The Naked Scientists

The Naked Scientists Forum

Author Topic: How much memory does a brain have? How much information can it store?  (Read 25927 times)

Offline Alan McDougall

  • Neilep Level Member
  • ******
  • Posts: 1285
  • Thanked: 14 times
    • View Profile
Seany,

Does this give a perspective, three times the total data in the Library Of Congress
 

Offline graham.d

  • Neilep Level Member
  • ******
  • Posts: 2208
    • View Profile
Ahh we are all showing our age here. First computer I worked with was the Manchester University Atlas. Later on the first mini-computers came into being like the DEC PDP8 (12 bit word).

Although the brain has a very different architecture to today's computers, it should still be theoretically possible to "map" a brain into a Von Neumann architecture (for example) even if not very efficiently. In that way the number of bits of memory could be determined. This would probably result in many more memory bits being used than the brain equivalent though, because of the inefficiency. Mapping the brain's plasticity and fluidity (the ability to forget for example) would take up memory in the software to do this. The brain's structure is very important in itself and not just the amount of memory.

An electronics example is associative memory (or Content Addressable Memory) used in specialist applications in the communications industry. Here you present a 64 bit word (for example) and it returns the memory address (or a sequence of addresses) for which there is a match. Partial matches can also be looked for by applying mask bits. It is a sort of backwards operating memory and has some similarity to the way the brain can sometimes operate. Implementing such a memory on a computer with software is very inefficient (a simple way is to simply search sequentially looking for matches) but takes a many times longer to complete.

Another issue with determining the number of bits of memory there is in the brain is how to translate analog level sensing into bits. Neurons work on analogue levels not just 0s and 1s. This is very efficient in the brain, but a digital computer would have to represent each of these analog levels with a number of bits (depending on the resolution required), which is not efficient. Counting neorons does not give you the number of memory bits therefore.

The question of how much memory is in the brain does not address the structural complexity issues which really dominate the problem. On one hand a computer could be built with sufficient memory (achievable today I would expect) but without any software or really too much guidance on how to organise the system behind any software, or, on the other hand, we could map various subsystems on to dedicated hardware and software sections and connect them into some overall hierarchy. The first is the more economic approach but would not produce an efficient system. But both systems depend on a deep knowledge of how the brain is structured. We are not even close to this level of understanding.
 

The Naked Scientists Forum


 

SMF 2.0.10 | SMF © 2015, Simple Machines
SMFAds for Free Forums
 
Login
Login with username, password and session length