Are Sense Data Images?
It is tempting to think that SDT should take (visual) sense data to be (visual) images of some kind, but it is not a good idea. It might sometimes be useful to use the concept of image metaphorically to explain something about sense data and their functions. For example, John Locke's appeal to the camera obscura in the Essay might serve to explain something about ideation. Antonio Damasio's appeal to a sort of movie theater might, too. It is more likely that they are invitations to disaster.
First, it should be obvious that if the camera obscura or movie theatre is used as a metaphor for perception, then, unless some restrictions apply, delimiting its precise metaphorical function, we are at risk of falling into a homunculus regress. The camera obscura and movie work because of their engagement with viewer's sensory system. If we say that a perceiver's awareness of sense data is like a viewer's using the camera obscura, we suppose an inner viewer to account for the outer viewer. Careless use of these metaphors vindicates the criticism of SDT that it assumes a homunculus regress.
Second, and more fundamentally, the basic assumptions of SDT clearly imply that sense data are not images in the way that, say, photographs and paintings are images. This is true not only of SDT, but any generically representational account of experience, such as Intentionalist-Representationalism, for example. In order to see this, it will be helpful to consider first how images are supposed to work when we conceive them from a Direct (Naive) Realist framework.
Direct Realists should say, roughly speaking, that a decent image of a scene should be an artefact that suitably resembles the scene. Thus, oversimplifying, a colour photograph of a golden beach with green-leafed trees by a blue lake on a sunny day would have a golden section and a blue section with some green patterns. Further, the Direct Realist should say, we can see the similarity of the photograph and the depicted scene by holding up the photograph when at the beach. Trompe l'oeuil art gestures at the possibility that the image might be taken to be the reality.
Representational approaches, including SDT, must say something different. No particular premium is to be put on similarity of image and depicted scene. The crucial element for the successful image is that it causes a perceptual state in the viewer which is suitably similar to the perceptual state produced by the scene. In some cases, images will be radically different in character from what they depict.The night sky and a canvas covered in heavy metal chemicals look the same because they cause broadly the same visual condition in the subject, not because they resemble each other in any essential way. If they look black, it is because, in the sky, the photons never come back because they keep on going, and, for the canvas, they vanish into the orbital chemistry of the oils.
The sense datum theory says that sense data are not artefacts which cause us to have the same perceptual state that we have when we experience what the artefact depicts, as photographs, drawings and paintings are. They are, rather, elements of the very perceptual states which are the respective effects of image, on the one hand, and depicted scene, on the other. According to SDT, sense data are not images.