Jeff Hawkins’ 1000 brains intelligence

Setting a valuable path for brain-based AI

Are robots able to be made to act in a similar way to the human brain’s function? Here a robot appears set up to move by motor calculation, not patterns, just as is done by today’s limited machines. Photo by Possessed Photography on Unsplash

Jeff Hawkins book, On Intelligence, was published in the early 2000s. As my interest over the previous 20 years was how to emulate a brain on a slow digital computer, I found his book refreshing and competitive.

Why competitive? Nobody wants to live in a world made better by someone else! (joke from the sitcom Silicon Valley). Since then, Hawkins has also published a second book to do with ‘A Thousand Brains’ that sounds similar to my stadium model of the brain. Let’s find out more.

On Intelligence comparison

The main difference between my model, Patom Theory, in which brains are modeled by decomposed, atomic patterns of only sets and sequences and Jeff’s prediction model is levels. Patom theory starts with its learned representation — patterns. If you have a pattern (say some situation) and it has sequences that follow it, the brain can anticipate one of those subsequent situations. That anticipation is a prediction. Prediction comes from the selection of a sequential pattern.

Patom theory is a level below a prediction model. It explains representation, but not as much the use of representation. For example, if you are hit with a strap, your body will remember the subsequent pain. If you…