some things were once the case in extra-cinematic reality and, subsequently, that such and such is the case in the fictional story of the character Holly Golightly.
Another big objection would be that the visual specificity of secondary seeing is canceled by the role of inference. I think this charge is groundless. Secondary, like primary, cognitive perception involves the generation of cognitions and beliefs about objects and situations by visual means. The processes by which these judgments are acquired may well be, at some level of description, reasoning-like or computational. But cognitive perception in general, and secondary seeing in particular, need not involve any conscious intellectual procedures of forming judgments on the basis of visual evidence. Seeing that the cat is on the sofa, seeing by the gauge that the gas tank is empty, and so on do not entail acts, however swift, of deliberation or justification. On the contrary, they short-circuit inferential exercises and discursive processes that might otherwise stand between seeing and knowing. 58
Hence the sensory state's informative properties might straightaway cause a suitably constituted viewer to recognize a depiction of Hepburn and to see by the image that she had stood before Tiffany's. Maybe a good rule of thumb is that where the boundary between cognitive perception and epistemic achievements in general is vague, "seeing that" and "seeing by" locutions fail to denote distinctly visual accomplishments to the degree that conscious inferential processes intervene between seeing O and coming to any realizations about O or a situation systematically related to it.