Blog Coordinator

Knobe's X-Phi Page

X-Phi Grad Programs

« Even More Expertise! | Main | Experimental Philosophy of the Self at NYU »

11/06/2014

Comments

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

Clayton

Hi Wesley,

Thanks for writing this up, it's a really helpful guide to the recent literature.

There are two things I wanted to ask about. If you or anyone else had the chance to clarify this, that would be great.

In the papers that you co-authored, you make a pretty convincing case that there are lots of importantly different things going on in the cases described as 'Gettier cases'. In suggesting an alternative more fine-grained taxonomy, is it important for your purposes whether cases that fit into the newer categories are similar in, say, all being cases of knowledge or all being cases of non-knowledge? Some of the data cited above suggests that the respondents were willing to ascribe knowledge in Gettier Category 1 cases. If it turns out that variants that fit the description offered reliably trigger non-knowledge intuitions, would that show that the category would need further refinement?

I also have some questions about the characterization of Gettier Category 1 cases:

GC1: An agent perceptually detects the truth, and there is a salient but failed threat to the truth of her judgment.

One worry has to do with the notion of perceptually detecting a truth, perceiving a truth-maker, and similar ideas that figure in these discussions. I'm somewhat skeptical of the idea of perceptual detection of truth if that means something different from, say, a true belief being produced by perception. Since there's considerable controversy in the literature on perception as to whether truths and truth-makers are things perceived, I don't see why it figures in the discussion. (It's worrisome, in part, because there's a pretty convincing set of arguments in the literature that propositional seeing is sufficient for knowledge, so it's not surprising that cases that are described as seeing that something is so or perceiving that something is so are cases of knowledge as seeing that something is so just is knowing that something is so in a visual way.)

Another worry has to do with the idea that there's a salient but failed threat to the truth of the subject's judgment in the relevant cases. Suppose the subject judges, say, that this is a barn (where 'this' picks out some visible structure). In the standard fake barn case (and failed threat cases) there's no salient but failed threat to the truth of _that_ judgment in the relevant cases. Such a threat would be a threat that the object picked out by the demonstrative would be something other than a barn. That's not what's happening, though, is it? It's not as if the building that the subject sees is a barn but some salient threat could have rendered it a non-barn. Does that seem right to you?

Wesley Buckwalter

Hey Clayton, in general I suspect that there is much more work to do in further refining categories beyond the basic schema we provide above. The finding of knowledge denials in what we were labeling Category 1 might provide a reason for such a revision. On the other hand, other features of such a case, pragmatic effects, small variations in wording, questioning, and so on, might all trigger denials in cases that appear to fit that basic category but which might not provide good reasons to question its fidelity. So it would very much depend on details of the potential discovery.

I think there are some really interesting theoretical questions about the taxonomy relating to the philosophical niceties of perception. The present proposal was simply to try and track knowledge ordinarily understood. With respect to knowledge ordinarily understood, perceptually detecting the truth appears to be a very important variable in the following way. If all else equal people think the subject saw, say, a ground squirrel or barn, they think she knows. And conversely, convince them in the same basic case that she doesn't see those things, and they'll likely think she doesn't know.

We do not test fake barn cases in that particular paper, but do speculate that they seem very structurally similar to Category 1, with recent results supporting the same ascription patterns found there. However perhaps further distinctions could be made to reflect various subtleties in the natures of different failed threats.

Clayton

Hi Welsey,

Thanks for your response, that's helpful. Part of the reason I'm worried about these papers is that I worry about the way that the cases and their significance is described. Maybe this is all harmless, but maybe it isn't. I hope you won't mind if I press a little bit further.

Consider the way that you've refined the unnatural category of 'Gettier-case':
GC1: An agent perceptually detects the truth, and there is a salient but failed threat to the truth of her judgment.

I have three sorts of general worry about the refinement.

Worry 1: You've addressed this somewhat above, but I wondered whether the refined category would be more useful than the blanket term Gettier case if species of the genera marked out by GC1 differ significantly in terms of whether they're classified as knowledge or non-knowledge cases. There's supposed to be some data suggesting that some of these cases are classified by some respondents as cases of knowledge, but I think we agreed that this case is a GC1 case that is a non-knowledge case:

Lucky Penny: Jill has a lucky penny. She hasn’t seen other pennies before. Her brother stole her lucky penny and took it with him to school. He dropped it. Someone picked it up but later dropped it. It worked its way across the city. A week later Jill was on a school trip when she looked down and saw a penny that happened to be her lucky penny. “It’s my lucky penny!” she said.

If it turns out that this is a non-knowledge case and it turns out that other GC1 category cases are cases of knowledge, I'd worry about whether GC1 is any more helpful as a category than, say, the highly disjunctive notion of a Gettier case.

Worry 2: I don't think you've addressed this yet, but standard fake barn cases aren't GC1 cases as GC1 cases are categorized here. Why is that? Well, in a standard fake barn case, the target belief is something like this:

(1) That is a barn. [Where 'that' is a demonstrative that picks out a barn.]

If you have a judgment with the logical form, , you'd think that a salient threat to the truth of the judgment would be a case in which there's a salient threat that threatens to make it the case that a is non-F (or, perhaps, destroy a so that a doesn't exist). There's no such threat in the relevant range of cases. In the relevant range of cases the salient threat is one that would lead the subject to classify some second thing, a thing distinct from a, as a case of F when it isn't. Introducing some b where b is numerically distinct from a where b is non-F is just not a way of threatening the truth of a is F.

Worry 3: This is a worry about perception. You say:

"The present proposal was simply to try and track knowledge ordinarily understood. With respect to knowledge ordinarily understood, perceptually detecting the truth appears to be a very important variable in the following way. If all else equal people think the subject saw, say, a ground squirrel or barn, they think she knows. And conversely, convince them in the same basic case that she doesn't see those things, and they'll likely think she doesn't know."

This doesn't address the worry I had, which is a worry about the coherence of the notion of perceptually detecting the truth. The worry was actually only compounded by the way that you've just moved from perceptual detection of truth to perceptual awareness of a particular.

There's controversy as to whether there's any sense to be made of perceptually detecting the truth if that means something different from this:
(a) S perceptually detects the truth if S forms a true belief on the basis of perceptual experience.

If you want to say that this is all that perceptual detection of the truth comes to, you might have some worries about a model that uses perceptual detection of the truth to cash out perceptual knowledge because there's so little present here about how a perceptual belief connects to experience. If you want to flesh out the idea of perceptual detection of the truth, you could say something along these lines:
(b) S perceptually detects the truth that p if S forms the true belief that p on the basis of a veridical experience with p as its content.
(c) S perceptually detects the truth that p if S sees that p.
(d) S perceptually detects the truth that a is F if S believes that a is F because a has a look that's distinctive of Fs and a's having this look explains S's belief because it triggers a disposition to classify visible things as being F on the basis of a thing's look.

It doesn't look like (b) will be helpful because there can be veridical hallucination cases in which you wouldn't want to say that perceptually detecting the truth is closely connected to knowledge. Moreover, if a subject is told that many of the nearby non-F objects are visually indistinguishable from Fs, she still might have a veridical experience and believe that a is F correctly but it doesn't look like this is a case of knowledge.

It doesn't look like (c) will be helpful because seeing that p just is knowing that p. The people who contest this often say that propositional seeing is some purely visual affair that doesn't require belief, but then if that's right, we should think of the Lucky Penny case as a case in which the child sees that the penny on the ground is her lucky penny. I thought that you didn't think that was a good description of the case, but even if you think that it is, the Lucky Penny case looks like a non-knowledge case, which suggests that the purely perceptual phenomenon at play in that case isn't sufficient for putting the child in the position to know.

You could go with something closer to (d), but then it's an open question as to whether subjects in the relevant range of cases perceptually detect the truth. The Lucky Penny case illustrates this as well as it needs to be. The look that triggers the child's disposition to judge isn't distinctive of her penny, so we wouldn't credit her perceptual capacities with classifying the case correctly on the basis of how the penny looks.

It's hard to tell from what you said whether you had some other model in mind, but you suggested that knowledge might be closely connected to object seeing. The problem with this is that object seeing is usually understood in such a way that you count as seeing an object even when you're in no position to recognize it. If your father is wearing a hood and is so unrecognizable, you see the hooded man before you iff you see your father. If a barn has been disguised to look like a bank or a Pizza Hut, you see the barn iff you see the disguised building. And so on.

I fear that the utility of your model for what's going on in GC1 cases will involve a careful specification of what's going on perceptually in the relevant range of cases and the notion of perceptual detection of truth isn't going to help us see what precisely your model is and so won't help us test to see whether you've identified something that ordinary subjects pick up on that disposes them to classify certain cases as cases of knowledge.

Clayton

"If you have a judgment with the logical form, , you'd think that a salient threat to the truth of the judgment would be a case in which there's a salient threat that threatens to make it the case that a is non-F"

Bah! The logical form of the judgment: a is F. I tried to use brackets and that just rendered the thing invisible.

John Turri

Hey Clayton,

I though it might be useful to mention this here: we already know that your "Lucky Penny" example is judged to be a case of knowledge. I ran it and reported the results here:

http://certaindoubts.com/more-x-phi-on-fake-barn-intuitions/#comment-173172

Clayton

Hi John,

Known to be a case of knowledge by whom? I've run it by a handful of philosophers and they all judged it to be a clear non-knowledge case. I know that you consider it to be a knowledge case, but I think (Wesley can correct me if I'm wrong) that he agreed that it was a non-knowledge case.

John Turri

Hi Clayton,

Just to be clear, I said that we know that this is how people judge the case. I didn't say that it's known to be a case of knowledge. (Though you are correct that, on the most natural way of understanding the case, I do think this is clearly a case of knowledge.)

You wrote, "If it turns out that this [i.e. Lucky Penny] is a non-knowledge case and it turns out that other GC1 category cases are cases of knowledge, I'd worry about whether GC1 is any more helpful as a category than, say, the highly disjunctive notion of a Gettier case." The taxonomy is much more helpful than the highly disjunctive notion partly because it does a much better job of capturing the ordinary concept of knowledge. That remains true even if the ordinary concept misclassifies the case in question.

Clayton

Hi John,

That's helpful. I couldn't tell whether your aim was to isolate something that would be sufficient for knowledge or triggering a knowledge verdict in a sufficiently high number of cases.

Wesley Buckwalter

Hey Clayton, thanks for raising these interesting questions. Given that the taxonomy is organized by ordinary psychological responses, those are what will be relevant for evaluating the place of cases like Lucky Penny, which data show emits a similar ascription pattern. One might disagree that this is the philosophically correct answer, as you seem to, while still finding the taxonomy useful in sorting all these cases according to that criteria.

I’m not sure I follow why a salient threat category can’t be interpreted broadly to include both failing to make F not-F and failing to make someone classify b as F. Perhaps more distinctions could track the difference, though ordinary reactions don’t seem sensitive to this, which again is what the taxonomy was meant to reflect. Also, interesting you say that one kind of threat is “destroy a so that a doesn't exist”, the text from Colaco et al: “This tornado…did destroy most buildings…local townspeople built house facades in the place of destroyed houses.” Part of the story definitely involves the threat of the tornado destroying the house so that it doesn’t exist.

You again raise several very interesting questions about the nature of perception, seeing, recognition, etc. With respect to your worry of the utility of GC1 however, we classify both the Knowledge Control and GC1 partly in terms of agents perceptually detecting the truth. The utility of GC1 involves the failed threat aspect, specifically that judgments given failed threats resemble those of paradigmatic knowledge in the Control. The complex interpretation of perceptual detection wouldn't be a problem for that particular category over what is required for paradigmatic cases of knowledge.

Clayton

Hi Wesley,

A couple of quick points.

"I’m not sure I follow why a salient threat category can’t be interpreted broadly to include both failing to make F not-F and failing to make someone classify b as F."

Because b's being non-F isn't a threat to the truth of a's being F.

"we classify both the Knowledge Control and GC1 partly in terms of agents perceptually detecting the truth."

Right, I can see that, but what is perceptually detecting the truth? I've offered a number of extensionally different accounts of that and I'm not sure which one you have in mind. Your suggestion is that we can use the notion of a failed threat, but I've just explained why that notion isn't well-defined. If I wanted to test whether your suggestion about perceptual detection of truth and knowledge, I'd be at a loss as to how to construct cases because I don't yet know what it means to say that some subject perceptually detects a truth.

Also, just a quick question, didn't you agree that Lucky Penny was a case of non-knowledge?

Wesley Buckwalter

Hey Clayton, the failed threat notion in GC1 is general but it’s perfectly well defined in the paper. It refers to “lucky events that threaten but ultimately fail to change the explanation for why a belief is true”. There are, of course, many different types of lucky events that fit that description. There was robust ascription in the particular ones we tested fitting this general description, and the same seems to be happening in other work done on FBs, which also seems to fit this general description. That led us to suspect that FBs belong in GC1. Perhaps tests of different cases will inspire need for refinement. In any event, I would have thought very similar patterns in psychological responses between the former and the latter suggest similar taxonomic treatment even if there was some mild dissatisfaction over labeling.

In the relevant cases, the suggestion was that whether someone knows depends essentially on whether they detect (i.e. visually see) the object (i.e. the squirrel, diamond, or barn). What we discovered is that in those cases, and seemingly including typical FBs, the intuitive thing to think is that (unless there is added reason to think otherwise) the person sees all they need to in order to know. I am not sure which of your extensionally different philosophical accounts of the nature of perception inform this intuitive psychological response, and I am curious to learn more. However you need not be at a loss for how to (at least begin to) test the suggestion about perceptual detection. You can begin to evaluate this claim by simply noting the details of and comparing the results from the well matched detection controls in the studies provided.

John Turri

Hey Clayton,

Your continued engagement with this work is much appreciated. I've been considering the questions you're asking about the nature of the perceptual relation, and it's clear that these are important and worthwhile questions.

I think that the natural, simplest, and most straightforward answer to your question is that, in fake barn cases, perceptually detecting the truth amounts to *seeing that the object is a barn*. People very strongly agree that Henry sees that the object is a barn (as reported here: http://certaindoubts.com/more-x-phi-on-fake-barn-intuitions/#comment-172919).

When explaining the taxonomy, we use the more general "perceptual detection" language to allow for other perceptual modes.

Does this answer your question?

Jennifer Nagel

By the way, we should add to the roundup Hamid Seyedsayamdost's carefully executed study, "On Normativity and Epistemic Intuition: Failure of Replication", now forthcoming in Episteme. Available here: http://philpapers.org/rec/SEYONA-2 .

Wesley Buckwalter

Hey Jennifer, Thanks for the update! I did not know this had been published, I will add it to the list above. While I think some caution is required in interpreting negative results in experimental philosophy lacking power analyses, effect sizes, and so on, the latest balance of evidence definitely seems to suggest that cross cultural variability is not a significant factor in these kinds of judgments.

Clayton

Dear John & Wesley:

Sorry for the slow response, but the last few days have been packed. The responses have been really helpful in sorting out some issues about how the folk think about the cases.

John,

Your suggestion is that we might use propositional seeing (i.e., S's seeing that p) as a way of cashing out this idea of perceptual detection of truth. That seems right to me. In fact, I don't think there's any other useful notion of perceptual detection of truth besides this one, but I think that that points to a potential worry.

Suppose I wanted to try to test this hypothesis:

H: Perceptual detection of truth is sufficient for knowledge/being in a position to know.

To get a fix on the idea of perceptual detection of truth, I'd use the notion of propositional seeing, so H is really:

H' Seeing that p is sufficient for knowledge/being in a position to know.

Here's the worry. It seems that this equation holds generally:

Identity: S's seeing that p just is S's knowing that p in a visual way.

If I don't have an independent way of thinking about propositional seeing, I don't know how to go about constructing the right sort of cases to test H. Is there some independent way of characterizing propositional seeing? If I don't have some independent grasp on the notion, I don't see how I can test H just as I don't see how I could test H'':

H'': Knowing that p in a visual way is sufficient for knowledge/being in a position to know.

Until I have some independent way of characterizing propositional seeing, I don't see how to test the hypothesis H. That's why I tried to offer some alternative construals of perceptual detection of truth above, but while each construal allowed me to characterize the relevant range of cases without using the notion of propositional seeing, it seemed obvious that the cases so characterized wouldn't be cases in which vision puts a subject in a position to know.

Hi Wesley,

Maybe I shouldn't have used the term 'well-defined'. What I meant was that the relevant kind of case was described in such a way that it didn't apply to the cases that it was supposed to.

Gettier Category 1 was defined as, "An agent perceptually detects the truth, and there is a salient but failed threat to the truth of her judgment." As I explained above, this doesn't fit the fake barn case (although on pp. 10 in section 4 of "Gettier Cases: a Taxonomy", you claim that fake barn cases are GC1 cases. The case differs from News, which does fit GC1. Thus, GC1 doesn't lump the cases together that you lump together in the paper. I don't know if this is a significant mistake, but it is a mistake.). In a fake barn case, there's nothing that threatens the truth of the judgment that this particular object is a barn. The judgment that a is a barn is false if a is a non-barn, but not if some numerically distinct object b is a non-barn.

John Turri

Hi Clayton,

I don't know if thinking about this in terms of logically sufficient conditions is suitable to the ordinary concept of knowledge (or most anything, really).

One straightforward way to test H or H', without worrying initially about providing "an independent way" of understanding propositional seeing, is to devise a range of scenarios and gather judgments about whether the agent (i) sees that P, and (ii) knows that P. Then we could see whether, for any scenario, people are willing to judge that the agent sees that P while denying that the agent knows P. Relatedly, we could look at the extent to which propositional-seeing attributions outstrip knowledge attributions.

If the results I linked to above are any indication, it's somewhat likely that people will be more generous with propositional-seeing attributions than knowledge attributions. If things did turn out that way, then it would be evidence that Identity is false (in folk epistemology), even if propositional seeing and visual knowing are intimately related.

PS: I'm somewhat unsure how these questions about characterizing propositional seeing are in any way especially worrying for our work. In our cases, we don't say that the person "sees that P" and then ask whether the person "knows that P." At most, we characterize the seeing objectually. This is right in line with Goldman's original presentation of the fake barn case, in which he writes, "The object [Henry] sees is a genuine barn." Goldman also says that the "object is fully in view, Henry has excellent eyesight, and he has enough time to look at them reasonably carefully." (Goldman then claimed that, despite that information, in the fake-barn case, we will be "strongly inclined" to not attribute knowledge. But this behavioral claim is clearly false.)

Clayton

Hi John,

I agree with much of what you said in your response, but I'm still at a loss as to how to make sense of the taxonomy provided in terms of 'perceptual detection of truth' because that appears to be a technical notion. I don't think I understand how the notion has been defined. If I wanted to test to see whether subjects are willing to ascribe knowledge in cases in which there's perceptual detection of truth (with or without some salient threat), I'd need to construct cases in which we could agree that there's perceptual detection of truth going on and then see whether subjects ascribe knowledge. I don't think that perceptual detection of truth can be characterized _simply_ in terms of seeing objects for reasons sketched above (e.g., cases of seeing barns where the barns have been disguised to look like banks are probably not going to be the kinds of cases that would be helpful for testing the conjecture), so I'm looking for some guidance so I can see how to set up novel cases to test conjectures.

"This is right in line with Goldman's original presentation of the fake barn case, in which he writes, "The object [Henry] sees is a genuine barn." Goldman also says that the "object is fully in view, Henry has excellent eyesight, and he has enough time to look at them reasonably carefully." (Goldman then claimed that, despite that information, in the fake-barn case, we will be "strongly inclined" to not attribute knowledge. But this behavioral claim is clearly false.)"

It might be in line with Goldman's telling of the case, but I think he and others have been criticized for underdescribing the case. On some descriptions and for some ways of assigning a semantic value to 'we', he seems to be right. On others, he seems to be wrong. I fear that we're not going to make much headway in understanding a potentially diverse range of cases until we say something more about the cases than that it's a case of seeing a barn (e.g., something about the link between what's present in experience and the perceptual belief, say, or something about belief and looks/appearances).

Wesley Buckwalter

Hey Clayton, you write, "In a fake barn case, there's nothing that threatens the truth of the judgment that this particular object is a barn." I have to admit I'm still having trouble following your objection to the category as defined. Doesn't a tornado destroying various objects in the area constitute a salient but failed threat that this particular object is a barn?

I'm also sorry to hear you're still having trouble with the psychological taxonomy due to the fact that there's many philosophical interesting notions of perceptual detection. In this particular circumstance, it might not be too helpful to think of 'perceptual detection' as a complex technical notion. Conversely, it is a label that captures a general feature of cases that seemingly makes a very important contribution to ascription practices. I have attempted to describe what that is above, and noted how you could evaluate it by reviewing the various experimental controls in which it is present and absent. I'm not sure how this is an objection to this (initial) taxonomy, unless the point is just that the psychological responses in hand don't readily discriminate between various philosophical proposals of perception, which again I would agree with.

Clayton

Hi Wesley,

I don't think I'm doing a very good job getting my point across. Let me try this in a slightly different way.

Consider this argument:
P1. Goldman's fake barn case is a Gettier Category 1 case.
P2. In a Gettier Category 1 case an agent perceptually detects the truth, and there is a salient but failed threat to the truth of her judgment.
C1. In Goldman's fake barn case, there is a salient but failed threat to the truth of her judgment.
P3. The judgment in Goldman's fake barn case is the judgment that a is a barn (where 'a' is a singular term that picks out a barn that the subject sees).
C2. In Goldman's fake barn case, there is a salient but failed threat to the truth of the judgment that a is a barn.

I think that (C2) is false, but it validly follows from the premises above. I think that you accept all three premises in the taxonomy paper.

If you think that there is a salient but failed threat to the truth of the judgment that a is a barn, what is it?

John Turri

Hi Clayton,

We are not using 'perception' in any technical sense, really. These are just normal cases of perceiving an object (along with tightly matched controls that do not involve perception of the object).

You wrote, "If I wanted to test to see whether subjects are willing to ascribe knowledge in cases in which there's perceptual detection of truth (with or without some salient threat), I'd need to construct cases in which we could agree that there's perceptual detection of truth going on and then see whether subjects ascribe knowledge." Actually, you would not need to do that. In my last comment, I described an experimental design that avoids the need for that.

Clayton

Hi John,

One general worry I had about the taxonomy was this. If it's a useful taxonomy, it should be one that extends to a new range of cases but I couldn't see how to do that because the notion of perceptual detection of truth was unclear but was all the guidance you guys gave in specifying the category Gettier Category 1.

If we drop the talk of perceptual detection of truth and just stick with object perception, that's helpful in some ways (we have an independent grip on what that is) but then you said that there were certain things we'd have to control for. Do these cases count as Gettier Category 1 cases?

Red Diamond
To hide her diamond, Agnes colored her diamond red and placed it into a bucket of rubies. A jewel thief broke into Agnes' flat looking for diamonds. The thief had read in a reliable but flawed guide that there were red diamonds and white diamonds. (There was a picture in the book that showed a ruby but was described as a red diamond. In fact, there are no red diamonds.) She saw Agnes' diamond in the bucket of rubies, formed the belief that it was a diamond, pocketed Agnes' diamond and left. Did the thief see that it was a diamond? Did the thief know that it was a diamond?

White Diamond
Agnes had a bucket of fake diamonds marked 'diamonds' that she left in her flat to attract the attention of any jewel thieves that might break in. Each stone in the bucket looked like a real diamond. What she didn't realize is that one of the hundreds of stones in the bucket was indeed a real diamond. That stone happened to be sitting on top. A thief knocked the bucket over clumsily, saw the stones spilled across the floor, saw that the bucket was labeled diamonds, and grabbed the first stone that she could believing it to be a diamond. She left the others because she thought that she heard someone coming. She happened to grab the only diamond in the flat. Did she see that it was a diamond? Did she know that it was a diamond?

Experience Machine
Alice, a subject in a Nozickean experience machine, would have the "experience" of waking every morning at 8 and it looked just as if she was staring at the ceiling of a flat. The ceiling appeared to be about 8 feet away and painted white. One morning, the technician accidentally turned off the machine, so rather than have the computer simulate the experience of seeing the ceiling Alice saw the ceiling for the first time in twenty years. It just so happened that the experiences she had on this morning were indistinguishable from the experiences she had the past twenty years. Did she see that the ceiling was white? Did she know that the ceiling was white?

It seems that these are all perfectly good cases of object seeing in which the experience of the object causes a true belief, but they don't seem to me to be clear cases of 'perceptual detection of truth' and they strike me as being clear non-knowledge cases. I get the feeling that you want the GC1 cases to be object seeing + x where I'm not quite sure what the value of x is, but maybe that's not right and these are all perfectly good GC1 cases.

John Turri

Hi Clayton,

I think it's an interesting question which factors are relevant to the ordinary understanding and attribution of propositional seeing. I don't know how to classify these new cases you're proposing. In order to start getting a handle on precisely what's going in with them, we'd need minimally matched pairs differing in only a single respect. (I'd suggest starting with simpler cases.) Then we could see which manipulations caused significant changes in the perception attributions. Initially, at least, it might be worthwhile to just aside the knowledge attributions, until we had a better grip on what was going on with the perception attributions.

Clayton

Hi John,

One of my main worries about the papers linked above was that the notions of perceptual detection of truth, detection of truth, perceiving a truth-maker, etc. weren't sufficiently clear for me to get a grip on and that it wasn't very helpful to bring in the notion of seeing objects because that's a highly diverse range of cases. One reason that I'm curious to learn more about this is that I'm curious about the relationship between vision and knowledge. If someone though that visual contact with an object, truth-maker, etc. was nearly sufficient for knowledge, I think that the last three cases I've described are a nice challenge to that view. So far as I can see, these cases are similar in these three respects: there's object seeing, there's veridical experience, and there's the formation of a true belief in response to this experience where the true belief is about the object the subject sees. On its face, that would suggest that these three cases should all be classified as GC1 cases, but I'm open to an argument to the contrary. I take it that such an argument would have to bring out some difference between these cases and the standard fake barn case that doesn't simply speak to whether the subject knows but also speaks to the suggestion that all these cases are equally good in terms of whether the subject perceptually detects the truth, perceives a truth-maker, etc.

John Turri

Hi Clayton,

People's judgments about these things are not monotonic. The cases you presented are complicated and, to me at least, confusing. Without minimally matched controls, there would be no way to tell what people were responding to and, hence, whether the cases point to useful adjustments to the taxonomy.

Clayton

Hi John,

Fair enough. I didn't think that the cases were that much more complicated than the cases you'd find in the Gettier case literature. All I did to construct them was take the visual relations to objects that seemed to be present in your cases and then tweaked the following features:

* The subject saw an F, had a veridical experience, classified it as an F on the basis of experience using a feature that the thing had that's not typically a features of Fs.

* The subject saw an F, had a veridical experience, classified it as an F on the basis of experience using a feature that the thing had that is typically a feature of Fs while seeing a lot of non-Fs that also had the same visible properties. (A fake barn case where lots of fakes were visible, not merely potentially visible.)

* The subject saw an F, had a veridical experience, classified it as an F on the basis of experience using a feature that the thing had that is typically a feature of Fs after having undergone a series of indistinguishable hallucinations.

My aim was to preserve the visual relations to the objects that the subject saw with the aim of creating new cases that had object perception, veridical experience, the formation of a true belief on the basis of veridical experience with the aim of seeing whether this was sufficient for any of the following:

(a) perceiving a truthmaker
(b) perceiving a fact
(c) perceptually detecting the truth
(d) seeing that
(e) knowing that

I think there are some philosophers (e.g., Colin McGinn) who take (a)-(d) to be purely visual affairs, but these are the kinds of cases that convince me that he's mistaken. Anyway, I had read you (and the others) as defending a similar view, a view on which object perception + veridical experience + true belief grounded in the experience in the normal way would be sufficient for (a)-(c) and possible sufficient for (d)-(e). I had designed these cases to test these two sufficiency claims that I thought that you were interested in, but if they're not claims that you're interested in, then the cases won't be very helpful for your purposes.

The comments to this entry are closed.

3QD Prize 2012: Wesley Buckwalter