This is not at all correct. There are hierarchies created ephemerally, but the organizaition of knowledge in the neocortex most certainly is not hierarchical. (Possibly some form of heterarchical but definitely not hiearchical). What is true is that every 'perception or action' subsystem (e.g., visual perception, throwing a ball, speaking English, understanding English (or any other natural language), always create hierarchical emphemeral organizations of information. You see this in the fact that you can easily graph the 'grammatical structure' of any sentence hierarchically, or can throw or catch a ball, or know what is in your "field of view".

All that machine learning stuff is garbage speak. We have known Hebbian networks for 70ish years.

To this day we do not know what of the infinite possible organizations of discrimination networks are actually at work in the brain. The coding still eludes us. Part of the problem is that the "experts" are not educated in the right way to understand how to ask the right questions. AI guys are exploring the space but frankly have no motivation to do the hard fundamental work, nor do most even care to. So they enumerate different 'systems of convenience' to win various ill-framed contests and make money.

Like the world of navigation before Newton, or electricity and magnetism before James Clerk Maxwell. We still don't have a good idea about how the brain works in detail. Just generally. And most of our knowledge is pretty rudimentary.

Here is what I say about Numenta's stuff...

https://medium.com/liecatcher/a-thousand-things-about-a-thousand-brains-ec44598ef363

We are not living in a simulation. We are living in an incredibly fast, precise, discriminated, emulation of reality. And, amazingly, natural language proves all our brains do the fast, precise, discrimination exactly the same way, with detailed agreement among people driven by what they can say to each other in natural language (and NOTHING ELSE).

You miss the fact that predications are fundamental to all neocortical computation.

https://medium.com/liecatcher/lie-diseases-and-implications-for-ai-nlp-53e8bd2abce2

Also, finally, the evolution of this computation was evolved over BILLIONS of years. Ion channels among cells gave us multicellular organizations, left-right-symmetry, and other symmetries, and yes, these evolved into the 'ultimate' ion channel that sparks, the neuronal synapse. Even the most rudimentary, reptilian, worm, real-life neural network performs discriminations of great precision and speed. As do plants with THEIR ion channels among cells.

Just say'n.