We take it for granted that all cultures see as we see. Deregowski questions this assumption. The conventions used in western art are not recognised world-wide. Robert Laws, was a Scottish missionary working in Malawi at the end of the 19th century. He reported that natives could not easily interpret black and white photographs. They recognise parts that make up the whole, and with effort are able to combine the parts in order to see the whole picture. Mrs Donald Fraser who taught health care in Africa in the 1920's reported that the natives could not appreciate that profiles were an accurate representation of people. The natives would point out that as the picture showed only one eye, it could not be a picture of a person. Another report suggested that natives could understand 2D pictures easily enough. In this case, tribespeople were frightened when a life-size picture of an elephant was projected onto a sheet. These unsystematic reports are hardly scientific evidence upon which to draw conclusions. William Hudson, a researcher, used a much better controlled technique. He presented pictures that had the usual western depth cues.
Hudson would ask questions like "What is closer to the man?", or "What is the man doing?". In the above described picture, the hunter appears to western eyes (brains!) as though he is about to spear the antelope, but many native Africans would say he was about to spear the elephant. This was presumably because the elephant was viewed as being on the same plane as the antelope, and thus in the line of fire. To make sure this mistake was not owing to a more basic problem; a control picture was also constructed. This depicted the objects as being on the same plane (i.e. the same distance from the observer), and all drawn in proportion to each other. This picture did not depend on western depth cues, for interpretation. Natives interpreted this picture correctly. Another control in this experiment was for the experimenter to ask the subject to name each object in the picture. This was just to make sure that any problems were not owing to the poor recognition of the objects. The sticks and modelling clay experiment.
In another test, Zambian primary school children and unskilled
workers were given sticks and modelling clay. The clay could hold
the sticks together. The subjects were asked to build a model as
depicted in a 2D drawing. 
Westerners would have had no difficulty in
interpreting the drawing as that of a cube, however, many Zambian
subjects produced a 2D model that matched the drawing, but was
not 3D! The subjects were also given Hudson's test (described
above). It was found that if the a subject gave incorrect answers
in the Hudson test (a two-dimensional perceiver) he was likely to
build a 2D model. Likewise, a subject who correctly answered
Hudson's test (a three-dimensional perceiver) would build a cube.
Two-dimensional perceivers preferred a drawing of two squares
joined on the same plane by a single line, rather than a western
drawing of a cube, when asked to chose the drawing that best
represented a model cube. 
A drawing of an impossible object (an ambiguous trident) was
given to Zambian children to copy.
The amount of time the children could
see the drawing was controlled and recorded, by the lifting of a
flap that hid the drawing. The children were not allowed to draw
when the flap was open. After the flap was closed the children
had to wait another 10 seconds before continuing. The children
were also given the Hudson test. It was found that two-dimensional
perceivers spent less time looking at the drawing then three-dimensional
perceivers. Most westerners (three-dimensional perceivers) find
it difficult to copy the ambiguous trident, owing to our
knowledge of real objects interfering with what is really
depicted. The Zambian two-dimensional perceivers presumably do
not suffer from this interference effect and are therefore able
to copy the drawing far more easily.
Richard L. Gregory of Bristol University, has designed an apparatus that can allow an objective read-out of how far away an object depicted in a picture appears to be. This ingenious device has a Polaroid filter that allows the picture to be seen by only one eye, whereas an adjustable spot of light can be seen three-dimensionally with both eyes. Hudson's pictures were displayed, using this apparatus, to African unskilled workers. The subjects were asked to adjust the light so that it appeared at the same distance as objects depicted in Hudson's pictures. Three-dimensional perceivers adjust the light in accordance with the perceived depth of objects in the pictures. This gives a variety of readings for one picture. Two-dimensional viewers (many of the Africans) gave similar readings for all the objects depicted in a picture, suggesting that they perceived the objects as if they were all the same depth. The readings for the control pictures (objects on one plane) were similar for both three-dimensional perceivers and two-dimensional perceivers.
Hudson also found that African children and adults prefer
pictures of elephants, viewed from above, that show most of the
elephant's features (e.g. all the legs). 
Even though, in an overhead view of an
elephant, the legs would not be visible, the Africans preferred a
picture of, what appears to western eyes as, an elephant doing
the splits, showing all four legs! It was mentioned above that
Zambian workers prefer the split drawing representation of the
model cube. Split drawings appear in cultures world-wide.
American Indians of the north-western coast, people of the Sahara
desert, Siberia and New Zealand, all have used the split drawing
style in their art. Explanations
Deregowski, doubts these theories. He feels that it is difficult for one culture to understand the art of another. It seems that different cultures adopt different conventions for representing the outside world through their art. The results of the above perceptual tests support this view.
Good web site about Visual Perception