0
$\begingroup$

When writing a futuristic story I always try to minimise the impact of computerization. Its rather uncomfortable writig a story when technology is advanced enough that humans arent in a position to come into conflict with one another.

One thing about my current setting is that they use biological robots to compensate for the lack of computer power to do their bidding. As a consequence they would be able to build enormous brains to do their research or control large scale projects. Which in turn raises the question: how much brain power could I expect from such a thing?

If we look at brains of the greats like Steven hawking, Einstein or John Von Neumann we could expect great things if we scaled their brains up. But when comparing brains to computers, its the computers that always take the cake. This is a bit of an unfair comparison as much of our brains isnt busy with the important calculations but with unimportant stuff like keeping your chemical balance in check, staying upright, thinking of how to act in social situations, handling the complexity of laguage, how to have sex with a good partner, gathering food and other trivial things.

So I'm wondering, if we assume the minimum brain power for survival is met, how much brainpower can I expect from each KG/Volume of brain?

If that makes it easier (it probably wont), lets say we have a brain of 6 metric tons (and a few tons of biomaterial to keep it alive and interacting in some form with the world).

$\endgroup$
5
  • 1
    $\begingroup$ "technology is advanced enough that humans arent in a position to come into conflict with one another". Citation needed. $\endgroup$ – Kepotx Apr 23 '20 at 12:33
  • 2
    $\begingroup$ A human brain has a volume of about 1.2 liters and weighs about 1.2 kg. So at best you would expect a brainpower of about 0.8 VonNeumanns per liter or kilogram. If you would like a better unit of measurement than the VonNeumann, you must specify what you mean by brainpower. $\endgroup$ – AlexP Apr 23 '20 at 13:11
  • 1
    $\begingroup$ @AlexP since it is very hard to find a good way to measure brain power it is not useful to try and constrain the question even more. Any way to get an idea or quantify the amount of brainpower is good. $\endgroup$ – Demigan Apr 23 '20 at 13:22
  • 1
    $\begingroup$ @AlexP perhaps a better argument: if I had asked about a laser powered by an engine how much power I would need to destroy something particular, you wouldnt hesitate to use watts. But an answer in horsepower the engine would need to power the laser including energy loss would be equally fine right? I THINK that brain power is usually done by computations per second. But thinking is different from knowing. And if someone can offer a clear and understandable answer in Von Neumann's per liter then thats all we need right? $\endgroup$ – Demigan Apr 23 '20 at 13:52
  • $\begingroup$ Your question sound very much like Craniometry, mainly used accoriding to wikipeadia for "Theories attempting to scientifically justify the segregation of society based on race", although it has a modern day revival it is not useful to measure intelligence. Brain size does not say anything about brain power, the density of neurons en the density of interlinking and network cluster nodes seem a much better representative of intelligence. Much the same as CPU's my old 365 chip was bigger than my current I7 CPU. $\endgroup$ – D.J. Klomp Apr 25 '20 at 13:42
1
$\begingroup$

A quick search reveals, that the human brain with an average volume of 1.2 litres can achieve about 10^13 analog cacluations per second.

Assuming we have an evenly distributed neuron desity, this results in 8.3 * 10^12 calculations per second (or 8.3 tera flops but not really, since flops stands for "float point operations per second") per 1 litre of brain volume.

$\endgroup$
5
  • 2
    $\begingroup$ Those numbers are ridiculous, from an IT engineer's point of view. The only way to reach such absurd numbers is to count all processing performed by individual neurons; but this is utterly irrelevant. Transferring this to a low-power ARM processor and counting all processing done by individual gates would reach even more absurd numbers; for example, counting in the same irrelevant way, we could say that an Apple A7 processor with its billion transistors performs over 10^15 operations per second. $\endgroup$ – AlexP Apr 23 '20 at 13:06
  • $\begingroup$ @AlexP You leave out the fact, that the brain only does analog calculations, which are way faster than the float point operations modern processors perform, since every neuron can perform an operation. But it also limits the calculation precision. $\endgroup$ – OneSaltyAceTanker Apr 23 '20 at 14:13
  • $\begingroup$ Do you know what a logic gate is? I can assure you that it does no floating point calculations. (And what's this obsession with floating point calculations? Some processors do floating point operations some of the time. Most of time they don't.) $\endgroup$ – AlexP Apr 23 '20 at 18:04
  • $\begingroup$ It also depends on how the neurons (or transistors) are structured. If you are able to design a neural network or circuit for a single task, you can achieve that kind of efficiency. (For instance how our brains are able to predict the arc of a ball and catch it.) But in order to do any kind of general work or intelligence you need a large group of those neurons or transistors dedicated to processing general instructions, and you will get far fewer actual calculations per second. This is why the human brain can only think about a few abstract math calculations at a time, while computer=faster. $\endgroup$ – Brianorca Apr 25 '20 at 18:58
  • $\begingroup$ Is this is effectively saying each neuron can perform some floating point operation individually? Because that is ridiculous. $\endgroup$ – BMF Apr 27 '20 at 0:15
0
$\begingroup$

they use biological robots to compensate for the lack of computer power to do their bidding

Well, if this would mean straightforward calculations then the electronic processor has a clear advantage.

If instead the 'bidding' involves taking complex decisions based on fuzzy and incomplete data then a biological brain may have an advantage. Especially if you consider its self-programming capabilities.
Training allows these super brains to solve similar problems but also elaborate an advantageous behavior in case of unexpected scenarios.

$\endgroup$
2
  • $\begingroup$ It is always nice to see that the biological brain thinks the biological brain has an advantage :D. $\endgroup$ – D.J. Klomp Apr 25 '20 at 13:31
  • $\begingroup$ @D.J. Klomp Just for the meantime..... :-D $\endgroup$ – Duncan Drake Apr 25 '20 at 13:39
0
$\begingroup$

There is a problem with the way this is conceptualized. Brain power is not explicitly (certainly not linearly) related to brain volume. There is a critical volume in the hominid line around 1100 or 1200 CCs (if I remember correctly) in which we start seeing the capacity for abstract thought, but above that critical value brain volume doesn't seem to have much direct relationship to intelligence. In part this relates to the fact that high-level cognition occurs on the surface of the frontal lobes, and that this surface area increases more as a function of the convolutions of the brain than its actual volume (the way a large piece of paper can be crumpled down into the same volume as a small piece of paper); in part it related to the fact that the complexity of neural connections is likely more important than the quantity of neural material. But the point is that simply making brains larger does not necessarily make them more powerful.

Brains are not simple bit-processing computers, though I know that analogy is often used. They are actually more akin to quantum computers in that they process information holistically, with local changes having global impact without passing through some defined central processing unit. The real advantage of the human brain isn't its 'power' per se, but rather its flexibility: intuition, creativity, analogical reasoning, shifts in perspective, etc. These things allow the brain to 'see around corners' (as it were), and make plans for the unknown or adapt to unexpected events. The limitations of linear (machine) processing is that it is algorithmic in the final analysis, and algorithms are non-adaptive. Would a Big Giant Brain be smarter than your average bear? Maybe, depending on how it was educated. But it seems more likely that it would just be more 'human', with a mind and personality of its own based on its own idiosyncratically creative perception of the world. It would lose the thing we value in modern computers — that sheer, mindless, myopic dedication to tasks — and start developing 'traits'. What will we do if the organic super-mind we created to solve the mysteries of the cosmos gets bored and decides it wants to take up landscape painting?

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.