Tuesday, November 27, 2012

A Story in Which I Acquire Fisheyes, or The Paranoid Jeweler and the Sphere-Eye God

I become paranoid and quit philosophy. But I need to make money somehow, so I become a jeweler. The problem with being a paranoid jeweler is this: Since I am doing close work, I must focus closely through a magnifying lens, which drastically reduces my ability to keep a constant wary eye on my surroundings. Fortunately, being the clever sort, I hit upon a solution: fisheye lenses!

Above each ear, I mount a fisheye lens, each with a 180-degree field of view, together drawing light from the entire 360 degrees of my environment. Mirrors redirect the light to a screen mounted eight inches in front of my eyes, entirely occluding my normal forward view. Reflections from the right-mounted lens fill almost the whole space before my right eye. Reflections from the left-mounted lens fill almost the whole space before my left eye. Thus, binocularly, I simultaneously see my entire 360-degree environment. No one can sneak up on me now! By shifting attention between my right and left eyes and by shifting my gaze direction, I can focus on different objects in my visual field, thus doing my jeweler's tasks. My best view is of objects near my ears, which in fact are magnified relative to what I could see unaided, while objects near midline, directly in front of me, behind me, or above or below me, are radically shrunk and curved. This is no big problem, though, because I can rotate my head to put whatever I want into the privileged field of view. I quickly acquire the habit of walking with my head turned ninety-degrees sideways and with my ear slightly angled toward the ground.

At first, of course, everything looks radically distorted, given what I am used to. Here's how things look on the right side of my visual field when I turn my right ear toward the sky:

(image from http://www.sandydan.com/photo/wide/fish/ftest5.jpg)

After a while, though, I get used to it -- just like people get used to convex rearview mirrors in cars and just like, given long enough, people adapt to lenses that completely invert the visual field. I come to expect that a gemstone of constant size will look like [this] when held before my forehead and like [that] when held before my ear. I learn to play catch, to ski, to drive a car (no need for rearview mirrors!). After long enough, things look right this way. I would be utterly stymied and physically incompetent without my fisheye lenses.

After long enough, do things go back to looking like they did before I put on the fisheye lenses, the way some people say (but not others!) that after adapting to visual-field inverting lenses things eventually go back to looking the way they did before having donned the lenses? That doesn't seem possible. Unlike in the inverting-lenses case, there doesn't seem to be sufficient symmetry to enable such an adaptation back. I see much better near my ear than in front of my forehead. I see a full 360 degrees. Things might come to look just like I expect, but they can't come look the same as before.

Do things nonetheless look illusory, though now I am used to the illusion? I'm inclined to say no. With full adaptation they will come to seem right, and not illusory -- just as it seems right and not illusory that the angle a person occupies in my visual field grows smaller as he walks away from me, just like it seems right and not illusory that the car behind me appears in my central rear view mirror in my forward gaze as I am driving (and not like a small car somehow elevated in the air in front of me, which I only intellectually know is really behind me). There's nothing intrinsically privileged, I'm inclined to think, about the particular camera obscura optics of the human eye, about our particular form of refraction and focus on an interior retinal hemisphere. Another species might be born with fisheyes, and talk and fight and do physics with us -- do it better than us, and think we are the ones with the weird set-up. A god might have a giant spherical eye gathering light from its interior, in which mortals dwell, with objects occupying more visual angle as they approach the surface of the sphere that bounds its indwellers' world. There is no objectively right visual optics.

From this, I think it might follow that our visual experience of the world as like [this] (and here I inwardly gesture at the way in which my office presents itself to me visually right now, the way my hand before my eyes looks to me right now), is no more the one right veridical visual phenomenology of shape and size and distance than that of my fisheye future self or that of the sphere-eye god.

25 comments:

Zach Barnett said...

As usual, I agree. Indeed, if there were a group of beings who grew up seeing this way (360 degrees), and there were a device that had precisely the opposite effect (enables them to "see" the way that you and I do), then probably our way of seeing would seem distorted and seriously deficient to them.

I'd add that this all seems to me rather straightforward (both your post and my corollary), but in your defense, I have worked with good, respectable philosophers of mind who seem to be tempted by views incompatible with this sort of 'perceptual egalitarianism.' Maybe it's not so simple as it seems to me. Maybe I'm deficient.

clasqm said...

Presumably there's a reason they are called "FISH-eye lenses". Perhaps if you were a fish. this would be pretty much how you would see the world.

And if you were a sufficiently articulate arthropod, you would be able to tell us what it was like to observe the world through hundreds of crystalline lenses in your large compound eyes.

If you were an intelligent bat, you could tell us how to observe the world using reflected sound waves. ditto for cetaceans. Vision and its analogues have evolved independently around half a dozen times on this planet if I recall correctly.

I'm sorry, Eric. I'm sure there's a point in here somewhere, but I seem to be missing it.

Eric Schwitzgebel said...

@ Zach and clasqm: I agree with Zach's corollary. I don't mind if it seems pretty obvious. That will help it serve well as a first step down a garden path that will, I think, create problems for what most philosophers and theoretical perceptual psychologists say about half-submerged oars and Stratton's inverting lenses. Maybe even farther down the path will be something like a secondary-quality or transcendental idealist view of spatial experience. I'm not sure I do want to go all the way down either of those paths myself, though....

Tad said...

This post made me think of this upcoming conference:

http://www.gla.ac.uk/schools/humanities/research/philosophyresearch/cspe/events/sensorysubstitutionandaugmentationconference/

Scott Bakker said...

I’ve thought about this very topic off and on over the years, but more in terms of what 360 degree vision means for the MARGINS of our visual field. The photo you provide actually illustrates a strange, and I would argue, profound complication of your thought experiment: the 360 degree view can only be pictorially represented *against* some kind of background, in this case, a blacked-out square situated on your blog page. To visually represent 360 degree vision you have to shrink it to fit within a couples of degrees.

Actual as opposed to simulated 360 degree vision would arguably have no margin whatsoever, no point where the visual field just ‘runs out.’

One of the fascinating things about target-tracking ‘predatory’ vision systems like our own is the way we neglect the margins of our visual field, how we have so little sense that we’re looking at the world throw a retinal straw, simply because of the way, presumably, our visual systems are integrated with what might be called a *cognitive* fish-eye lens–where the margins are occluded altogether.

A couple of not so obvious things come out of this, I think. In a sense, true 360 degree vision, target-detecting ‘prey’ (paranoid) vision systems simply speaks to brute sensory access, and not ‘vision’ in any phenomenological sense. The cognitive fish-eye is what’s important. The lens just speaks to the information delivery mechanism. Where eyes-forward target-tracking systems use saccades, movements of the head, etc., eyes-around target-detecting systems would presumably rely on exogenous *attention* mechanisms to isolate and assess threats. Given evolutionary thrift, they would likely be 360 degree ‘motion detectors,’ which when triggered alert their cognitive systems, which then generate ‘vision.’ In a sense, you could say there really is *no such thing* as ‘360 degree’ vision, just a variety of biological (or in the case of your thought experiment) strategies pertaining to the efficient delivery of visual environmental information. And this suggests that there’s something ‘honest’ about eyes-forward target-tracking systems like our own: we at least, aren’t gulled by the absence of margins into thinking we can see our own backs!

Grobstein said...

Presumably "illusion" in this context means something like, perceptual features that are susceptible to deceive us in our naive, immediate interpretation of how they relate to the real world. Once you adapt, your immediate interpretation of your fish-eye percepts will be generally accurate.

I agree with the comments above that I don't really see the other side of this controversy, if there is one.

Relatedly, an interesting discussion of whether the upside-down visual images projected on our retinas are "really" inverted in the brain: link.

Eric Schwitzgebel said...

@ Tad: That looks like a terrific conference. Fiona Macpherson has a good eye for interesting issues in this area!

Eric Schwitzgebel said...

@ Scott: Your point about attention shift rather than foveal shift seems right to me. And yes, in order for the 360-degree vision to come to *me* it has to be presented in a very unnatural way -- in two circles in my forward vision. A more natural 360-degree seer presumably wouldn't have to do this. I'm not sure, though, about your resistance to calling 360-degree sensory access to visual information "vision". It will be importantly phenomenally different, surely -- which is my point -- but is there reason to think it will be *so* phenomenally different as to not merit the term "vision"? And if so, is this a substantive issue of some sort or just a vocabulary issue?

Eric Schwitzgebel said...

@ Grobstein: Descartes is good on the retinal inversion issue, too! (In "The World".)

As for the other side of the issue: Are you ready to say that to a sufficiently skilled rower, the oar half-submerged in water in no sense looks bent? To the person sufficiently familiar with the Muller-Lyer illusion, the lines no longer look different sizes? If so, you're in a minority. If not, how would you distinguish those cases from the present one? In a nutshell, that's the issue -- though I admit that I don't really do a good job taking it home in the post.

Zach Barnett said...

Question: In the checker shadow illusion, tile A seems to be darker than tile B. In fact they are the same color. I am familiar with this illusion, and so I believe that the two tiles are the same color. I would bet loads of money on it. But the effect remains? Why?

A perceptual system in my brain continues to be tricked by the stimulus, even though *I'm* not. It continues to issue the judgment that A is darker than B. But I know better than to endorse my own perceptual system's mistaken judgment.

The same is true in the case of a partly submerged oar. The rower knows that the oar is not bent, but the relevant system of her brain continues to issue the judgment that it is. My guess is that this would continue to be true, even for an experienced rower. But in principle, it's possible that her perceptual system would "learn" to better unify inputs from above and below the water such that it would no longer issue the mistaken judgment that the oar was bent. If this happened, the oar would in no sense look bent to the rower.

Eric Schwitzgebel said...

@ Zach: So should we take that same perspective back to the fisheye case? Is there still a sense in which the objects near midline look curved, even though I intellectually know better?

Arnold Trehub said...

Eric: "Do things nonetheless look illusory, though now I am used to the illusion?"

I wonder about your use of the term "illusion". A visual illusion is a perceptual experience that does not conform with the "true" stimulus as justified by independent analysis. In the fisheye case, artificial lenses intervene between the visual world and your retinas, providing a true stimulus that presents the visual world in a way that is a distortion from your natural visual experience of the world around you. And you experience it as a distortion. I would guess that your ability to use the mechanisms of the post-retinal brain system that constitutes your perceptual experience (retinoid space?) to normalize your experience would be much more difficult in the fisheye case than in the case of the inverting glasses because of the much more complex egocentric spatiotopic adjustments that would be needed. However, if you could actually make the necessary spatial compensations in your post-retinal retinoid space, then your phenomenal world would seem different from what you remember, but not distorted. At least that's the way it seems to me.

Scott Bakker said...

"It will be importantly phenomenally different, surely -- which is my point -- but is there reason to think it will be *so* phenomenally different as to not merit the term "vision"? And if so, is this a substantive issue of some sort or just a vocabulary issue?"

You caught me riffing! What I wanted to push was how the two alternatives bear on the ability to cognize environmental ignorance. You would have a hard time convincing a sentient fish-eye creature there’s someone in the room they can’t see, simply because they would assume that they see everything there is to be seen. Not so much with a sentient monkey-eye creature. What makes inattentional blindness, for instance, so shocking is the assumption of sight. If you asked participants of the gorilla experiment to watch the basketball being passed around through a straw, the gorilla would be a simple matter of occlusion rather than blindness. It would be the same as asking them whether they saw the clown walking behind them. It would go without saying.

With the sentient fish-eye creature, the scope of possible ‘inattentional blindnesses’ is drastically increased, as is the scope of possibilities for attentional scrutiny. The thing is, the monkey-eyed creature possesses mechanical means to target everything the fish-eyed creature can without the attendant vulnerability to inattentional blindness. So in a sense, you could say the monkey-eyed creature would likely be easier to ambush, but more difficult to deceive (because the possibility of ambush would be something it knows). The fish-eyed creature, on the other hand, would be more difficult to ambush and yet easier to deceive (because it thinks it sees everything there is to be seen). This counts as a ‘veridical difference,’ doesn’t it?

Eric Schwitzgebel said...

@ Arnold: You write: "However, if you could actually make the necessary spatial compensations in your post-retinal retinoid space, then your phenomenal world would seem different from what you remember, but not distorted."

That is exactly the conclusion I am aiming for. Where to go with it after that is of course another question, but I'm thinking of it as a step down a garden path toward the half-submerged oar in no sense looking bent to the sufficiently skilled and adapted rower.

Eric Schwitzgebel said...

@ Scott: Yes, now I understand your thought better! That all seems right to me.

Jorge A. said...

As a philosopher of mind, you should be interested in this:

http://www.sciencemag.org/content/338/6111/1202

The tasks this virtual mind can perform can be found here:
http://nengo.ca/build-a-brain/spauntasks

It includes pattern recognition, which is nothing short of astounding.

Kester said...

I'm not sure how far the visual field on its own can support illusion – for that we need to be using it in perception (which requires cooperation with our senses of position, movement and balance, and our visual imaginations and visual memories, etc., etc.). For instance, using another of your themes, if you have your eyes closed in order to study the patterns produced, and suddenly something happens in your brain to invert left and right in your visual field, then the patterns will be distorted in some way, but there's no question of any illusion occurring because you aren't using the field for perception at that time. Insofar as your experience of the world like [this] or like [that] can be illusory/veridical, [this] and [that] are not states of your visual field (only). This is hardly in conflict with your position that no one type of visual field is privileged, of course. I think I don't disagree with that position, but could you maybe give an example of a philosopher privileging ordinary visual experience in a way you oppose?

Kester said...

You ask in comments above “how would you distinguish those cases [like the oar in water] from the present one?”. My suggestion would be that there is a difference between what you might call global and local illusions. We can change e.g. which shapes in the visual field we use to recognise straight lines, which would be part of the process of adapting to the fish-eye set-up and, in doing so, dissolving a global illusion (that all straight things are bent). But no such change will enable our faculties to distinguish bent and partly-submerged oars in all situations - their indistinguishability is a case of merely local illusion. In the global case, we aren't using our visual field right in relation to e.g. our other senses, though that can change; in the local case, our visual field can't provide enough information to distinguish the two things, so we'll be vulnerable to illusion however we use it.

For instance, when you say “it seems right and not illusory that the angle a person occupies in my visual field grows smaller as he walks away from me“, that rightness is a case of absence of global illusion, resulting from the way you're using your visual field (the visual field alone can't be right or illusory). Even when you use it right, though, that fact about how distance alters appearance allows for local illusion - two very different things can look the same (see: youtube.com/watch?v=vh5kZ4uIUC0). It is the structure of the visual field that allows for this local illusion, in that some information has to be lost in translating differences in position in 3D space into differences in the visual field, so merely using it right doesn't help you. You might say that, in the global case, getting used to the illusion dissolves it, because there getting used to it amounts to coming to use the visual field right, which is a change in perception; but, in the local case, getting used to it only involves being able to rely on extra-perceptual self-correction, which doesn't dissolve the illusion.

Arnold Trehub said...

Jorge" "The tasks this virtual mind can perform can be found here:
http://nengo.ca/build-a-brain/spauntasks

It includes pattern recognition, which is nothing short of astounding."

The tasks that this virtual mind performs are surely not astounding. The stimuli are presented in isolation and in a normalized position with respect to the camera. A cognitive brain has to parse stimuli out of a clutter of objects in a visual scene. The cognitive neuronal mechanisms detailed in *The Cognitive Brain* (TCB) (MIT Press 1991) are able to accomplish this essential task in order to learn and recognize objects. See TCB, Ch. 12, "Self Directed Learning in Complex Environment", here:

http://www.people.umass.edu/trehub/thecognitivebrain/chapter12.pdf

also

http://people.umass.edu/trehub/sparscodtre.pdf

Arnold Trehub said...

Eric, here is a true illusion that might interest you. No special goggles intervene between the stimulus on the computer screen and your different perceptions of the rotated table, a visual object having a constant physical length and width over rotation:

http://people.umass.edu/trehub/Rotated%20table-2.pdf

Eric Schwitzgebel said...

Thanks for the continuing comments, folks!

@ Arnold: Yes, that is a very cool illusion. I've seen versions of it before, but not with the background lines. It seems especially strong to me in your presentation.

@ Jorge: It will be interesting to see where this kind of thing is in 15 years!

Eric Schwitzgebel said...

@ Kester: Another interesting case is bifocals. I'm not sure if this would be "global" or "local" in your way of thinking, and I'm not sure what it's really like to wear them daily. When an object crosses the field from one lens curvature to another, is that psychologically similar to the bent-oar case in the relevant way?

Another thought experiment: We live in an aquatic environment, with our heads always 2 feet above the water and unable to raise or lower our chins. Is the refraction at the waterline now global?

Arnold Trehub said...

@Eric: The interesting thing is that this illusion (like others in 2D perspective drawings) is predicted by the neuronal structure and dynamics of the neuronal mechanisms in the retinoid model. These brain mechanisms expand perceived vertical dimensions relative to perceived horizontal dimensions in perspective drawings depicting 3D objects. See "Why 2D Perspective Works", pp. 10-13, here:

http://theassc.org/documents/where_am_i_redux

Anonymous said...

http://people.umass.edu/trehub/Rotated%20table-2.pdf

so is it that the first table is on a stand of some sort such that its "far" (in a horizontal sense) side is no further from you than it's close side?

Eric Schwitzgebel said...

Anon Dec 26: I would have said that the natural interpretation is the opposite of that. I'm not sure I get your point.