What is the computational power of our brain?

Complete neuron cell diagram. Based on latest experiment dendrites are more than wires connecting neurones, they have computational capabilities. Credit: LadyofHats/CC

Image of a neurone with its dendrites and their spines (red dots). Experiments at UCLA indicate that computation happens both in the neurone and in its dendrites. Image credit: Shelley Halpain/UC San Diego

Has neuroscience been on the wrong track for centuries? Image from the article on the Conversation, link in the text. Credit: Justin Pickard/Flickr, CC BY-SA

Many years ago, when I was young and computer have just begun to enter the lay person imagination, the computer was called "electronic brain". The back mind was that a computer was like a brain, just made by silicon and way slower. However, as technology progressed the computational capacity of the "electronic brain" have increased enormously and many scientists begun to wonder about the point in time (the singularity) when the computational power of a computer would equal and then exceed the computational power of a human brain.

Whilst we have a standard way to define (and measure) the computational power of a computer we do not have an accepted way to measure the computational power of a brain and this basically is because we do not know how the brain, as a whole, compute.

Kurzweil estimated, in his book "The singularity is near", that a 1,000$ computer in 2020 will have the same processing power of a human brain and that by 2045 the intelligence of a computer will exceed the intelligence of a human being. Others claim, with well articulated reasoning, that the computational power of a brain will not be matched by a computer within this century.

This diversity in "opinions" underlines the fact that we do not know enough about our brain.

Now experiments performed by a team of researchers at UCLA have found that what we thought were just passive wires used to transport signals from one neurone to another, the dendrites, have computational capability. 

There are 100 billions neurones in our brain and the estimate of a brain computational capabilities have been based on those numbers. Each neurone can have more than 100 dendrites (and each dendrites is composed of "spines" that increase its connectivity potential), hence one could say that if computation also takes place in dendrites as well as in neurones the overall capacity is far larger.

Researchers that carried out the UCLA experiments suggest that the brain computational power may be a hundred times larger than what estimated before. 
I am not sure that this is true. Before coming to an accurate estimate of the brain computational power we need to understand how the brain really processes information and the paradigm it uses to do that. 

In a brain there is no separation between memory and processing, processing changes memory and memory changes processing at a fundamental level. My personal opinion is that we already have computers (and their software) that are more intelligent than humans in certain areas and the number of these areas will grow over time. We will reach a point when, for all practical reasons, computers will be more intelligent than humans although we will not realise when that point will be reached. Computational power surely is needed to increase a computer "intelligence" but it may not be necessary to have more computational power than a brain to achieve that point. And, conversely, achieving larger brain computational power in a computer may not lead to computers outsmart human brains.

I am not alone in feeling that we know too little to be able to evaluate the actual computational power of our brain. In a recent article on The Conversation it was claimed that a radical rethink of the way we look at the brain is needed to understand it.

Author - Roberto Saracco

© 2010-2018 EIT Digital IVZW. All rights reserved. Legal notice