I certainly agree with my understanding of what you are saying but, as a computational cognitive neuroscience with a focus on intermodal processing in the human brain, you should consider other large areas of empirical science as well. Here are a few articles to challenge your thinking with empirical facts that your explanations must also address.

In my world, normal human natural language is a much more direct window on neural computation than neurophysiological consideration and thereby provides a richer understanding than you are getting, at least so far, in this first article.

https://medium.com/liecatcher/how-your-brain-computes-41ebe7428ff9

I believe you are making a common error in ignoring human natural language as an object of study that should not be relied on as a way to explain computations. I discuss this in my review of Mark Solm's book on consciousness, qualia, and such.

https://medium.com/liecatcher/off-on-mark-solms-3d87462ab327

Finally, this is how I see the computations of natural language and the brain. I see an organ in the brain for human natural language based on findings in your field, and a number of other equally informative fields of empirical inquiry.

https://medium.com/liecatcher/natural-language-and-your-brain-237185770b00

Finally, I think human natural language directly observed as an object of inquiry best reveals the massive and precise unconscious intermodal neuronal processing in any animal brain, and you are seeing this illustrated by our modelling successes in AI/ML/NLP/Vision. This agrees with the work of Terry Sejnowski best understood in his book talking about the Deep Learning Revolution from the point of view of neuroscience.

Here would be my take on the evolution of the uniquely human brain. I agree with Chomsky who supposed that natural language evolved in a evolutionary instant. By "instant" we might see it evolve in 10,000 years, not a million. The natural language organ, and natural language itself, is simply a means to use existing computations that can serialize and deserialize massively parallel knowledge computations using similar computations present in the most primitive animals that deal with serializing sensory inputs and serializing motor performance for the sake of survival.

https://medium.com/liecatcher/the-natural-evolution-of-human-lies-655e983ee6c6

I don't particularly have much interest in "consciousness" and would prefer for science to simply account for whenever someone uses that word (or word sense in any natural language) in a context understood by someone else.

It is a important word to understand for its computation, as are many other words, like "truth" and "lies" for example. This is the approach of Language Neuroscience.

I'm very happy to follow your writing and your observations which, as I said, in my understanding of them, are quite apt and quite correct.

Thanks for this nice article.