- Cases of limited standard context
In a scientific text, a good illustration is simple and necessary. Forget the second part for now. What is "simple"? Some things are better presented not as diagrams or graphs, but as detailed images that never are fully discussed because that would distract the reader.
For example, X-rays of broken bones or scans of stained electrophoretic gels are a step more detailed than doodles: they tell something about quality of the process of picturing. They can be blurry or crisp, maybe unevenly, and it matters. They have context, and people who routinely read the literature will be looking for it without thinking, while people who only start to do it need to pay conscious attention. It is teachable, up to an expected level of competency and perhaps beyond. Hard to say.
But textbooks (in my experience - YMMV) don't stop to define the proper context. You're supposed to see the break in the bone, or the molecular weight of the protein, and move on. You are expected to waste a few gels of your own and gain the "clarity of vision" by trial and error, even though the skill of recognizing untrustworthy images is very important and you might need it even if you've never personally done something similar. Especially if you work in a remote field (like plant hormone research) and don't even plan to do something similar, but happen to need to check something (like what paleontologists think of how leaves evolved).
At least you often can ask a friend what to look for. (Seems that Before Infrastructure, researchers just had to re-invent that wheel now and again. Are there any ways to get insight into it? A history of science source?)
- Cases with some unknown amount of context
It's when you don't know whom to ask. It's when you can't name what you see. It's when you suspect the picture goes with the text, but have no clue what the connection is. Lastly, it's when you stop staring at it and try to fit questions to force it to make sense.
I think there should be a moment, at the beginning of studying something in depth, to have an order of understanding the image. Like, 1) what is this, 2) why is it here, 3) what was before it (in time or space) and what will be after, and 4) what is missing. (Maybe "what is missing" should go first. Sometimes the meaning of a chart became clearer to me when I looked at the axes and not at the plot.)
And what do you think? What questions should one train oneself to ask of illustrations - not just to read, but also to present?