When drawing, if I only use the mental imagery then I tend to draw blobs of value and symbolic shapes without respecting the true proportions of the object. What I fill in the details with instead are learned contours: motion patterns are easy to train with repetition, and when they are combined with an arrangement of primitive forms, more complex stuff can be described without visualizing it completely at any step.

If you do some contour drawing exercises, you'll notice right away how much you can fill in just by getting a decent silhouette: a lot of systems of figure drawing simply go into more detail with that and aim to codify rules for what lines are necessary.