[photo-3d] Re: Visual Cliff experiments

[Bruce Springsteen <bsspringsteen@xxxxxxxxx>]

Jim Crowell wrote:

> These experiments (usually called the Visual Cliff experiments) were
> originally done by the Gibsons at Cornell (by Eleanor Gibson, I think).
> They were particularly interested in motion- and texture-related cues
for
> surface layout, which they considered to have been neglected in the
study of
> vision.  So they designed the apparatus to emphasize those cues,
presenting
> the babies with a large dropoff (so there'd be a lot of motion parallax)
and
> putting checks of equal size on the two levels, so the more distant ones
> would project to a much smaller retinal size.
> 
> In other words, they deliberately designed the thing not to answer the
> question we're interested in... :-)

Aha!  So how *do* we answer the question we're interested in?  Abram has
indicated that there are plenty of good ways to verify stereopsis in
infants - what are they I wonder?

I'm looking at Bela Julesz' monograph "Foundations of Cyclopean
Perception."  In the fifth chapter, he states:

"Stereopsis is innate.  Hess (1956) showed that newly hatched chicks
wearing distorting prisms would peck too close and might starve in the
midst of plenty, without relearning.  Bower (1968) showing random-dot
stereograms to six-day-old human infants could elicit a stable eye
fixation if a cyclopean bar appeared in depth; otherwise the eye movements
ocurred at random.  This is most remarkable since the eyes of the babies
were covered with silver nitrate (required by law) prior to these
experiments (and were cleansed by the experimenter), and the first
encounter with patterned stimulation resulted in stereopsis.  Only about a
quarter of the babies exhibited this behavior, and Bower assumes that the
rest had problems with monocular vision."

I assume that there are other tests these days, but the random-dot
stereogram is a handy thing for weeding out non-stereo cues.  In the case
of an adult subject who has never previously experienced stereopsis,
verbal responses might also be hard to verify unless other cues were
eliminated - otherwise how are we certain that what they are seeing is
"stereo", to report it to us?  Devising experiments for our friends,
relatives and babies may be trickier than we can easily do, but thinking
about it is good excercise for the would-be experimenter, at any rate.

Bruce

__________________________________________________
Do You Yahoo!?
Get Yahoo! Mail - Free email you can access from anywhere!
http://mail.yahoo.com/