Back in Chapter 6, we considered the problem of “space perception”, which is the problem of how we can visually perceive a 3D world given only 2D input. The retinal image is like a photograph, it is a 2D snapshot of the world. One solution involves touch. The idea is that touch has “privileged access” to the 3D nature of the world, because the input is inherently 3D, not 2D as with the retinal image.
This theory caused a philosopher named William Molyneux to pose a question in a letter he wrote to John Locke. He noted that it was, of course, normal for a person born blind to be able to distinguish the 3D shapes of objects via touch. For example, a blind person has no trouble distinguishing between a cube and a sphere held in the hand. However, what if this person was somehow given the ability to see? If we sat this newly seeing person down at a table and placed a cube and a sphere on the table before him, and without letting him touch them, would he be able to recognize the shapes? Recall that according to the aforementioned theory, 3D shape and depth perception requires learning to see in 3D by coupling touch experience with visual experience, that is, by looking at and touching things at the same time. Here’s the wikipedia article on Molyneux’s problem:
To help understand this problem, consider the cube in image found at this link:
The image is 2D. But we see it as a 3D cube. What shape are the sides of the cube? They are all square, of course. At least, that is how see perceive the sides. But what shape are they in the 2D image? The front and rear faces are square, but the sides of the cube are actually parallelograms in the image. Representing a 3D square as a 2D parallelogram is called a “perspective projection”. So the problem, for vision, is to learn how to deal with perspective…how to understand that the parallelogram is really a square. So, supposedly, touch is required in order for the visual system to learn this. Note also that the touch system is not subject to perspective problems, and that is why it is seen as having privileged access to 3D shape. When you touch the cube, all of the faces are registered directly by the fingers as square shapes, they never feel like parallelograms. So when you touch and look at the cube at the same time, you feel that all sides of the cube are square, and this tells this visual system that the sides that “look like” parallelograms are really squares. So if you read the wikipedia article on Molyneux’s problem, you’ll see that Locke and Berkeley agreed that the answer was “no, the person would not be able to visually recognize the objects”. Surely Lotze, with his theory of local signs, would agree. And Helmholtz, who argued that prior experience is needed, would also agree. However, Kant emphasized that 3D perception came from the innate categories of thought – so to whatever degree this is true, learning would not be required and Kant might say the answer is “yes, the person would recognize them.”
While it has not been possible to test the question scientifically because very few people are born blind, and for those that are, it is not possible to just “flip a switch” and “turn on” their eyes and ask them this question. But consider a different question. Could a person born blind understand perspective? Remember, touch is not subject to perspective projections; vision is the only sense that must deal with 3D to 2D projections. So the empiricists would say “no, a person born blind could not learn about perspective.” A rationalist might say “yes, because innate knowledge about depth would allow a blind person to understand perspective.” Consider these paintings:
They portray depth in perspective, including the cues of linear perspective, relative size and atmospheric perspective. These paintings were made by a man born blind, who received no education, especially none about art and perspective, and who figured out how to draw like this entirely on his own. More info about him here’s
A psychologist at the University of Toronto made this video about him:
So what do you think? What do you think is the answer to Molyneux’s problem? Or to this other problem: how does a blind person learn to understand perspective?
edit: Esref Armagan’s site has recently been suspended, so for now the links to his painting gallery do not work. For now, try this link: