I think it's like this: there's multiple layers to the process of seeing something. Light enters the eyes, it's converted to an electrical signal in the nervous system, that signal is formatted into the "video feed" in our brain, and that picture is matched with existing patterns we know. Visualizing something is working backwards from the resultant pattern to produce possible inputs that would match. It doesn't go to the video feed level, but it goes to the intermediary stage where patterns are being matched to a picture. Less "what are traits I associate with an apple" and more "how does it feel when I recognize an apple?"