1

EDIT This question is extremely badly worded and I would appreciate it if you could ignore it's content. I cut and pasted parts of two related but unfortunately distinctly different questions into one post and was unable to edit it to my satisfaction for a period of time. In particular, the Moore's Law references are not germane to the question. My apologies for this error.

I have voted to close my own question, due to its inherent unfixable problems, and if anybody else with sufficient reputation wants to vote to close, please feel free to put it out of its misery:) Again, my apologies for my posting, before checking for errors. END EDIT

There are estimated to be 86 billion neurons in the human brain. I am ignoring a further 90 billion non neuron support cells. It is difficult to establish a comparison (as regards "packing density") between the number of transistors in an integrated circuit CPU of a current computer and the neurons in a human brain.

As of 2015, the highest transistor count in a commercially available CPU (in one chip) is over 5.5 billion transistors, in Intel's 18-core Xeon Haswell-EP.

My obvious problem here is that I simply don't know how directly we can compare a human neuron with a presumably much simpler man-made transistor. I will assume they can be directly compared, but I realise this is a completely ad-hoc assumption. Human Brain Energy Use may be of relevance here.

My question is, can we estimate how closely our brains conform to Moore's law? My rationale is that, although a neuron is different in many ways to a transistor, if they both ultimately follow quantum mechanical principles, then I have assumed they both follow Moore's law, in the context of "you must keep the wires a certain distance apart".

In particular, it would be interesting to know, assuming a neuron is much more complex than a transistor, if our brain's cells actually follow the Moore's law.

Community
  • 1
  • 1
    Don't edit the content out of your post. That goes against our policies on vandalism. You can edit to clarify the question, but not to get rid of it entirely. (If the "delete" link appears below it, you can use that.) – David Z Sep 11 '15 at 13:36
  • I will give you the tip that evolution certainly does not follow Moore's law type scaling laws in general, but this is a biology and not a physics question and you should ask the details about that in a different forum. – CuriousOne Sep 11 '15 at 14:18
  • 12
    The question as stated makes no sense - the brain obviously doesn't double in capacity every 2 years. Are you asking if the growth of neurons in children is exponential? Or maybe if the growth of the average human brain over millennia was exponential? Or if the growth of the human brain's average capacity has slowed to due the artificial-selection caused by society? What do any of those questions have to do with Physics? – BlueRaja - Danny Pflughoeft Sep 11 '15 at 16:25
  • 2
    @BlueRaja-DannyPflughoeft you are perfectly correct. I prepared two related questions this morning, then I cut and pasted them in a rush, mixing them up and posted this one by mistake. When I realised that it was extremely badly worded, I tried to delete it but could not, and I asked the moderator to delete as it had received an answer, and you can read the comments above regarding deletion. I apologised on the post and put an edit pointing out the chain of events, asking users to ignore it. I then deleted the vast majority of the question, but this was rolled back. –  Sep 11 '15 at 16:35
  • I guess you could interpret your question as "has the brain had a steady increase in density over the last x million years" at a rate that's obviously no where close to Moore's Law. But that would probably be more biology related than physics. – David Zech Sep 11 '15 at 18:15
  • 1
    @DavidZech I am unfortunately unable to edit it, currently anyway, to what I originally intended it to mean, but I won't forget Moore's law anytime soon, for sure:). My own fault, for rushing it..regards –  Sep 11 '15 at 18:26
  • 6
    I'm voting to close this question as off-topic because I don't really see how it's a physics question. The reference to quantum mechanics seems like a red herring. – DanielSank Sep 12 '15 at 04:44
  • @AcidJazz to be clear: you can edit the question to clarify it. Just don't use edits to completely change what it's asking, and don't use them to remove the question entirely. – David Z Sep 12 '15 at 07:24
  • I'll chip in and offer the last close vote. It's been suggested that this question be deleted, which is another issue.... – David Z Sep 14 '15 at 04:49

3 Answers3

15

Moore's law from the Wikipedia article:

Moore's law is the observation that the number of transistors in a dense integrated circuit doubles approximately every two years (since the invention of integrated circuits).

His prediction concerned the rate at which the density of transistors in circuits increased with time. So I'm not sure it makes sense to apply Moore's law to the human brain. Even if you could compare neurons (or groups of neurons) to a transistor, the human brain evolved over many millions of years and isn't reinvented on a regular basis the way that integrated circuits are.

Siraj R Khan
  • 1,958
  • Thanks, I may have worded the question badly, I will try to rephrase to reflect the title better, that is, do we know if the brain follows Moore's law. –  Sep 11 '15 at 12:43
  • 3
    I would say the answer is that we do know and that the human brain does not conform to Moore's Law. The law is about how the transistor density increases with time, with each new iteration of the circuit. Assuming there are features of the brain that we can identify as transistors, the human brain evolved over an extremely long time. Also I'm not sure how much the densities of neurons increased with time but I highly doubt the density of transistor-like structures doubled every two years as human evolution just doesn't work that quickly. – Siraj R Khan Sep 11 '15 at 12:56
  • 4
    Did you mean to ask: `Is the human brain limited to further increases in neuron density the same way that we think circuits will be/are restricted to increases in wire density, due to the effects of quantum tunnelling?' – Siraj R Khan Sep 11 '15 at 13:03
  • Thanks, no, I think I just got caught up on the wrong train of thought, as you say, evolution does not work that fast. –  Sep 11 '15 at 13:08
  • Also, transistors implement binary logic but the function of a neuron is not quite so simple (pulse trains may seem equivalent but amplitudes and action potentials are valued continuously, not discretely). Neural networks in computing can be used to model the type of activity patterns seen in biological brains but each neuron is represented by a data structure in a digital computer and only simulates continuous variables (with things like double-precision floating-point variables). These do not scale linearly with CPU transistor counts at all, the processor just simulates time passing. – Darren Sep 11 '15 at 23:56
  • ...continued, neural networks can be implemented with analog components but afaik (in my very limited knowledge and experience) the performance gains to be had with modern digital computers have eclipsed other approaches. – Darren Sep 12 '15 at 00:00
  • 1
    The brain is increasingly more neuroplastic than we've ever given it credit for. So it's plausible that there IS a neuron "Moore's law" that acts over larger timescales. – Daddy Kropotkin Jan 16 '19 at 17:50
  • Hey @N.Steinle Thanks for responding. I actually don't disagree with you, you raise a very interesting point. I'm not a neuroscientist so I can't really comment on neuroplasticity, you have me at a disadvantage there. I certainly don't want to mislead anybody on here into thinking I'm a brain expert! The point I was trying to make in my original answer is this: Moore's law has a very specific definition. It's about exponential growth of density of transistors on integrated circuits by a factor of 2 every 2 years (roughly). Continued below: – Siraj R Khan Jan 17 '19 at 20:18
  • By replacing transistors with neurons we're already no longer technically talking about Moore's law. If we remove the doubling every two years, it's even further from Moore's law. It's just an example of exponential growth. That being said, maybe there is a process in the brain where the doubling in density of neurons happens every 2 years. Maybe at some point during growth from baby to adult there's a similar increase in density. Or when neurons are repaired after damage. I honestly couldn't say. The answer to this question really depends on how much we decide to dismember Moore's law! – Siraj R Khan Jan 17 '19 at 20:19
  • I think we can all agree it's not really a physics question. – Siraj R Khan Jan 17 '19 at 20:19
3

Although it isn't possible to answer your question, as the Human brain doesn't evolve its physical appearance in short time spans comparable to Moore's Law, I found some information that may help with the gist of your question.

There are many differences between the human brain and a computer that make a comparison between their potentials difficult. Volume occupied by the brain may be the best measure of its potential.

Difference #2 in the above link explains that memory retrieval in the brain is different from byte-addressable memory retrieval in a computer. The brain seems to have a more efficient retrieval system than a computer, IMO.

Difference #7 says the processing that takes place at brain synapses is chemical and spatial, as well as electrical, and is much more complex than the logic switches made of transistors. Synapses are not at all comparable to transistors.

The brain also is a good deal more adept at self-repair and self-change of its circuit utilization and design. It may be in this area that the brain has the most promise for increasing its intellectual capacity in a manner comparable to Moore's law. After all, human brains make Moore's law possible.

The human brain underwent a spurt of rapid development from 800,000 to 200,000 years ago, during a period of volatile climate change. The rapidity of climate changes from cold to hot and back again favored adaptable tool-makers who could adjust their social and material lifestyles quickly. The brain had to "earn its keep", because it uses an enormous amount of energy and leaves humans at a disadvantage in physical ability compared to other animals. This link from the Smithsonian Institution shows the period of rapid development and compares it to the previous 4 million years of relatively slow development (click on the chart comparing brain size with climate change volatility).

Ernie
  • 8,516
0

In addition to the physical constraints on size, one has to consider the fact that the parts of any machines will degrade. When comparing a conventional machine to the brain, you need to consider the self repair and also the fact that the system is supposed to be copied to produce the next generation. The system has to be reproducible from scratch.

If you consider a machine consisting of macroscopic parts then there is a fundamental problem with maintaining the integrity of the system. The system itself needs to contain the information about its own formal description in order to be able to maintain itself or to be able to make copies if itself. However, each macroscopic part of the system can be in an astronomically large number of physical states (the entropy of the part), while the system only keeps track of a few macroscopic parameters. There is then no way for the system to prevent itself from degrading on the long term.

The only way this problem can be averted is if the system is build out of parts whose formal description will be encoded in the system such that there is no missing information that the system cannot act on.

Count Iblis
  • 10,114