Recently Stephen Hawking apparently said “I think the brain is like a programme in the mind, which is like a computer, so it’s theoretically possible to copy the brain onto a computer and so provide a form of life after death.”, which understandably generated a fair amount of noise on Twitter and elsewhere.
I think it’s safe to say that he was either misquoted or that there was a translation error, given his mode of communication, so that what he really meant to say was that the mind is like a programme in the brain, which is like a computer. Even so, many people saw this as another example of people from other fields (not neuroscience) clumsily throwing bad metaphors around. As was semi-seriously pointed out these people are often physicists, who, I guess, seem to think they know it all.
Personally I too think it’s a limited metaphor but the notion of copying your mind onto a computer at some point seems perfectly reasonable to me. Granted, the computer in question may look nothing like the laptop I’m typing on right now, may have an independent body, may need to evolve and grow via experience rather than being programmed and may even depend on physics that we do not yet understand but it will still be a computer in the most mundane sense. A piece of technology invented by humans capable of processing information.
The capture and transfer of your current mind state is a whole other world of technical problems but again, perfectly reasonable to assume that it is doable.
What happens next is a very interesting question and I think that it is not clear that the resulting mind will evolve from that point forward in the same way as if they had remained in their original organic body. From their point of view I guess they will never know and possibly never care as their new reality will be the only one they know.
I find it surprising though how many neuroscientists or interested scientists make the case that this is a ridiculous notion and that the simple physicists simply don’t understand the complexity of the issues. Some comments referred to our woefully limited ability to model the human brain right now or the staggering compute power needed to implement these as proof of how daft this idea is.
One particular blog article from 2007 came up again where the “10 Important Differences Between Brains and Computers” are listed. From the comments it seems that the author is actually talking about a PC (as in one you can buy today) when he says computer but if so the case hardly needs to be made.
In any event I thought it would be interesting to briefly look at each of the 10 points, written more than 6 years ago, to see what the state-of-the-art is now (with a focus on EU research) and if they are still relevant.
1. Brains are analogue; computers are digital
I would disagree on both counts here, I think “analogue” is an oversimplification of how brains actually process information (Shu et al. 2006) and computers are not always digital. Analog computers were relatively common until the 1960s and recent research has revived the idea for biologically inspired computing (FACETS project).
2. The brain uses content-addressable memory
So do electronic computers if we wish (See RAMPLAS project).
3. The brain is a massively parallel machine; computers are modular and serial
Computers can be massively parallel also and are increasingly leaning in this direction.
The Neuromorphic Computing Platform from the Human Brain Project will be massively parallel by any standards and will address many of the points raised here, building on the FACETS project mentioned above.
4. Processing speed is not fixed in the brain; there is no system clock
I think we can agree that processing speed is not fixed but there are clock like mechanisms. And as mentioned in a previous post some go so far as to put a number on it (40Hz in case you are wondering).
5. Short-term memory is not like RAM
It is undoubtedly different but the metaphor is, functionally at least , not a bad one. As the author points out they both need “power” and from an architectural point of view this seems to me an important similarity. The key issue being that it is not permanent unless energy is expended to transform the information.
6. No hardware/software distinction can be made with respect to the brain or mind
This is a good point but bio-inspired architectures are designed in the same way. The “code” is generated based on experience. See Biologically Inspired Computingfor some nice examples.
7. Synapses are far more complex than electrical logic gates
This relates to point 1 above, if we accept that brains are not simply analogue or digital it makes little sense to make this comparison. It also overlooks the quite complex physics involved in a “simple” electronic logic gate.
8. Unlike computers, processing and memory are performed by the same components in the brain
9. The brain is a self-organizing system
Both true for a PC but once again ignoring bio-inspired architectures.
10. Brains have bodies
This one is a very interesting point and extremely relevant for the metaphor. I think we all agree that embodiment is essential for the future of AI but again there are already many groups working on this since Brooks in the 80’s. A nice overview of the topic can be found in Wilson & Golonka 2013
Bonus: The brain is much, much bigger than any [current] computer
This one is the least relevant I think as we know that computational power increases exponentially. We will have all the computational power we need assuming we figure out how to use it.
Interestingly this example was given via twitter recently of how far we are from simulating the human brain. I actually see this as a sign that we are almost there.
As for copying our minds onto a computer? I think that if the laws of physics allow it, someone will make it happen and a lot sooner than you might expect.
