It may be easier to think of it in contrast to just looking at your reference and drawing what you see. It's not about "I see a shape, I draw a shape". It's about taking in the visual information the image provides and thinking about what it's telling you about how each individual bump (as an example of a textural form) exists in space. Is the bump tall, is it short, is it wide, is it pointy? How are the bumps arranged across the surface, are they arranged in clusters and groups or uniformly?

A mistake many students end up making is that they try to find the cast shadows in the reference image, then draw exactly the shadows they see. This does not allow them to rearrange the texture, to apply that texture from one kind of surface to another, because they're fully reliant on the photo reference as dictating every aspect of that texture, and end up limited only to copying it as they see it. What the texture exercises in this course get into however is being able to "unwrap" that texture from one structure and apply it to another, to move the light source around as needed to suit their needs.

That's why it requires us to understand what the reference is telling us about what exists in 3D space, not merely replicating it exactly as we see it.