Saucy Jack
Brian sent out the following question. I was going to just reply via email, but I thought it better to put it up here, in case anybody else wanted in on the action.
Do you guys get a difference here?
(12) If Prince Albert committed those murders, he is Jack the Ripper.
(13) If Prince Albert had committed those murders, he would have been Jack the Ripper.
If so, how would you make sense of the difference? Any thoughts?
My response (a little rough, but hopefully readable):
The most natural reading of "Jack the Ripper" is as a name, and the most natural reading of (13) is as a counterfactual conditional, (12) as a material conditional. So (12) could be true, and if so is necessary in virtue of the necessity of the consequent (on the other hand, if false, it's necessarily false). If someone's having committed the murders is good evidence that they are JtR, then the antecedent is even good evidence for the consequent, so it could plausibly be given some kind of relevance (non-truth-functional) reading. I know very little about relevance logic, but it seems to me that some kind of Bayesian (evidential probability) sorta guy might do that job for cases like (12) (ordinary, non-counterfactual conditionals).
(13) is weird, since counterfactual conditionals sound so much like relevance conditionals (perhaps even more so than do ordinary indicative conditionals); that's perhaps reason to believe that the Lewis/Thomason semantics for them doesn't capture English usage (since it's open to the same "paradoxes of the conditional" as the ordinary strict conditional is). On the L/T reading, (13) is false only if Prince Albert is not Jack the Ripper (that identity being necessary if true and impossible if false) and it's possible that Prince Albert committed those murders (and so it's probably false). I can't make sense of an evidential relevance reading of (13), probably because I can't make much sense of such readings for any counterfactual conditionals (more on this below). Furthermore, I'm not sure I can give ANY kind of coherent relevance reading for (13), or at least a coherent and true reading. That's because relevance readings typically express exclusivity, and so a conterfactual relevant conditional implies the falsity of its consequent as well as its antecedent. Given the falsity of the consequent (which is then impossible), nothing could have made it true, not even murder. I'm not sure if that's a false relevance reading or just not a coherent relevance reading since there's no relevance. Counterfactual + relevance + identity (or another necessary/impossible proposition) = weird, and probably just false. As I said, I think a relevance reading is the most natural reading (caveat: I know almost nothing about relevance logic), and taken in that way, (13) is either false or incoherent.
The natural readings of (12) (material or evidential relevance) allow that it be true. However, given an evidential relevance reading of (12), I'm not sure what COULD count as evidence for the consequent, since I assume that we (in place of the dudes from Scotland Yard) aren't so strange as to lack the belief that JtR is JtR or that PA is PA, and so the belief that JtR is PA (iff that's true). So I definitely don't want to say that the antecedent could be evidence supporting BELIEF in the conclusion, although perhaps 'evidence' is broad enough to include things that make you recognize stuff (where that's not a matter of belief so much as having the OBJECTS that are involved in your beliefs "lined up" in a certain way in thought).
As for an evidential reading of (13), I have trouble making sense of such a thing because evidential evalutation seems to be tied to my perspective in a way that ordinary truth-evaluation is not. In other words, if I consider a hypothetical situation and what would be evidence for what for me, I can't but evaluate it in terms of my actual evidence; thus counterfactuals can't be given evidential readings that aren't rather weird. In the case of (13), I can only consider my actual evidence for the antecedent and the consequent; if PA is not JtR, then NOTHING is evidence for the consequent, no matter what I would think if I knew/believed different things. This understanding of counterfactual evidence may all depend on my not much understanding a sense of evidence (or reason to believe/recognize, if there is such a notion) that isn't factive. I'm inclined to say that if P is false, then you don't have evidence for P or a reason to believe that P, although you may think you do. Now perhaps there's a coherent category of stuff that's like evidence (or reasons) but not factive. Perhaps one might be inclined to say that it's one's "qualitative evidence", or "seemings", and furthermore that the antecedent of (13) should be understood as specifying conditions in terms of this "kind of evidence". I think that's a rather theory-laden (i.e., poorly motivated) response; I'm not currently convinced that it's incoherent or anything, but I'm mostly inclined to think it's false. In other words, I don't think such a notion (if coherent) is sufficiently evidential to give us a coherent and possibly true reading of (13). I think (13) just doesn't work like that, nor does other evidential talk in English.
Do you guys get a difference here?
(12) If Prince Albert committed those murders, he is Jack the Ripper.
(13) If Prince Albert had committed those murders, he would have been Jack the Ripper.
If so, how would you make sense of the difference? Any thoughts?
My response (a little rough, but hopefully readable):
The most natural reading of "Jack the Ripper" is as a name, and the most natural reading of (13) is as a counterfactual conditional, (12) as a material conditional. So (12) could be true, and if so is necessary in virtue of the necessity of the consequent (on the other hand, if false, it's necessarily false). If someone's having committed the murders is good evidence that they are JtR, then the antecedent is even good evidence for the consequent, so it could plausibly be given some kind of relevance (non-truth-functional) reading. I know very little about relevance logic, but it seems to me that some kind of Bayesian (evidential probability) sorta guy might do that job for cases like (12) (ordinary, non-counterfactual conditionals).
(13) is weird, since counterfactual conditionals sound so much like relevance conditionals (perhaps even more so than do ordinary indicative conditionals); that's perhaps reason to believe that the Lewis/Thomason semantics for them doesn't capture English usage (since it's open to the same "paradoxes of the conditional" as the ordinary strict conditional is). On the L/T reading, (13) is false only if Prince Albert is not Jack the Ripper (that identity being necessary if true and impossible if false) and it's possible that Prince Albert committed those murders (and so it's probably false). I can't make sense of an evidential relevance reading of (13), probably because I can't make much sense of such readings for any counterfactual conditionals (more on this below). Furthermore, I'm not sure I can give ANY kind of coherent relevance reading for (13), or at least a coherent and true reading. That's because relevance readings typically express exclusivity, and so a conterfactual relevant conditional implies the falsity of its consequent as well as its antecedent. Given the falsity of the consequent (which is then impossible), nothing could have made it true, not even murder. I'm not sure if that's a false relevance reading or just not a coherent relevance reading since there's no relevance. Counterfactual + relevance + identity (or another necessary/impossible proposition) = weird, and probably just false. As I said, I think a relevance reading is the most natural reading (caveat: I know almost nothing about relevance logic), and taken in that way, (13) is either false or incoherent.
The natural readings of (12) (material or evidential relevance) allow that it be true. However, given an evidential relevance reading of (12), I'm not sure what COULD count as evidence for the consequent, since I assume that we (in place of the dudes from Scotland Yard) aren't so strange as to lack the belief that JtR is JtR or that PA is PA, and so the belief that JtR is PA (iff that's true). So I definitely don't want to say that the antecedent could be evidence supporting BELIEF in the conclusion, although perhaps 'evidence' is broad enough to include things that make you recognize stuff (where that's not a matter of belief so much as having the OBJECTS that are involved in your beliefs "lined up" in a certain way in thought).
As for an evidential reading of (13), I have trouble making sense of such a thing because evidential evalutation seems to be tied to my perspective in a way that ordinary truth-evaluation is not. In other words, if I consider a hypothetical situation and what would be evidence for what for me, I can't but evaluate it in terms of my actual evidence; thus counterfactuals can't be given evidential readings that aren't rather weird. In the case of (13), I can only consider my actual evidence for the antecedent and the consequent; if PA is not JtR, then NOTHING is evidence for the consequent, no matter what I would think if I knew/believed different things. This understanding of counterfactual evidence may all depend on my not much understanding a sense of evidence (or reason to believe/recognize, if there is such a notion) that isn't factive. I'm inclined to say that if P is false, then you don't have evidence for P or a reason to believe that P, although you may think you do. Now perhaps there's a coherent category of stuff that's like evidence (or reasons) but not factive. Perhaps one might be inclined to say that it's one's "qualitative evidence", or "seemings", and furthermore that the antecedent of (13) should be understood as specifying conditions in terms of this "kind of evidence". I think that's a rather theory-laden (i.e., poorly motivated) response; I'm not currently convinced that it's incoherent or anything, but I'm mostly inclined to think it's false. In other words, I don't think such a notion (if coherent) is sufficiently evidential to give us a coherent and possibly true reading of (13). I think (13) just doesn't work like that, nor does other evidential talk in English.
posted by Luke M @ 9/14/2006 11:04:00 AM
3 Comments:
At 8:03 AM,
Luke M said…
Thanks for the comment, Phil, and welcome aboard. I'm not terribly sure what to make of the category of false justified beliefs. This discussion is sort of tangential to Brian's original question, but it's interesting in its own right. Is there a notion of justification that isn't factive? Is it the same in cases of true belief and false belief? Or do we perhaps want some kind of "disjunctive" account, where fundamentally different things are going on when you're justified and when you aren't but think you are? I'm partial to the disjunctive approach, though I don't think about this subject much and know very little about the literature.
Truth be told, one of my reasons for bringing this up was the vaguely Two-Dimensionalist idea that what matters for certain kinds of justification (a priori justification?) is a qualitative ("internalist"?) description of one's evidence or epistemic state. If that notion is coherent and really is a kind of justification, then perhaps it could be employed in a true reading of (13). So this really may be the heart of the issue, and not a tangent. I'm up really early, so I can't tell right now.
At 8:45 AM,
J said…
Both of the pseudo-conditionals demonstrate another truth: if it's not an established fact (or mathematical relation), don't use it in a conditional. However much "modal logic" appeals to some hipsters, probability ain't logic (nor is physics, which the supposed "causal conditional" generally reduces to).
At 11:50 AM,
Pliny the not so old said…
Seems the obvious interpretation is missed here: the two statements reflect different attitudes of the speaker about the proposition in question (in the second, the speaker does not believe it).
Post a Comment
<< Home