There's a version of visual work that happens at the end of a process. You've figured out what you think, and now you need to communicate it. You make a diagram, a slide, a render, and it explains something already clear in your head. This is how most people think about visual output: as communication, not cognition.
There's a different kind of visual work, and it's rarer partly because it's harder to defend professionally. You make something visual before you know what you think, in order to find out what you think. The image isn't a record of reasoning that already happened. It's how the reasoning happens.
This isn't a new idea. Rudolf Arnheim was arguing for it in the 1960s. His book Visual Thinking makes the case that perception isn't passive — that looking at something is an active cognitive process, and that visual representations aren't just pictures of ideas but carriers of thought in themselves. The diagram on the page has properties that exist nowhere else. You can examine them, find implications you didn't intend, and revise your thinking in response.
What Schön was watching
Donald Schön's work on design cognition gets at something similar from a more empirical angle. He spent time observing designers at work and noticed that drawing and talking were running in parallel — not that designers figured something out and then drew it, but that drawing and figuring something out were the same activity.
His phrase for this was "reflection-in-action." Designers hold a conversation with the materials of their work. You make a mark, the mark talks back, you respond to what it says. A sketch isn't a neutral record of intent. It has its own geometry, its own implications, and those implications constrain and suggest what comes next. "Doing extends thinking" was how he put it. The action generates information that wasn't available before the action.
This is hard to square with how visual work gets presented in most professional contexts. A portfolio shows finished work. A presentation shows conclusions. The exploratory drawing — the one where you were still figuring out what the problem was — rarely makes it into any public account of the process. The method stays invisible, and people new to the field often don't learn it exists.
The render as instrument
The hardware concept renders on this site came out of something close to this approach, though I wouldn't have described it that way at the time. The research phase — reading patent filings, looking at supply chain component photos — produced a rough set of constraints. Port placement, approximate dimensions, the profile suggested by CAD leaks. But constraints aren't a design. There's a gap between "the logic board is probably this wide" and "what this product will actually look like."
Building the 3D model was where that gap got explored. The model forced decisions the source material didn't specify. How does the lid hinge meet the base? What's the transition between surfaces? How do the port cutouts affect the visual weight of the side panel? You couldn't answer those from the CAD leaks. You had to build something and see what it implied.
That's the sense in which the render was research rather than illustration. It wasn't depicting a known thing. It was generating a possible thing and then evaluating whether it made sense — whether it looked like something Apple would make, whether the proportions tracked with other products in the line, whether the design language held together. Sometimes you'd build something and immediately see that a proportion was wrong. Not because you'd calculated it, but because you'd made it and looked at it.
Barbara Tversky on external representations
Barbara Tversky's research on spatial cognition adds another angle. In Mind in Motion she argues that spatial representations — diagrams, sketches, maps — don't just communicate structure, they create it. The act of arranging things spatially on a surface changes how you think about their relationships. Once something is external, you can see it from outside your own reasoning process and evaluate it in ways that are harder to do internally.
She makes a point about working memory that's relevant here. Our capacity to hold complex relationships in mind without external support is limited. Externalizing through a sketch or diagram doesn't just help you show others — it helps you hold more of the problem in view at once. The surface becomes a cognitive prosthetic.
This is one reason early-stage visual work pays off beyond what you'd expect from just getting something down on paper. The drawing creates an artifact that sits outside your working memory and can be examined, revised, and related to other things. You can compare two sketches in a way you can't compare two mental images.
The limits of the approach
None of this means visual thinking is always the right tool, or that all problems benefit from being approached spatially. Some problems are better handled through calculation, through language, through formal logic. The question is which kinds of problems respond well to visual exploration — and those tend to be problems with spatial properties, relational structure, or design constraints that interact in ways that are hard to reason about abstractly.
Hardware design fits that description well. The form factor of a laptop isn't arbitrary — it reflects a set of constraints (battery size, keyboard area, screen proportion, thermal requirements) that all interact. Building the thing spatially lets you see how those constraints push against each other in ways that are genuinely difficult to reason about without a physical or visual model.
The method also requires tolerance for ambiguity that not everyone finds comfortable. You're committing to a visual direction before you have a clear sense of where it leads. Some people find this productive — the constraint of having committed to something specific forces the question of whether that thing is right. Others find it disorienting, because the thing they made doesn't match what they were imagining and they're not sure which was wrong.
What this looks like in practice
Using visual work as research means building things earlier than feels comfortable. Before you're confident about direction. Before the idea is fully formed. The goal isn't a good sketch — it's a sketch that forces questions you didn't know you had. Rough is fine. Precise is counterproductive at this stage, because precision forecloses exactly the ambiguity that makes early-stage exploration useful.
It also means paying attention to what the visual work tells you. It's easy to treat a sketch as a delivery mechanism for an idea you already had, and to stop looking once it's good enough to communicate. The more productive habit is to look at what you made and ask what's wrong with it — not in a critical sense, but in a diagnostic one. What does this tell me about the constraints I'm working with? What has this made visible that wasn't visible before? What would I have to change to make it feel more right, and what does that tell me about what "right" means here?
Arnheim's basic claim is correct, I think: perception and thinking are not separate faculties operating on the same material. Visual representations don't just express thought — they shape it. The most useful thing visual work does isn't communicate to others. It's think for you. The sooner in a process you start using it that way, the more useful it becomes.