Two-Dimensional Sensory Input

CSGnet Archive CSG2002
1

I was participating on another list when someone made mention of ecological
psychology. That led me to some web pages, from one of which I extracted
the following:

<snip>

"Gibson, James Jerome

"In his last book, The Ecological Approach to Visual Perception, James
Gibson (1904-1979) concluded with a plea that the terms and concepts of his
theory "...never shackle thought as the old terms and concepts have!" He
was referring to the framework of traditional perception, as was reflected,
for example, in the classical problem of space perception Bishop Berkeley
posed more than three hundred years ago (Berkeley 1963). How is it possible
to perceive three-dimensional space when the input to our senses is a
two-dimensional retinal surface in the case of vision, or a skin surface in
the case of touch? Logically, it seemed this inadequate stimulation had to
be supplemented somehow to account for our ordinary perception of a
three-dimensional world. There have been two general proposals for the
nature of this supplementation. An empiricist proposal, advocated by
Berkeley himself, based the supplementation in the prior experience of the
individual. The alternative nativist proposal based the supplementation in
the innate functioning of the mental apparatus which intrinsically imposes
a three-dimensional structure on two-dimensional stimulation. These two
alternatives in only slightly modified forms persist to this day."

<end snip>

I'm probably way out of my league here but it seems to me that the
preceding paragraph doesn't stand up to the slightest scrutiny. It doesn't
seem to me that the retinal surface is two dimensional; it has depth or
thickness. Nor is the skin a two-dimensional surface. In fact (and I know
I use that word loosely), it seems to me that there are no true
two-dimensional surfaces; all surfaces have a third dimension, even the
archetypical sheet of paper. Or, in different terms, a true
two-dimensional surface is really just an abstraction made for the purposes
of analytical thinking or mathematicaly reasoning but doesn't really have
any counterpart in physical reality (or whatever it is of so-called
physical reality that we can perceive).

Anyway, I thought the members of this list might have some thoughts about
the basic question posed above; namely, " How is it possible to perceive
three-dimensional space when the input to our senses is a two-dimensional
retinal surface in the case of vision, or a skin surface in the case of touch?"

Regards,

Fred Nickols
nickols@safe-t.net

2

[From Bruce Abbott (2002.11.12.16015 EST)]

Fred Nickols (2002.11.12) --

I'm probably way out of my league here but it seems to me that the
preceding paragraph doesn't stand up to the slightest scrutiny. It doesn't
seem to me that the retinal surface is two dimensional; it has depth or
thickness. Nor is the skin a two-dimensional surface. In fact (and I know
I use that word loosely), it seems to me that there are no true
two-dimensional surfaces; all surfaces have a third dimension, even the
archetypical sheet of paper. Or, in different terms, a true
two-dimensional surface is really just an abstraction made for the purposes
of analytical thinking or mathematicaly reasoning but doesn't really have
any counterpart in physical reality (or whatever it is of so-called
physical reality that we can perceive).

The quoted material was referring to the fact that the retina or skin
surface registers only a two-dimensional image of the three-dimensional
world. The fact that the receptive surfaces themselves have depth is
irrelevant.

With respect to vision, information from which depth can be reconstructed
is available from a variety of sources. The most potent of these is
retinal disparity -- the differences between the two-dimensional images
formed on each retina owing to the fact that the two eyes view the scene
before them from somewhat different positions. The closer an object is to
the viewer, the greater is the apparent difference in the apparent position
of the object, relative to a distant background. This disparity is used by
the visual system to reconstruct the distance of the object from the viewer
and is the basis for the "3-D" or pop-out effect one gets while viewing a
scene with both eyes open and which disappears when one eye is closed. It
would be of considerable interest to know how the visual cortex performs
the computations, which are facilitated by the fact that the neural
representations of the left visual fields of both eyes lie next to one
another in the right occipital lobe and similarly for the right visual
field of both eyes in the left occipital lobe. The process that
accomplishes this is innately wired into the brain but requires appropriate
visual input if the wiring is to be properly maintained (i.e., "use it or
loose it"). For this reason it is recommended that children whose eyes
don't align properly (e.g., cross-eyed) have their condition corrected as
soon as practical, as later correction of the physical problem not correct
the problem in depth perception.

Bruce A.

3

[From Fred Nickols (2002.11.12.1632)] --

[From Bruce Abbott (2002.11.12.16015 EST)]

Fred Nickols (2002.11.12) --

I'm probably way out of my league here but it seems to me that the
preceding paragraph doesn't stand up to the slightest scrutiny. It doesn't
seem to me that the retinal surface is two dimensional; it has depth or
thickness. Nor is the skin a two-dimensional surface. In fact (and I know
I use that word loosely), it seems to me that there are no true
two-dimensional surfaces; all surfaces have a third dimension, even the
archetypical sheet of paper. Or, in different terms, a true
two-dimensional surface is really just an abstraction made for the purposes
of analytical thinking or mathematicaly reasoning but doesn't really have
any counterpart in physical reality (or whatever it is of so-called
physical reality that we can perceive).

The quoted material was referring to the fact that the retina or skin
surface registers only a two-dimensional image of the three-dimensional
world. The fact that the receptive surfaces themselves have depth is
irrelevant.

Uh, hmm, gee, well, I dunno. I guess that's the essence of my question. I
think I get the bit about binocular vision but is the image on the retinal
surface truly a two-dimensional image or is there more to it than that?

Fred Nickols
nickols@safe-t.net

4

[From Rick Marken (2002.11.13.0920)]

I'm probably way out of my league here but it seems to me that the
preceding paragraph doesn't stand up to the slightest scrutiny. It doesn't
seem to me that the retinal surface is two dimensional; it has depth or
thickness. Nor is the skin a two-dimensional surface.

The "classical problem of space perception" is really only a "problem" if
perception is conceived of as a process of correctly representing the reality
that is presumed to exist outside of oneself. This problem doesn't exist in PCT
because the role of perception in PCT is not to provide an accurate map of
reality; in PCT the role of perception is to provide a controllable construction
of reality. Indeed, in PCT it is "reality" that is the mystery, not perception.

"Reality", in the form of a physical model of the environment, is one of the
assumptions (guesses) made by the PCT model. The PCT model essentially says "If
reality is assumed to be, say, the projection on a two dimensional retinal surface
of a ball moving in three dimensional space, and if a control system constructs a
perception of that reality in terms of the two dimensional projection of the angle
of the ball relative to a fixed point in three space and if the system moves in
that reality so as to keep that perception constantly increasing then the system
will intercept the ball". In PCT, the question isn't how "veridical"
representations of reality are constructed. In PCT the question is "what is the
nature of the representation of reality that _is_ constructed". What, that is, is
the function that relates what we perceive to what we _assume_ (based on "physics
and chemistry") to be the reality that is the "argument" of that function. We
study this aspect of perception using the test for the controlled perceptual
variable.

Best

Rick

···

--
Richard S. Marken, Ph.D.
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

5

[From Rick Marken (2002.11.13.1130)]

Wolfgang Zocher (2002.11.13.1857 CET)

> [From Rick Marken (2002.11.13.0920)]
>

Thanks for this wonderful and "constructivistish" description of
perception. It's really GREAT!!

Gee, thanks Wolfgang! Great to hear from you. I hope the fact that you are able to post
to CSGNet means that you are feeling better! It would be so great if you could come to
the conference in LA.

Anyway, best regards to you and Marion.

Best

Rick

···

--
Richard S. Marken, Ph.D.
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

6

[From Wolfgang Zocher (2002.11.13.1857 CET)

[From Rick Marken (2002.11.13.0920)]

Thanks for this wonderful and "constructivistish" description of
perception. It's really GREAT!!

Wolfgang

···

The "classical problem of space perception" is really only a "problem" if
perception is conceived of as a process of correctly representing the reality
that is presumed to exist outside of oneself. This problem doesn't exist in PCT
because the role of perception in PCT is not to provide an accurate map of
reality; in PCT the role of perception is to provide a controllable construction
of reality. Indeed, in PCT it is "reality" that is the mystery, not perception.

"Reality", in the form of a physical model of the environment, is one of the
assumptions (guesses) made by the PCT model. The PCT model essentially says "If
reality is assumed to be, say, the projection on a two dimensional retinal surface
of a ball moving in three dimensional space, and if a control system constructs a
perception of that reality in terms of the two dimensional projection of the angle
of the ball relative to a fixed point in three space and if the system moves in
that reality so as to keep that perception constantly increasing then the system
will intercept the ball". In PCT, the question isn't how "veridical"
representations of reality are constructed. In PCT the question is "what is the
nature of the representation of reality that _is_ constructed". What, that is, is
the function that relates what we perceive to what we _assume_ (based on "physics
and chemistry") to be the reality that is the "argument" of that function. We
study this aspect of perception using the test for the controlled perceptual
variable.

Best

Rick
--
Richard S. Marken, Ph.D.
The RAND Corporation
PO Box 2138
1700 Main Street
Santa Monica, CA 90407-2138
Tel: 310-393-0411 x7971
Fax: 310-451-7018
E-mail: rmarken@rand.org

--
Wolfgang Zocher

-------------------------------------------------------------------
www: http://wzocher.bei.t-online.de/
-------------------------------------------------------------------

7

Fred, the rotating trapezoid illusion provides some interesting experiences
in this regard
David Wolsk.

···

----- Original Message -----
From: "Fred Nickols" <nickols@SAFE-T.NET>
To: <CSGNET@listserv.uiuc.edu>
Sent: Tuesday, November 12, 2002 12:24 PM
Subject: Two-Dimensional Sensory Input

Anyway, I thought the members of this list might have some thoughts about
the basic question posed above; namely, " How is it possible to perceive
three-dimensional space when the input to our senses is a two-dimensional
retinal surface in the case of vision, or a skin surface in the case of

touch?"

Regards,

Fred Nickols
nickols@safe-t.net

8

[From Bill Powers (2002.11.15.0828 MST)]

Fred Nickols" <nickols@SAFE-T.NET> --

> Anyway, I thought the members of this list might have some thoughts about

> the basic question posed above; namely, " How is it possible to perceive
> three-dimensional space when the input to our senses is a two-dimensional
> retinal surface in the case of vision, or a skin surface in the case of
touch?"

Has anyone given a direct answer to this question? Maybe Bruce Abbott did.
Anyway, the answer is that the eyes provide two images from (slightly)
different points of view, and this is enough to permit judging depth from
differences in the relative positions of objects in the two eyes. This
works up to 20 feet or so, depending on the sharpness of vision.

It's also possible to judge depth using only one eye if the point of view
is shifted laterally (at right angles to the line of sight). Doing this
shows the relative positions of objects changing as the point of view is
moved. As noted in B:CP, it is even possible to see depth when one image is
presented to one eye on one day, and the shifted image is presented to the
other eye a day later -- at least one woman was able to do this! The test
was to pick out an elevated patch in a two fields of random dots normally
viewed binocularly. This woman was obviously able to store images at a very
low order of perception, perhaps second order, and compare a remembered
image with a real one.

In the Little Man Version Two, the simulation perceives the radial
distances of the hand and of the target by computing the disparity of
retinal positions between the two eyes. Optical computations find the
points on the retina where the fingertip and the target will appear in each
eye, relative to the center of the retina. A simple perceptual computation
then produces a nonlinear and approximate measure of distance from the
image disparity. Reaching out to touch the target is controlled by making
the fingertip image disparity the same as the target image disparity, so
the errors in computing actual physical distance don't matter -- the same
errors and nonlinearities occur in computing both distances.

Actually, the eyes also individually move to center the target image, so
the target image disparity is reduced to zero. But the convergence of the
eyes needed to do that directly adds to or subtracts from the disparity
between images of the fingertip, so the depth control system still works
the same way, by making the fingertip image disparity the same as the
target image disparity -- in this case, zero.

I don't know how the brain actually computes distance from the two sets of
image disparities, but the Little Man simulations clearly shows that this
is possible to do by at least one means.

Best,

Bill P.

.