You can think abstractly about geometry problems, or graph tgeory but even about linear algebra. Feynman describes how he felt while holding and analyzing problems in his mind. I'm less eloquent.

Also you can think about the narrative and the valence of a scene for a book, without blabbering to yourself while planning the story. I can narrate it like as if I imagine a friend being there and translating my thought into language but it's an extra step.

It also exhausts me more. Which makes coding with LLMs more tiring for me. Before it, I could code without language use exhausting my brain, but talking to LLMs requires putting all of it into English, while before I could just imagine a certain refactor as one coherent thought in my mind without turning it into sentences.

Of course above a certain level of difficulty, putting thoughts down helps for problem solving, but a lot of that can be various types of sketch drawings and formulas too.

It's likely a verbal and spatial heavy thought difference, the lawyer / engineer type of thinking difference.