Re: Learning to see in 3D

[P3D Jim Crowell <crowell@xxxxxxxxxxxxxxx>]

At 9:25 AM -0600 1/15/97, P3D John W Roberts wrote:
>
>I agree that all of this is likely to be beneficial, but there's a
>physiological factor that comes into play at a very early age, probably
>too young for Viewmasters. It has been reported in the past on this list
>that vision researchers state that there's a certain critical age at which
>small children either develop the inherent ability to see in 3D, or they
>don't.

Very young, about 5 months if I remember right.

>
>I came across an interesting article in Scientific American several years
>ago, which may shed some light on how this could be. The article was about
>the vision systems in young cats. Cats have visual centers for processing
>the incoming left and right views, but when they are very young, the visual
>signals from both eyes go to both processing centers. (The "left view"
>processing center receives input from both the left and right eyes, and
>so does the "right view" processing center.) As a cat gets older, and
>providing that the cat has full use of both eyes, the cross-connections
>are gradually shut down, so each processing center pretty much receives
>signals only from the corresponding eye. Above a certain age, the ability
>to shut down these cross-channels appears to go away.
>

Actually, this isn't quite right.  Initially, a given neuron takes input
from a range of positions in the two eyes.  During normal development,
those connections are refined so that the neuron only responds well to a
fairly narrow range of disparities (differences in horizontal positions of
a feature) between the two eyes' images; this is the mechanism that makes
stereo possible.  This refinement requires experience with a normal range
of disparities; if the animal or person has amblyopia or strabismus
(crooked eye) or some other condition that prevents this experience, then
each neuron tends to become dominated by inputs only from one eye or the
other & stereo doesn't develop.

>
>As to why the the brain should start with the cross-connections and then
>rewire them as the visual system is used, perhaps there's an evolutionary
>advantage - if something's wrong with one eye or part of the brain, it
>provides the best chance to provide the best possible visual processing
>with what's available.
>

I think it's more a problem that the degree of specificity required is too
great to be coded in the DNA.  That's a very common sort of story in
development; the DNA sets up some initial conditions & a certain "normal"
set of conditions & experiences in early life is required for fine-tuning.

-Jim C.

----------------------
Jim Crowell
Division of Biology
216-76
Caltech
Pasadena, CA
(818) 395-8337
jim@xxxxxxxxxxxxxxx



------------------------------