, etc. If Yale's mailing address changed, for whatever reason, I would track that and change where I mailed my application materials. Whatever roles it's characteristic of intentions to play, there seems to be no problem in their being played out successfully in the Yale-alone case. By parity of reasoning, the same applies in the Stanford-alone case. So what changes when I apply to both schools? Nothing at all, as far as I can see. I write different essays and send them off to different addresses; I provide my recommenders with multiple envelopes (actually IIRC you do this nowadays entirely through LSAC so that wouldn't be an issue, but whatever), and so on. All your fancy cross-temporal and social coordination problems that intentions are meant to solve seem to get solved just fine, and intentions are just the things that solve those problems (& do other things, perhaps, I'm not actually looking at any relevant material at the moment, but whatever other things they do, I'm not sure that the case changes much). I think that if you don't approach the case ready to rule it out on the basis of agglomerativity, there's no real reason not to allow that one can intend to be admitted to both schools, when intentions are defined by their functional roles.
So what's agglomerativity really about? I think something like this is happening: if you look at intentions individually, then you get the requirement that they be means-end coherent; you've got to fill in the intention (or plan) such that it actually gets executed, and that means settling on, sticking to, and actually executing sub-plans as means. I've got no quarrel with that aspect of Bratman's account. If you look at intentions together, then you might think that there's some different requirement, agglomerativity, forcing them to be jointly executable, just as a question of whether the ends are compossible. But I think it makes just as much sense to take agglomerativity as a statement about means-ends coherence among intentions. If I plan to be in Atlanta at 3pm next Friday, and in Boston at 3pm next Friday, I'm obviously in trouble, but it's not obvious that the reason for that is that the ends are not compossible, full stop. (Examples relying on the same person being in multiple locations at the same time probably lend credibility to the independence of a requirement of agglomerativity, I think, because they can seem to be judged not compossible on metaphysical grounds, though that probably depends on your commitments regarding personal identity. I'm going to ignore that, though.) The real problem is that there's just no means I can choose that will put me in Atlanta at the relevant time that also leaves me free to choose a means that will put me in Boston at the relevant time. In fact, modulo supposed metaphysical impossibilities, there doesn't seem to be much more to the claim that two intentions aren't agglomerable than the claim that there's no selection of means such that each intention can be carried out (for what are likely contingent reasons; I can't buy X and Y because I've only got enough cash for one (and can only use cash), say—in the case of impossible agglomerations, of course, it's quite plain that there will be no correct selection of means).
Now, for just about any plan an agent might want to execute, its being successfully carried off will depend not just on actions of the agent h/hself, but on scads of other agents' actions as well. So it's probably better to talk, not of selections of means to the end, but of selections of means to the end that I can carry out. After all, it's the role of my intentions in settling my actions and enabling coordination with others and my future self that's characteristic of them. Two intentions will be agglomerable if there's a selection of means to each, available to me, such that I can carry out those means jointly [insert some hand-waving here about how this isn't just pushing the agglomerativity requirement down into the means, because the necessity of being able to jointly execute the means is already required by means-end coherence, of which I'm arguing agglomerativity is a fallout anyway]. But then agglomerability has been separated from the joint realizability of the ends themselves. We can still say that I can't rationally intend to be in two places at once, or to give one person $5 and another $10, when I've only got $12, but can allow that when the thing that makes two intentions not jointly realizable was never something over which I had any power anyway, I can still intend to do both things.
If the above goes through, then the argument for treating "intend to A" and "intend to try to A" on pp 121–22 of Intentions, Plans, and Practical Reason seems to fall by the wayside, since it depended on the original conception of agglomerativity (there called strong consistency). All to the better, says I; there is much wisdom in Bart Simpson's saying "I can't promise I'll try. But I'll try to try.".
Comments
on 2007-05-05 22:41:59.0, bitchphd commented:
I don't understand a word you're saying.
Luckily it doesn't matter because you probably didn't intend anyone to actually comment on this.
And if you did, well, intentions are irrelevant.
and, further, on 2007-05-06 2:22:11.0, Mike J. commented:
That's very nice. The reasoning in the last sentence of the second paragraph is perhaps compressed in a way that makes it a little misleading. Nothing in the sentence depends on the principle of agglomeration. Rather, it depends on the hypothesis and on the principle that you can't intend what you think is impossible. Since (implicitly) you can still intend to be admitted to each, it follows that the principle of agglomeration is false.
In the second line of the last paragraph, did you omit 'as equivalent' or something like that?
and, further, on 2007-05-06 9:43:49.0, ben wolfson commented:
With regard to the second paragraph: the hypothesis that you can't intend to do what's impossible, in the case of multiple intentions, seems to be just what the principle of agglomeration is supposed to be for (whereas with individual intentions to the impossible, means-end coherence). At least, when discussing this example (I don't know if he's done so in print, but he has done so in classes, where he presents it as an alternative to the linked video games argument from chapters 8 & 9 of IPPR), Bratman himself argues directly from agglomerativity: we know that you don't intend to do each, because you'd then have to intend to do both, but both can't actually be done. So agglomerativity does enter into it, I think.
And yeah, something's missing from the last paragraph, but it's actually "as distinct", not "as equivalent".