Monday, November 12, 2012

New Essay: The Problem of Known Illusion and the Resemblance of Experience to Reality

I'll be presenting a new essay at the Philosophy of Science Association meeting in San Diego on Friday in the morning sessions. I've been drafting it out since 2010, in various shapes and lengths, and I presented it orally in 2011 at UMSL, but the thing always seems to crumble in my hands, and until now I haven't been comfortable posting a circulating draft. However, by stripping it down to 2000 words for a brief oral presentation, I can conveniently decline to delve into the issues that keep stymieing me and present the core idea fairly simply, I hope, with a couple of examples.

Abstract: If Locke is right, when I visually experience a cubical thing and judge rightly that it is in fact a cube, then there is a mind-independent thing out there the shape of which in some important way resembles my experience of its shape. If Kant is right, in contrast, we have no good reason to think that things in themselves are cubical; there's nothing independent of the human mind that has cubical properties that resemble the properties of my visual experience of cubes. I believe we can start to get a handle on this dispute empirically through introspection. Suppose that there are multiple different ways of veridically experiencing the same object and that it can sometimes be the case that there's no good reason to think that one of the two different experiences more closely resembles things as they are in themselves. It would then seem to follow that there's a kind of looseness between features of experience and features of things in themselves. Things in themselves might be more like this or they might be more like that or somewhere in between; but we can no longer say that we know they are like this -- a miniature Kantian victory over Locke. And then the question would be: How far can we push this type of argument? In this paper, I consider two test cases: convex passenger-side car mirrors and inverting lenses of the sort invented by George Stratton.

Full paper here. As always, comments and criticisms welcome, either on this post or by email.


Kevin Reuter said...

Dear Eric,

Very nice paper, but I think there is a difficulty you don't go into.

Let me use my own example to make the point and then apply it to your examples. If you look at a white wall that is illuminated by red light, then you might judge that this is a white wall. However, you do use other information (cues, backgroundknowledge, etc.) to make that determination. If I had blindfolded you before and had asked you to immediately answer what the colour of the wall is, once you I had unblindfolded you, you would have (most likely) said 'red'. So, if I had given you more time, you would have been able to pick up cues from the environment and use backgroundknowledge to determine the real identity of the colour of the wall. But using this additional information seems not acceptable to me if you want to respond to the Locke/Kant question. Making inferences (even if they are unconscious) should not be involved in determining how things experientially look to you.

I think the same applies to the oar, lenses and mirror cases. If you are given backgroundinformation (the oar is in water, the mirror is convex, you wear inverted lenses), then you might say that things 'look' the same way that they look in normal circumstances. But this 'looks' language is not any longer a pure experiential look, but rather a partly epistemic one.


Eric Schwitzgebel said...

Thanks for the thoughtful comment, Kevin! I agree with your general point, and indeed this is one of the (many!) things I really need to think through better before this project is fully mature.

Here's one preliminary thought, though: It doesn't seem to me that the convex passenger mirror case is like what you describe -- at least that's not the interpretation I'm most drawn to. It's not (I think) that our visual experience is of the car as small (or distant?) and knowledge that we are looking in the mirror then compensates for that. And if it's not the case for the passenger's side mirror, then I can construct a series of cases to the bent oar, starting with hypothetical pieces of car windows that use refraction to open up a wider field of view -- a garden path style argument to the treatment of the oar.

And yet, I agree, that what I get at the end of the garden path strikes me as wrong. It would be cool and elegant if it were true, though, so I'm definitely tempted! Partly, I do think it's an empirical/introspective matter....

Kevin Reuter said...

Hi Eric,

let's focus on the mirror case. You present three arguments for your case, and I think I disagree with each of them.

1. Your second solution to the question, "objects in convex mirrors are larger, not closer than they appear", seems to me to be clearly wrong but only because the object is not presented on its own. The image of the car certainly takes up less space on the convex mirror, and consequently less space in the visual experience. But the car is not presented on its own, so we immediately relate the size of the car to the road, trees and so on, and thus it seems the car cannot be larger than it appears (it could therefore not be a 3-foot-tall car). This seems to be similar to what Peacocke has in mind, when saying that trees along an avenue look the same size although they take up different portions of the visual field.

2. You talk about skilled drivers, but what about the unskilled drivers? And what about the reasons for the Federal Government to put the warning in place? This gives us some strong reasons to believe that cars in convex mirrors appear to be further away to the unskilled driver. What makes a skilled driver adjust to the convex mirror; is it that the experience itself changes or that the inferences he draws from having these experiences change? To me the latter sounds much more plausible.

3. My intuitions diverge from yours. I am not a regular driver, and when I do rent a car, the car
does look further away to me when I look at the convex mirror.

Eric Schwitzgebel said...

Hi Kevin, thanks for following up!

On 1: Yes, clearly the road and trees would also have to look smaller, not just the car. Geometrically, that seems to me as much of a possibility as that they look farther. Admittedly, though, if we integrate over time incorporating head motions, things get more geometrically complex. On the non-obviousness of Peacocke's claim, consider the fact that many philosophers have asserted otherwise over the years (as detailed in my 2006 article and ch. 2 of my 2011 book).

On 2: The grounds for the warning were dubious even when it was first instituted, by the standards of research at the time. Setting that aside, my Multiple Veridicalities view accepts that the experience in some importance sense stays the same -- cf. my discussion of "teavy" and "toovy" with inverting lenses. It stays "toovy" but now that you're an expert, "toovy" is a way (a passenger's side convex mirror way) of a car's looking 6 feet tall and 60 feet behind.

On 3: I agree that such introspective testimonials are dubious -- and I include my own. However, I also think they are essential. Hence my general tendency toward doubt and skepticism on these matters. Not that I think we should give up!

Anonymous said...

I am a regular driver and the images in the mirror are as far as i can tell appear exactly the right distance away.. A mirror without that adjustment would annoy me (and so i have been annoyed many times by that writing on the mirror)


Kevin Reuter said...

Dear Eric,

thanks for responding to my comments again.

A quick rejoinder if I may:

You say that your "Multiple Veridicalities view accepts that the experience in some important sense stays the same". You also say though (in the paper) that the mind gets used to and adapts to the toovy way. But then whatever it is that changes seems to be a conceptual matter and not experiential.

So, how can we experimentally test the multiple veridicalities view? What about following experiment: The mirror that is normally convex we change to a flat mirror, and vice versa. If drivers do not get confused, and do not misjudge the distance, it would seem that you are right in holding that things can look the same distance although the corresponding experiences can be different. If, however, people misjudge distances, it seems more difficult for you to deny that inferential reasoning plays an important part in determining the distance.

Best, Kevin

David Duffy said...

I was reminded of the Gibsonian "direct perception" versus the computationalist "indirect perception" model. The former seem to be discussed a bit in the philosophical literature. I guess the DP/ecological psychologists would say that the distorted mirror image has been completely integrated into your auditory and peripheral visual dynamic model of cars in space around you. You extract the affordances (here dangerousness to you based on position, size, speed and distance) from how the image changes.

Eric Schwitzgebel said...

Kevin, I agree that those results would put pressure against the view I favor, if people couldn't be trained to get used to the flipping, though at first it might throw one for a loop. Partly too, I think, the evidence has to be introspective -- which introduces serious methodological problems, in my view!

Eric Schwitzgebel said...

David -- Yes, I agree with that, though I also don't think the general Gibsonian perspective delivers an unequivocal result on this issue. Gibson does in fact briefly discuss the bent oar case and adaptation to inverting lens, but not committing explicitly, consistently, and self-consciously on the teavy/toovy issue, as far as I can tell from my reading so far. (I am open to correction on that point!)

Larry Franz said...

Hi Eric. Thanks for posting your paper.

I think that the image in the flat mirror is more veridical since it more closely resembles how the car behind us will look if we turn our head and look at it directly (which would seem to be the standard of visual veridicality, if anything is). I've had the experience of looking at a car in the convex mirror and then turning my head and being surprised at how close the car actually is (although no accidents so far). I don't remember ever having that feeling when looking at the flat mirror and then turning my head.

On the other hand, maybe I'm just more accustomed to flat mirrors, as you say. But if flat mirrors and convex mirrors are equally veridical, how about the kind of fun house mirror that produces a wavy image? Are those mirrors just as veridical as the other kinds? We can look at ourselves in a fun house mirror and take into account that the mirror is creating waviness where there isn't any, but it still seems to me that a wavy fun house mirror is significantly less veridical than direct observation. If looking at an object without a mirror is the standard of veridicality, the flat mirror is next in line (it does reverse your image). Then comes the convex mirror, followed by the wavy mirror.

At any rate, I can't think of a good reason for saying that convex mirrors and flat mirrors are equally veridical, but fun house mirrors aren't. Direct observation and the three kinds of mirrors all exhibit some veridicality, but it doesn't seem that they provide the same amount.

David Duffy said...

Eric Schwitzgebel said...

Larry: Yes, that's a good point about flat vs. convex, but I don't know how far it really takes us. By similar reasoning, might one say that the car *really* looks like it's up in the air in front of one?

The carnival mirror case is interesting too. I think what I want to say is that it also can be experienced as yielding veridical appearances, once one becomes accustomed enough and skilled enough. I agree there's no difference of principle here.

Eric Schwitzgebel said...

David Duffy: Thanks for the links! Those articles look to be both interesting and relevant.