Our books

Become a Fan

« Philosophical Books as Presents | Main | New Contributors »



Feed You can follow this conversation by subscribing to the comment feed for this post.

Marcus Arvan

Hi David and Moti: Thanks again for sharing your paper with us, and kicking off our Working Paper Group!

I just gave your paper a read. I think you're tackling an interesting issue, and I'm very sympathetic with the basic thrust of your argument (I've always found conceivability arguments dicey myself!). However, I have a worry about your premise (4): the premise that it's conceivable that CP is false. My first reaction was: I'm not sure CP's falsity is conceivable. So I think you need to give an argument in favor of (4).

Now, in order to conceive of CP's falsity (your premise (4)), one must be able to conceive of a scenario where a person can conceive something P but their conceiving it does not provide even prima facie evidence that P is metaphysically possible. Are there any such scenarios?

Here's one possibility: Chalmers' famous zombie case. Zombies, as we all know, would *themselves* think they could conceive beings just like themselves that "lack consciousness" -- which is of course absurd because *zombies* lack consciousness. What they think they can conceive (beings just like them without consciousness) is, their seeming to conceive it aside, *metaphysically impossible* by definition (since zombies don't have consciousness to begin with).

Alas, we all know what Chalmers would say (has said?) about this sort of thing. He would say zombies *fail* to successfully conceive the scenario in the first place. They *think* they can conceive of beings just like them that "lack consciousness" -- but in fact when they think they can conceive this, they fail. And in that case the connection between conceivability and metaphysical possibility is retained.

In short, in order to (better) defend your premise-(4), I think you need to (A) give a scenario in which it is conceivable that CP is false (the zombie case described above?), and then (B) defend the scenario against the objection that it's just a case of *seeming*-to-conceive.

What'cha think? I hope you find this comment helpful. I'm sympathetic with your general line of argument, and I've always found the Chalmers-type reply frustrating!

Moti Mizrahi

Thanks very much for these thoughtful comments, Marcus. They are really helpful. I agree that we need to say something in defense of (4).

Marcus Arvan

I've been wracking my brain all day trying to come to terms with the issue, and the following thought just occurred to me. Couldn't you say that it's clearly *conceivable* that Chalmers' argument is incorrect? Offhand, if *anything* is conceivable, I want to say, that is! ;) But now in that case, if it's conceivable that Chalmers is wrong, then it's at least *conceivable* that zombies can conceive what Chalmers says they cannot. But suppose that even though it's conceivable that Chalmers is wrong, he is in *fact* right (zombies cannot conceive beings just like themselves that lack consciousness). In that case, because it is *conceivable* that Chalmers is wrong (but he is not in fact wrong), we have a case where conceivability is not even prima facie evidence of possibility -- which is what you need to establish your premise (4). Does this work? I know it's convoluted, but I think it just might!

Clayton Littlejohn

Hi David and Moti,

I've really enjoyed your paper. I should say at the outset that I think that the line that you're pushing is fruitful. One thing you might address as you polish this paper up for submission is the link between (i) the reliability of conceivability as a guide to possibility and (ii) the suggestion that if someone conceives that p, this subject has a p.f. reason to believe p is possible. On some standard views of justification, it is possible to have a p.f. reason to believe, say, that things are as experience suggests even in circumstances where experience is reliable. It might help to say whether the argument rests on reliabilist assumptions (or not).

David Morrow

Thanks, Marcus and Clayton. Some initial thoughts:

Clayton--I've been wondering whether CP is more plausible on some interpretations of prima facie evidence than others. That probably is something we'll want to address. With regard to the phenomenal conservatism that you mention, though, is something's being conceivable a way of seeming metaphysically possible? If not, would the p.c. line still count it as prima facie evidence for something's possibility.

Marcus--It would certainly be a problem if premise 4 were false! I think I can conceive of a human(like?) being whose cognitive faculties are just not good at recognizing metaphysically (im)possible things. But citing a particular example seems like it might be question-begging. I think it helps to note that CP is kind of trivial if conceivability is merely a guide to logical possibilitym, which is in turn evidence of metaphysical possibility; or if the distinction between actually conceiving of something and merely seeming to do so were the same as the distinction between thinking of something that is metaphysically possible and something that isn't.

Robert William Fischer

David and Moti,

Here is another way to defend P4. (This may just be a variant of Marcus' argument; I'm not sure.)

P4 seems to be true. But suppose that it's false. If so, then the appearance of having conceived that p is not perfectly reliable evidence of having conceived that p. When shouldn't we trust the appearance of having conceived that p? Here a variety of answers are possible, but my guess is that they'll share this much in common: we shouldn't trust the appearance of having conceived that p when p's possibility is highly controversial. And if that's right, then this objection to your argument ("P4 is false") is cold comfort to those who would use conceivability arguments to establish philosophical theses, since just about *all* philosophical theses are highly controversial. (You might cite van Inwagen's "Modal Epistemology" here. Also, Peter Hawke has a more recent defense of van Inwagen's epistemic distinction between ordinary and extraordinary modal claims, which is essentially the distinction that my argument trades on.)

It's also worth noting that Williamson's counterfactual-based modal epistemology is supposed to deal with the evolutionary argument that you discuss in the Objections section. A number of people have argued (convincingly, by my lights) that Williamson's proposal doesn't actually manage this. But you might want to add a chunky footnote that refers to the relevant literature. (Tuomas Tahko has a paper on this. I can't recall the others at the moment, but I'll get you the names if they come to me.)

I hope this feedback is useful!


Moti Mizrahi

Thanks for the excellent comments, everyone.

Here is a thought about the issue concerning (4). Metaphysical possibility is usually taken to be distinct from logical possibility. But conceivability is usually supposed to be a guide to l-possibility as well. So, suppose p is conceivable, how do we know whether to infer 'p is l-possible' or 'p is m-possible'?

I take it that this is partly why MP is rarely explicated other than to say something informal along the lines of "how God might have made things." In that case, I think that David is right in saying that God might have made creatures whose cognitive faculties are such that they cannot "discover" modal truths. God could even have created creatures whose cognitive faculties are inverted such that if p is m-possible, they would consider p m-impossible, and vice versa.

So this is one way to go. I also like Bob's suggestion for defending (4). I think it is another good way to go about defending (4).

Clayton: Could you please say a bit more about why you think that our argument might "rest on reliabilist assumptions"?

T. Parent

Dear Moti and David,
Cool issue. I like very much the self-application of CP that occurs early on.

One observation is that premise I of the Nozickian argument seems unnecesarily strong. The problem is that, even if a modal module has no evolutionary advantage, it could be a spandrel or the result of evolutionary drift or something. Nevertheless, you could revise the consequent of I and the antecedent of II to read "there is no reason to think our modal module is reliable..." What results seems to enjoy some plausibility... (though there is still Clayton's sort of point to think about).

But on further thought, doesn't the Nozickian argument render the initial argument otiose? After all, conceivability is held to be one route to justifying (at least some) modal truths. But if modal beliefs are generally not justified, then CP in particular cannot be a way to justify them, yes? (This needn't be a criticism; it's perhaps just the point that you have two papers here instead of one.)

(I also have questions about the second reply re: physical possibilities vs. physical possibilities for us. But I'll stop for now in case I'm failing to track something...)

Moti Mizrahi

Thanks for your comment, Ted.

I am still not sure what Clayton is worried about. Perhaps you can help shed light on his concern. Our reductio is supposed to show that anyone who endorses CP is also committed to the claim that s/he doesn't believe that CP is true. So, strictly speaking, the issue is not reliability, but rather (Moorean) absurdity. But perhaps I just misunderstood what Clayton is worried about.

Clayton Littlejohn

Hi Moti & David,

Moti asked, "Could you please say a bit more about why you think that our argument might "rest on reliabilist assumptions"?"

I had P6 in mind -- 6. If it’s possible that (CP) is false, then there are possible worlds in which conceivability is not a reliable guide to metaphysical possibility.

Earlier in the proof, CP had to do with p.f. reasons. At this stage, the notion of reliability enters in and it's not entirely clear to me what the relationship is between (i) conceivability and possibility and (ii) the reliability of conceivability as a guide to possibility. Maybe these aren't the best views, but there are views on which unreliable experiences can justify beliefs (e.g., the experiences one would have if one were a BIV) and views on which actually reliable processes can justify beliefs even if it is contingent whether the process is reliable and one has no reason to believe the process to be reliable (e.g., a simple-minded reliabilist view on which a belief is justified if the process that produces it is reliable and the subject has no defeaters) (Although, perhaps the idea is that the (epistemically? metaphysically?) possible unreliability of the process is itself a defeater?).

Would it matter if we rewrote P6 and the rest of the proof so that there's no reference to reliability at all? That way, it would read something like this:
P6'. If it’s possible that (CP) is false, then there are possible worlds in which conceivability does not provide a prima facie reason...

My main concern in the last post was simply that there were some assumptions about justification operative in the argument that might be controversial and spelling them out could be helpful.

Having looked at the argument again, I do have a further worry about P8. It might help to clarify the argument by spelling out the modality of the "could" in:
"8. If there are possible worlds in which conceivability is not a reliable guide to metaphysical possibility, then the actual world could be one of those possible worlds in which conceivability is not a reliable guide to metaphysical possibility."

Is the idea that the actual world could _for all we know_ be one in which conceivability is not a reliable guide to metaphysical possibility?

That doesn't seem totally crazy to me, but it also doesn't strike me as being self-evident. I might try to motivate the argument as follows. Imagine a possible world in which everyone can conceive of violations of Leibniz's Law. That's a world in which conceivability is not a reliable guide to possibility. So, it's possible that conceivability is not a reliable guide to possibility. And so, P8 tells us that our world could be one in which conceivability is not a guide to possibility.

At that point, I'd object. Look, the specific grounds we had for saying that there's a possible world in which conceivability isn't a reliable guide to metaphysical possibility don't apply _here_. And so it doesn't follow from the mere fact that there's a PW in which conceivability isn't a reliable guide to possibility that for all we know _our_ world is one in which conceivability isn't a reliable guide to possibility. We don't have those worries about reliability because we don't typically take ourselves to conceive of counterexamples to Leibniz's Law.

Maybe I can put the point this way. If you claim that it's conceivable that there are worlds in which CP is false and that it's conceivable because these worlds have some specific feature, F, that doesn't give us any reason to think that our world could be one in which CP is false if our world is a non-F world. (By way of analogy, I might conceive of a world in which experience is unreliable because the experiences these subjects have regularly come into conflict with other experiences (e.g., perceptual illusions are the rule, not the exception). On this basis, I say that it's possible for there to be a world in which experience isn't reliable. I wouldn't then say, on this basis, that our world could be a world in which experience is unreliable because the specific reasons for concern about reliability that arose in the PW don't arise in our world.)

Justin Snedegar

Hi David and Moti,

I just read over the paper, and really enjoyed it. Just a few points:

First, you claim that the argument shows that defenders of CP are committed to the Moore's paradoxical claim, 'CP is true but I don't believe it'. But that isn't quite what it shows. Instead, it shows (if it works, of course) that defenders of CP are committed to something like 'CP is true, but I ought not believe it'. This isn't straightforwardly Moore's paradoxical, and in fact doesn't sound quite as odd to me (it sounds like the person is being epistemically irrational, or something, but not infelicitous.) Nevertheless, I think this is a small point, since this is still a troubling result for defenders of CP.

Second, premise 3 states: If it's conceivable that CP is false, then it's possible that CP is false. But all we can get from the CP, as you've stated it in its weaker form, is: If it's conceivable that CP is false, then we have a prima facie reason to believe that it's possible that CP is false. Then other premises would have to change, too, saying, for example, that we have a p.f. reason to believe that there are possible worlds in which conceivability does not give us p.f. reasons to believe in metaphysical possibility, and so on. I wonder if everything will still seem plausible, if we rewrite things like that?

I think that if you made these kinds of changes, you'd want to appeal to a principle like "If there's a pf reason to believe that p, then we ought to believe that p unless we have reason to believe that ~p, or the pf reason to believe that p is undercut". Then you could say that (i) we don't have a particular reason to believe that the actual world is one of the CP worlds, and (ii) there's no undercutter for our pf reason to think that it isn't.

Third, like T. Parent, I'm confused by the physical possibility-metaphysical possibility analogy later in the paper. I'm just not sure what you mean by the claim that what humans can do does not fix what is physically possible. Here are two propositions, one of which is true and the other of which is false, where the 'can' is read as physical possibility:

p1: Humans can fly unaided.
p2. Birds can fly unaided.

Surely facts about what humans can (phys. poss) do fix (/constitute) the truth of propositions like p2. They don't fix the truth of propositions about birds, like p1. But the 'what humans can (physically) do' in premise I of the argument about physical possibility seems to play a very different role than 'what humans can (physically/psychologically) do' would play in the analogous argument about metaphysical possibility. (I'm assuming here that what humans can conceive of is a physical/psychological fact about humans.)

Finally, there's a note you've inserted attached to the sentence "...there may be scenarios that are m.possible even though we humans cannot conceive of them". The note worries that the defender of CP can just point out that all this would show is that inconceivability doesn't give us evidence for impossibility. But I think that this kind of principle is implausible anyway, since it seems likely to me that there are lots of things that are actually true, or even necessarily true, that are simply too complicated for humans to conceive of. So I don't think the defender of CP needs to be, or even wants to be, committed to this principle.

That's it -- hopefully something there is helpful. Thanks for sharing the paper!

Moti Mizrahi

Thanks for the clarifications, Clayton.

Justin, thanks very much for your very helpful comments, too.

David and I are revising the paper. If you guys are interested, we could post a revised version on Dropbox.

Thanks again to you all.

Mark Alfano


It occurred to me while reading your paper that there might be an even more damning argument in the vicinity. I think that it's conceivable that (CP) is necessarily false. It would then follow that it's possible that (CP) is necessarily false. Then, by S5, (CP) is necessarily false, because p implies p.

I don't like ontological arguments, but that's what you have to deal with when you start doing serious metaphysics, I suppose.

Mark Alfano

oops, the last sentence of the first paragraph above should have said "Then, by S5, (CP) is necessarily false, because possibly necessarily p implies necessarily p."

David Morrow

Interesting, Mark. I imagine we'd get much the same retort that Marcus thought the friends of (CP) would offer: "Give me an argument that it's conceivable that (CP) is necessarily false." Another response, I suppose, would be to say that the conceivability of (CP)'s necessary falsity only provides a prima facie reason to think that it's possible (CP) is necessarily false, but this prima facie reason is self-defeating.

David Morrow

Clayton: Thanks for these comments. Regarding your last point, the idea is, as you say, that the actual world might, for all we know, be one in which CP is false. But the reason for this is not that we can point to some (non-controversial) case of a proposition that is conceivable but metaphysically impossible. It's that Moti and I (think we) can conceive of a possible world that is (otherwise?) exactly like ours in which our cognitive faculties simply aren't "hooked up" to metaphysical possibility in the right way. And so, if CP is true, we have prima facie evidence that our world could be one in which CP is false.

Justin: Between your comments and Clayton's, we're planning to replace reliability talk with "prima facie reasons" talk. Regarding your last point, I guess my comment wasn't clear about the point of my question. I agree with you that "inverse CP" (inconceivability is evidence of impossibility) is not very attractive. Friends of CP could easily deny it. I'm worried, though, that some of what we say only undermines "inverse CP." And if friends of CP can easily deny "inverse CP," then undermining "inverse CP" doesn't do much to undermine CP itself. I take it that you think we're right to worry about this.

Justin Snedegar


Regarding the last point about your comment: Sorry, I wasn't very clear about that point. Yes, I understood your worry, and think you are right to be worried about it for the reasons you point out -- "inverse CP" isn't very attractive anyway, so friends of CP should just deny it.

Moti Mizrahi

Thanks for the follow-up comment, Justin.

In “Conceivability and Possibility,” Szabo-Gendler and Hawthorne explicitly say that the conceivability-possibility move could go both ways (p. 2). Perhaps, when pressed, they would deny that. But I am not sure that they can do so consistently. For, supposedly, one is entitled to infer ‘p is metaphysically possible’ iff one can conceive of a scenario in which p obtains. In that case, if one cannot conceive of a scenario in which p obtains, doesn’t it follow that p is inconceivable, and hence metaphysically impossible?


David and Moti,
This is an interesting issue, and one that I've thought about myself. Your argument seems strangely familiar, but I can't remember where I might have seen something similar to it. Anyway, I'm curious to see how you develop this line of thought.

Now, I only glanced at the previous comments, and it seems that Clayton already brought this up, but the move to the intermediate conclusion (6) seemed very problematic to me, since CP only mentions prima-facie reasons, not that conceivability would be a reliable guide to met. poss. Indeed, I think that this is where the meat is in the conceivability/metaphysical possibility debate, for even if we grant the prima-facie reason, that entails no reliability whatsoever. Of course there is an overlap between what is conceivable and what is metaphysically possible, but the problem is to distinguish conceivable metaphysical impossibilities and metaphysical possibilities. Given that there's probably an infinite number of the former, this seems like an insurmountable challenge unless we have something more than a prima-facie case.

Whether this is a major problem for your paper is another question, but some note of the lack of connection between prima-facie reasons and reliability should probably be made.

Verify your Comment

Previewing your Comment

This is only a preview. Your comment has not yet been posted.

Your comment could not be posted. Error type:
Your comment has been saved. Comments are moderated and will not appear until approved by the author. Post another comment

The letters and numbers you entered did not match the image. Please try again.

As a final step before posting your comment, enter the letters and numbers you see in the image below. This prevents automated programs from posting comments.

Having trouble reading this image? View an alternate.


Post a comment

Comments are moderated, and will not appear until the author has approved them.

Your Information

(Name and email address are required. Email address will not be displayed with the comment.)

Open job-market threads