The idea that it’s going to be possible, in the foreseeable future, to “upload” a human mind to a computer is, I believe, quite wrong. The difficulties are both practical and conceptual, as I explained at length and in technical detail in my earlier post Your mind will not be uploaded.
I’ve now summarised the argument against mind uploading in much shorter and more readable form in a piece for The Conversation – a syndication site for academic writers. I’m pleased to see that the piece – Could we upload a brain to a computer – and should we even try? – has had more than 100,000 readers.
It’s led to another career milestone, one that I’m a little more ambivalent about – my first by-line on the Daily Mail website: Would you upload YOUR brain to a computer? Experts reveal what it would take to live forever digitally. There was also a translation into Spanish in the newspaper El Pais: ¿Podríamos cargar nuestro cerebro en un ordenador?, and into German in the online magazine Netzpiloten: Könnten wir ein Gehirn hochladen – und sollten wir es überhaupt versuchen?
Having read “Your mind will not be uploaded”, it seems that the argument presented doesn’t support the conclusion in the title.
– clearly molecules are important in the brain, and in individual cells. Molecules are also important in fluid dynamics, yet we are able to get the right behavior whilst ignoring them. Not everything that goes on in a brain cell is necessarily (a) relevant to the things you care about and (b) useful for simulating the brain. For example, we can be relatively certain that life support components such as mitochondria can simply be ignored – they do an important job, but that job is not an information processing job.
To be honest if I had to guess, I would speculate that molecular information processing in brain cells is used to make sure that they perform a well-specified role as an information processing unit. (The relevant sections of) our DNA doesn’t have anywhere near enough storage space for all 10^11 of our neurons to be unique special little snowflakes with relevant unique molecular dynamics going on inside them. Of course there will be such dynamics – but it must be generic in that very large numbers of neurons (millions) must have essentially the same (on average) internal dynamics, which can then be represented by an approximation, and the differences between these cells – what makes them unique – must be connectome level information, i.e. to whom they are connected and with what strengths. Furthermore, suppose that a significant fraction of the molecules in our brain are individually important from the point of view of human thought. Well, there are rather more such molecules than seems reasonable. 10^11 cells each containing 10^9 protein molecules (http://book.bionumbers.org/how-many-proteins-are-in-a-cell/). Why does the brain need 10^20 computational units to achieve performance on e.g. image processing tasks that we can do in computers with vastly fewer simulated neurons?
The simpler explanation is that these 10^9 molecules per cell are not each individually important, but rather their aggregate behavior is what is really relevant.
Anyway, thanks for a thought-provoking article.
Roko, perhaps I haven’t made my point about the molecular basis of biological computing clear. It’s not just the trivial point that neurons are made out of molecules; it’s that the individual molecules involved in signalling processes within the neurons, quite literally act as individual logic gates through mechanisms such as phosphorylation and allostery. So it’s not obvious that their behaviour can be aggregated to higher level laws, in the way that fluid mechanics reflects the interactions of the component molecules of the fluids. Dennis Bray’s book “Wetware” is very good on this.
Naturally, the number of protein molecules that are directly involved in information processing is still a lot smaller than the total number of molecules in the brain. However, how many information processing molecules there are is itself a dynamic quantity that responds to the information the brain is processing (for example it is this sort of behaviour that gives rise to the plasticity of connection strength at synapses).