This must have taken forever to write. Good job. I’ll just give some general thoughts on what I manage to read. I can’t comment much on the ideas presented in it, so I’ll focus on the writing. In my experience, improving writing often improves the ideas themselves, and all ways of improving it help a little because they make it more interesting and clear, so more vivid in your mind. My thoughts are not vivid at night, so these comments are coming from a lazy reader.
Before I start saying what I think you should change, I’ll say that I honestly don’t care about those things. I end up tossing 90% of my ideas and bitter towards 3/4 of the rest for wasting my time, and progress in AI and neuroscience have been historically really slow, so all that matters is what ends up working right. 99 pages is a lot of chances for great ideas and certainly a lot of good writing to build off of.
You should ask yourself what is your target audience. People who want to get lost in thought about high level frameworks or ways of thinking? People who want to think about specific mechanisms and their implications for high level frameworks? People who want exact mechanisms and results? Etc.
I recognized that the configuration of neurons in the cortex is too simple to store in the same manner as the platform.
I would reword this part. Do you mean individual neurons are too simple, and in what aspect? Whether or not it’s just because of biology being imperfect and evolving by trial and error, people spend decades studying just one part of the brain because there’s always more complexity. I think the solution to AI will be fairly simple, but getting there is hard. If you emphasize how slow progress has been despite how much we know about the brain and acknowledge that every idea about AI is far from complete, readers won’t take issue with whatever they disagree with.
You should explain the platform before saying you applied it to reverse engineering, even just in broad terms or what it is.
Start off with your ideas, at least the basics of them, after any broad / motivation-explaining / tone-setting stuff. That way, the reader gets a sense of where you are going with things like receptive fields. It’s a lot easier to remember those sorts of things when they are all part of the point when they are first described.
You describe some things in the introduction which are explained in On Intelligence, like the cortex being a sheet 2-4 mm thick. You could refer the reader to On Intelligence because I assume a lot of your ideas are inspired by it and Numenta.
Another feature of the cortex is its lack of features: it is uniform. That uniformity is taken to mean that neurons are the same, they are organized in the same way with respect to each other and they process signals in the same way.
The cortex has a lot of specializations depending on the species and region, but besides those tricks, implementation differences, newer advancements to core mechanisms, or variations between regions shared by all species (e.g. what/where pathway perhaps), it’s uniform.
