Sciencemadness Discussion Board

Comparison of human memory and computer memory

SupFanat - 22-5-2015 at 03:09

Which memory is better and for what purposes?
Thank you.

Non-volatile computer memory doesn't have to be always turned on in order to remain intact. DVD-ROM movies that are 10 years old can be still readable despite offline storage. If a human were left even for much shorter time without power, all his memory would be completely decayed and the atoms that used to be memory would be everywhere on Earth, even in Heard Island if you're in North America or in Chatham Island if you're in Europe... :o

[Edited on 22-5-2015 by SupFanat]

[Edited on 22-5-2015 by SupFanat]

phlogiston - 22-5-2015 at 03:27

IMO, you picked a very fascinating subject, but of all the possible questions that could be discussed, you picked such an uninteresting, unanswerable question.

There is no better. Each is best for its purpose.

An important difference is probably that human memory works by making associations between concepts. When you recall something, let's say a hamburger, you immediately recall many things that are associated with it. It's taste and smell, its spelling and pronunciation, the place you buy them at, it being 'fast food', the other things you often eat with it, etc.
When you try to retrieve information, you follow a path of such associations to get to the memory you need. e.g. you see someones face, and you recall his name.

Computer memory ofcourse doesn't work like that and it should not. It just recalls a byte of data from a given address. Always the same data from the same address, and no other bytes that are associated with it.

Varmint - 22-5-2015 at 04:05

The power isn't in the longevity of storage, but the cognitive power, the ability to take unrelated, seemingly useless bits of "data" and form belief systems, reasoning, and other things computers have a difficult time in doing.

It might sound like a computer could be made to do everything the mind does, but the mind does it in ways we don't understand and might never be able to "code" for. Certainly you can "code" for emotion, you define a set of inputs, the range of outputs those inputs affect, and develop an algorithm to emulate what we perceive as emotion. Smash the robot computer's "finger" with a hammer, and you can certainly program it to swear, and maybe even cry if you had the foresight to provide fluid reservoirs and a metering mechanism for the optical sensors (eyes). But will you program the robot to automatically search it's lifelong fuzzy storage and "remember" the time it had it's finger smashed when it first rolled off the production line? Can you program it to first shout profanities, cry, then remember the last time it cried and all the other robots pointing at it and laughing? Besides, wouldn't a crying robot just be a silly stab at crudely imitating humanity? Of course it would, and as the most base level would be illogical to program for such an occurance in the first place having determined early on that wet optical sensors isn't something to really plan for.

I hate to use a movie as an example, but have you seen Avatar? The power of the "forest" wasn't the ability to communicate from one plant to another, but the sheer number of interconnections in play, from one plant to many plants, those many plants communicating with others, and providing pathways from "downstream" plants back to our "original" plant, providing "data" that had the benefit of filtering/enhancement by nature of all the interconnection processing. Yes, it's a movie, but a fairly well thought out analog of the human brain and it's myriad of sequential and parallel "processing cores".

Then there is DRAM that forms the "RAM" in our computers. The "D" of course is for "dynamic", which implies readily changed, but also infers it needs to be refreshed periodically, else the bits lose their charge state and allow your data to decay. Miss several refresh cycles and you might lose a few bits (already catastrophic), miss one seconds worth? The memory is no longer memory at all, its a pseudo random collection of garbage.

Imagine the parallel processing that goes on in the human brain. There is an entire "core" dedicated to keeping the heart beating, and it begins processing while still in the womb, and essentially doesn't stop until you die of old age, some physical trauma, or disease. Imagine the number of parallel inputs to this "core": A sudden clap of thunder frightens you, and immediately the heart is commanded to beat faster in anticipation of an expected "flight" response generated by the summation of several other processing "cores". You see an attractive woman, the heartbeat increases, not to pre-process an expected "flight" command this time, but a gigantic collection of memories, emotions, wishes, all things no computer memory could attempt to fathom without millions of lines of code and a self adaptive means of processing the relevant data streams.

Human memory, while flawed from the stance of inability to remember an exact detail for millennia, is far superior in almost every other way. It might be illogical for me to swear when I accidentally hit my thumb with a hammer, but it happens automatically. If I'm a "good boy" I catch that swear in mid-stream and change it to "ouch, that was stupid", but I can pretty much guarantee it base reaction begins with "processing expletive"...

SupFanat - 22-5-2015 at 04:16

Neither DRAM nor SRAM is ever used as archival media. But biological systems use only such "archives" which can be either powered on or decayed.

Varmint - 22-5-2015 at 04:31

Pen and paper. Pen and sheepskin. Pen and tree bark. Cave walls.

Ah, there you go, Cave walls with drawn pictures (worth several hundred words each, correct?), lasting tens, hundreds of thousands of years. Millions? A timescale such that the storage layers in a CD/DVD will probably have decayed to the raw elements the dyes are formed from, while the plastic used to carry the entire thing might have turned to goo, or crystalized, crazed, and fallen to dust.

Obviously the cave wall isn't "human memory" in the organic sense, but it serves the purpose to preserve the thought process for timescales even the best of currently digital technology has zero hope in achieving.

SupFanat - 22-5-2015 at 04:37

Do you have any better idea for storage media, better than both living beings and computers?
In theory, a good idea would be making alternative Street View project (like Google) and launch some copies to space.

[Edited on 22-5-2015 by SupFanat]

phlogiston - 22-5-2015 at 04:40

Long-term memories are thought to be stored in the mammalian brain by the formation of new connections. You might consider that equivalent to permanent storage. They become built into the very structure of the brain.
Ofcourse, the brain cannot be powered-down like a computer, but that is not a specific property of memory. Biological cells in general cannot be 'turned off'. There is no biological off-switch because there is no need for one. So you can hardly use that to argue 'which is better'.

[Edited on 22-5-2015 by phlogiston]

SupFanat - 22-5-2015 at 04:44

Do you want to share your experience from journey (sailing around island) with me or someone else?
You can do it, if at all, only by technical means, not with just natural means.

Varmint - 22-5-2015 at 04:56

It depends what "better" means. For example, I could take some phenomenally inert material like ceramic, and use a computer to machine (laser or abrasion) dots and dashes to signify ones and zeros, and pretty much guarantee that stored in a place free from mechanical damage and the ravages of weather borne erosion (rain, freeze/thaw, sand abrasion) that it would last basically "forever".

But, would the bit density be high enough? Would ones and zeros still be interpretable as "data" tens of thousands of years from now? What if binary is found to be "too old school" and is abandoned, then we have future brainiacs trying to decipher this "strange, pre-historic code"?

What about DNA? Talk about storage density, that's a good starting point, and we can get bits and pieces of DNA that are millions of years old.

There's been talk of using lasers to induce phase changes in optical "cubes", where data is stored in planes in the x,y,and z axis. Purported to offer bit densities far surpassing anything we have today, but if you don't preserve the means to retrieve the data, that 1cm cube that holds the entire contents of the US Library of Congress and every movie made to date is little more than a curiously precise cube of glass.

And that's the rub: Any means of preserving data must also preserve the means to retrieve it, or, at the very minimum, contain clues on how to retrieve the data. And the method of encoding the retrieval must be obvious enough that our glass cube isn't discarded as a broken piece of ancient jewelry.

Stated another way, data density while offering the singular obvious advantage of storing more in a smaller space, also presents the biggest challenge in retrieving that data.

SupFanat - 22-5-2015 at 05:00

I agree, it's a scientific challenge.
If it works, I'd like to travel around and backup my trips to long-term storage. If no records of my trips are possible then why bother it at all...

DNA might be considered as well. But in order to consider long-term storage I need to have the data in the first place. And not simply driving around and getting...nothing. You can accept archiveless trip around the Earth but if you prefer archiveless trip then the entire discussion of storage means is pointless.

[Edited on 22-5-2015 by SupFanat]

aga - 22-5-2015 at 10:47

Quote: Originally posted by phlogiston  


An important difference is probably that human memory works by making associations between concepts. When you recall something, let's say a hamburger, you immediately recall many things that are associated with it. It's taste and smell, its spelling and pronunciation, the place you buy them at, it being 'fast food', the other things you often eat with it, etc.
When you try to retrieve information, you follow a path of such associations to get to the memory you need. e.g. you see someones face, and you recall his name.

Computer memory ofcourse doesn't work like that and it should not. It just recalls a byte of data from a given address. Always the same data from the same address, and no other bytes that are associated with it.

Hmm.

Interesting idea.

If data were stored (in a computer memory) exactly as it is now, yet with pointers to other associated data, much the same as linked-lists or relational databases are, then there'd be the advantage of Both methods.

I suspect the relating/linking would take up a huge amount of time though.

Perhaps we/our brains file data and crosslink data under the most immediately relevant 'threads', and that Sleep is required to re-process all of those temporary cross-linkages into the most appropriate places.

MrHomeScientist - 22-5-2015 at 11:33

Quote: Originally posted by SupFanat  
If it works, I'd like to travel around and backup my trips to long-term storage. If no records of my trips are possible then why bother it at all...

Even if you have a long-term storage solution, why bother with that at all? The sun will eventually engulf the Earth as a red giant. And if your data survives that (being carried to another planet by future colonists perhaps), eventually the universe will end. So if you think about it, in several trillion years no matter what you do, everything you and everyone else does is meaningless.

Have a good weekend! :D

SupFanat - 22-5-2015 at 14:53

The science isn't so far that we could make backup of Earth or even entire Solar System.
(Is this theoretically thinkable?)