[Martin Taylor 20000214 09:42]
[From Rick Marken (2000.02.13.1500)]
Happy Valentine's Day, Rick. Love you, too...
Me:
> The retina is not an encoding mechanism
Martin Taylor (20000212 13:38)
> Sorry, but I can't see a distinction between what I said
> and what you say, other than the wording. Here's what I
> said just before the bit you quote:
>
> > What gets transmitted up the optic nerve is a function of
> > the spatial and temporal patterns of light hitting the retina.
> > That function presumably favours those relationships of time
> > and space that usually help us control what we need to control
> > in order to survive, and inhibits temporal and spatial patterns
> > that rarely occur or that don't matter.
>
> How is this different from your:
>
> > The retina is a perceptual mechanism, producing time varying
> > scalar outputs as a _function_ of retinal array inputs.Does Bill's post [Bill Powers (2000.02.13.0216 mst)] help? If not,
I'll give it a shot. I guess my basic complaint is that I don't
see _encoding_ as being anything like _perceiving_; encoding
is data reduction (necessary for data transport when there are
bandwidth limitations); perceiving is data representation
(necessary for control of any particular representation of the
data).
As I understand the terms, encoding is a process that produces a data
representation. A "perceptual signal" is such a representation. What
"perceiving" is depends on the context, but, being a verb, it must be
a process. One of the meanings of "perceiving" is "to bring to
consciousness," but that's not understood within PCT. Within PCT,
"perceiving" usually seems to mean "generate the value of a
perceptual signal," or in other words, it is a synonym for the
"encoding" done by a perceptual input function.
[From Bruce Nevin (2000.02.13.2037)]
"Encoding" is a metaphor that implies discrete bits of information.
Is that your point (Rick and Bill?). I don't agree with Bruce, but I
can imagine that if you do understand "encoding" in this way, that
would account for our disagreement. I understand "encoding" to mean
the result of performing a function on several inputs, analogue or
digital. But Bill seems to have a different point...
[From Bill Powers (2000.02.14.0318 MST)]
Martin Taylor 20000212 13:38--
I thought Rick's point was quite valid. If the function of the retina were
merely to compress data, then we would assume that in the midbrain
we would have decompression to restore the original resolution.
Think this through a bit further. Why does the computer need to
decompress the picture data? There are two possible reasons, both
implying that the computer is unconcerned with the content of the
picture. One reason is that it must redisplay the picture so that it
is intelligible to the human eye. Usually, and most simply, this
means reproducing the original image. The other reason is that it
must perform some manipulation (such as your sharpening procedure) on
the pixel array. The brain usually performs neither of these
functions, but is concerned with the _meaning_ of the picture, which
implies no requirement for inverting the compression function. Rick
put it very well (2000.02.13.1330 PST) when he said said:
Held and Hein's results show that kittens can't
learn to control visual perceptions by just passively watching
those perceptions go by. The active kittens learn to control
visual objects (they avoid visual obstacles, move to a visual
destination, etc) by _trying_ to control those objects -- ie.
by actively producing outputs that do or don't have an effect on
the seen objects. The passive kittens can't learn to control
because they are prevented from producing outputs (via the
harnessing) that would have an effect on (and, thus, allow then
to learn to control) their visual perceptions.
and
PCT says that perception basically _is_ a "feature extraction"
process; the features that are "extracted" are aspects of the
"intensity" signals present at the sensory surface: features like
sensations, configuration, transitions, sequences, etc.
Both beautifully put. That's what I tried to say, but Rick's wording
is much better, as he describes one aspect of the data compression
process rather than simply pointing out that compression occurs, as I
did.
Finally, the function applied in the retina is a function of both
time and space. Bill (2000.01.14.0333 MST) rather oversimplifies in
his response to me. I don't know enough to be sure whether the _only_
encoding in the central retina is to improve resolution while keeping
a one-to-one mapping between pixel intensity and optic nerve fibre
impulse rate. I believe that to be incorrect, and it would certainly
leave the visual system very vulnerable to minor damage if it were
true. Such vulnerability is exceedingly rare in the design of
biological systems, so even if I had never studied the matter I would
tend to doubt that it occurs in the visual system. But since Bill may
be right, and I may be right, I'll leave it for people more up to
date in vision research to comment. But...
>if we can ever perceive the intensity of light (which I seriously
>doubt), we can do it _only_ by generating some kind of a function
>that codes the patterns of receptor outputs over time and space.You're overcomplicating the situation, in my opinion. Yes, there is a time
constant (about 0.1 sec at daylight levels of illumination) in the response
of the retina to light.
You deal only with momentary local fluctuations at a constant overall
level. And we can clearly perceive local differences from the local
average level. We are much worse at detecting difference over large
angles within a scene, and almost incompenetent at detecting
differences that occur over longish periods of time. We are very good
at _not_ seeing changes in intensity that are caused by changes in
illumination that occur slowly in space or time. We see what matters
for control, and unless the variation is great, the level of
illumination (and thus of light intensity at the retina) rarely
matters.
Martin