On Intelligence by Jeff Hawkins - Confirmed Prediction is Understanding

According to Jeff Hawkins in his book On Intelligence, human understanding is the act of our neocortex correctly predicting, in parallel across all our senses, the world around us. Confusion happens when our neocortex has incorrect predictions. For example, if you are walking on a concrete sidewalk, when you take a step forward, you expect the ground to be firm. If, on your next step, you were to sink into the ground like quicksand, you would be surprised. You would not be expecting soft ground. Your brain would have incorrectly predicted solid ground and confusion would result. This notion applies to all human interactions from opening doors to sitting in cars or interacting with other people and even breathing. We have a set of predictions in our brain and when they are violated it gets our attention.  Jeff’s book is interesting because in it he is describing his theory for how a neocortex works. His goal is to recreate the human neocortex using software algorithms. 

What will the future effect be if Jeff Hawkins is able to successfully create a neocortex using software?  How will the theories and products that Jeff Hawkins is working on change our predictions of what will happen in the future?

Humans not only predict second to second but they also predict tomorrow, next week, next month, next year and so on.  Predicting stock market movements correctly and acting on them can make you rich.  Predicting the best education for your children can give them a great start in life.  Predicting the reactions and thoughts of others can give you an edge in human interaction.  So how can we make sure we have the best information to make great future predictions?

I’m no expert, but I do like to seek out, read about, and discuss those individuals, trends, theories and ideas that are the most likely to have a large impact on our future.  And so I have created the Future Effects blog to write about these ideas and hope to find others that are interested in these ideas as well.

Jeff Hawkins' story continues, in March 2005, Jeff Hawkins, together with Donna Dubinsky and Dileep George, founded Grok (originally named Numenta). The company's goal is to simultaneously create a theory of how the brain works, and a computer algorithm to implement this theory. They have been using biological information about the structure of the neocortex to guide the development of their theory on how the brain works. So far, they have come up with two major algorithmic frameworks: Hierarchical Temporal Memory and Fixed-sparsity Distributed Representations. The frameworks can find patterns in noisy data, model the latent causes, and make predictions about what patterns will come next.

Two commercial software programs have been created from Jeff’s original theories so far: 

1)      SightHound Video which can recognize people and objects in the same way that a human brain works.

2)      Grok which automatically learns what patterns are normal for a data stream, then highlights any unusual behavior.

There is open source access to the Python/C++ implementation of Jeff's theories in software here.

Will Jeff Hawkins’ neocortex theory change the world?  It still remains to be seen, but I think it is a story worth following.