When we read a sentence we don't "pay attention to" each letter and the structure of each letter and the placement of the letter and color and lighting and all that. We just "end up with" the meaning of the sentence.
From a computational perspective, you would parse the sentence into "tokens", convert those tokens into a tree of nouns and verbs and such, resolve the ambiguities, etc. All of this process requires a lot of data structures. That is, you need trees pointing to strings, and trees pointing to other trees, etc. This could create dozens of intermediate data structures. In the end you want a final data structure with the meaning. But now we have all these intermediate structures we don't really need anymore. What is needed is some sort of garbage collection.
So I'm wondering how the brain deals with this. How the brain "loses" or "forgets" or "discards" the intermediate information when parsing a sentence (or interpreting other visual or auditory information). Wondering how it knows to not permanently store that information, which would end up overcrowding our brain quickly. Wondering how it "ends up with" the final appropriate data structure. At least at a high level.