Molyneux's Problem is whether a newly sighted man who is visually presented a cube and globe could “distinguish and tell which is the globe, and which the cube.” This seems to be about the perception of space. (Or so I have always thought.) Since the visual, tactile, and motor representations of space are different, the anti-nativist says that all correlations between them must therefore be learned. The problem for the newly sighted person is to compare the distribution of points in a visual representation of space with one in tactile or in motor space. How can he do this if he has had no prior experience of visual representation of space itself?
This, at any rate, was Diderot’s take on the problem.
"How does a man congenitally blind form ideas of shape? I believe that the notion of direction is given to him by the movements of his body, the successive existence of his hand in different places . . . If he runs his fingers along a taut thread, he will get the idea of a straight line; if he follows the line of a sagging wire, he will get that of a curve . . . For a blind man, unless he be a geometer, a straight line is nothing but the memory of succession of sensations encountered along a taut string . . . Whereas we combine coloured points, he combines only palpable ones . . ." (Diderot, “Letter on the Blind”)
Diderot's very interesting suggestion is that the sighted combine visible points by direct access to spatial relations, whereas the blind man infers spatial relations from temporal sequences. He cannot correlate these without first touching the two objects.
According to John Mackie, Locke had a different take on Molyneux’s Problem. Locke thought that our ideas of the primary qualities resemble those qualities (whatever ‘resemble’ might mean). If so, both the visual and the tactile idea of a cube resemble the cube. And though similarity is not transitive, it is not unreasonable in this instance to suppose that these ideas might resemble each other. On this reckoning, the newly sighted man ought to be able simply to look at a cube and note the resemblance of his new visual idea to the stored old tactile idea of a cube.
Yet Locke thinks that the newly sighted man would not be able to identify cube and sphere. Why? According to Mackie, the problem arises in the perception of depth. The newly sighted man can tell a circle from a square, says Mackie’s Locke. But he has not yet learned to infer depth from shading. Thus he cannot tell that a cube is a cube, or that a globe is a globe; he cannot perceive by sight alone that they have depth. He could distinguish and name a circle and a square, but not a globe and a cube.
Mackie’s Locke is questionable in two ways. First, a blind person is quite capable, even on Diderot’s account, of knowing that cubes never project a circle, not globes a square. That should enable him at least to distinguish if not to name the globe and the cube. He might think that the globe is a circle, but surely he would never never mistake it for a cube or a square.
Secondly, and more importantly, if Diderot is right, the tactile idea of a cube is a temporally united complex idea, whereas the visual idea is spatially united but temporally punctate. Not easy to correlate two such ideas without prior experience. Locke’s divergence from Diderot here arises from his treating CUBE as a simple idea, which violates atomism. Diderot, on the other hand, is explicitly an atomist in claiming that ideas of spatially extended things must consist of ideas of points.
On the other hand, Diderot’s statement of the problem neglects the basic question about intermodal comparison of simple ideas. We are not surprised to be told about Chiselden’s cataract patient that
catching the cat (which he knew by feeling) he was observ’d to look at her stedfastly, and then letting her down again, said So Puss! I shall know you another time.
Diderot puts the cube and the globe on the same footing as the cat; Mackie’s Locke puts them on the same footing as Diderot’s atomic sensible minima, i.e., his "sensations" experienced while running a hand over a string.
The question for Diderot, therefore, is this. Can a newly sighted man identify a tactile point and a motor point with a visual point? This seems like a very simple problem. Is there any empirical doubt that he can? Don’t newborn infants turn toward light? Don’t they immediately reach out to a parent’s finger? (Pawan Sinha’s patients in India could not immediately distinguish figures, but as far as I know, their visual abilities were easily equal to visually distinguishing two objects touching them on the arm.)
On Diderot’s take, then, the correct answer to Molyneux’s problem is: No, he cannot. But the question is poorly put. Properly put, the answer is: Yes, he can correlate visual and other representations of location. (This was the basis of my earlier take on this issue.)