Continuity of Memory

We are human because of our technology. Few of us can remove all of our technology and live outside in nature. In the winter most of the population would be dead of hypothermia in hours and in summer we would be pestered by insects and end up sunburnt. Our ability to live across this planet is due to our technology be that clothing, footwear or water and food storage. Critical to our continued evolution as humanity is our  continued utilization and evolution of technology. There is no going back unless we plan to not be human. Some people seem to have a blind spot about technology and divide it into arbitrary classes, and draw a line at a particular technology advance e.g. they stop at pre-steam 1800s technology or they stop at pre-electricity technology or they argue that any advanced technology is unnatural whilst forgetting the clothes they are wearing and in the end they have missed the point of what it is to be the human that made that first step Out Of Africa so many millenia ago.

Technology will complement all our person so whereas clothes act as a second skin then can we do this for the brain ? Is it remotely feasible that we could even store a full set of memories of the brain in the non-biological and in doing so maintain a continuity of that set of memories ? Various sources list the human brain as having 100 billion neurons with average of 7000 synaptic connections to other neurons. So that’s say about 700,000 billion connection records as a crude measurement but are we in the ballpark ?

If we optimize and have 4 bytes a record so we would be looking to store about 2800 Terabytes (2.8 Petabytes) and this article in the Scientific American suggests a similar number (2.5 PB) though the connection count is lower. Now today a retail hard disk is now 2 Terabytes (2,000 gigabytes) so we’re looking at 1400 hard disks of say the WD WD20EARX at around 100 GBP each so about 140,000 GBP or about 235,000 USD. The power used would be about 5W x 1400 or about 7 kW if it was all running at once. I remember my first retail hard disk – 10 Megabytes – in the early 1980s. Today I think I consume that in storage every hour of everyday.

We are in the ballpark. About every 2-4 years the bit density doubles. In two years time that 1400 disks will drop to 700 disks and so on. Somewhere by 2036 there could be a disk with the capacity to match the human brain in storage at 235,000 dollars in today’s money and somewhere by 2066 it could reduce to 1 dollar in today’s money.

It will not be using current technology (which is perpendicular magnetic recording (PMR) ) but we will evolve new solutions to solve problems as we humans have for all the millennium since we evolved  in Africa. The evolution of our technology is a punctuated gradualism in that there is a continuous change to an existing technology and then a rapid jump when a new technology obsoletes an older approach.

In the short term the bit density maximum of PMR of 1 terabits per square inch is replaced with technology by Seagate called heat-assisted magnetic recording (HAMR) which can get us to 10 terabits per square inch. This is a leap and will be available by 2016. Other new approaches here that could work are using bit-patterned media or self-arranging polymers for platter surfaces, or using antiferromagnetism in more exotic storage systems, or using non-disk approaches of solid-state technology with phase-change random access memory or memristor-based memory.  I don’t know what will win out. In fact we can’t know before you try as the properties of physical materials are deterministic but unknowable so until you actually do the science to make the material we don’t know what they really do.

Such a storage of memories is never a person as it is not functioning any more than a hard disk on its own is a functioning computer. This is a thought experiment to see if the technology we have can be in the right ballpark and I think it can. My vote on replicating our brain function will not be on traditional discrete media such as a disk or static RAM with a separate CPU but on a combined processor-memory with both the memory and simple processing logic being one and the same – smart memory – that mimics the existing functionality.

Hearts pump and brains think. From a functional point of view we can replace hearts with mechanisms so there isn’t much, other than knowledge of how to actually do this, that should stop us replacing the function of brains. Sure it’s not happening soon but it doesn’t seem impossible.