Slashdot Mirror
Ask Slashdot: Storage Capacity of the Human Brain?
STUR
submitted this interesting question:
"Since humans are supposed to have such great minds,
I would like to know how much storage capacity a human
brain has compared to, let's say, a computer's hard drive.
Hrm... If the capacity is high enough, do you think that
computers of the future could possibly use the brain as a
sort of hard drive or ram chip? Just a little something
to think about." I've heard that the brain uses a form
of holographic storage to archive its information and I don't
know if there is a direct mapping to that and say terabytes
of information (warning: I am not an expert!). What do
all think?
This is sort of a silly question for Slashdot, since most people are going to be talking out their asses.
That said, (and talking more around my ass, than out it), there isn't any sort of storage figure. Researchers do not have much understanding about how we remember things, but it IS fairly certain that there is no relationship to the way computers store information (ie, the concept of terabytes, etc).
Generally the brain remembers certain aspects of an experience -- wether an external experience, or an internal one. Its believed that the act of experiencing something, or recalling it later starts changing the relative levels at which nerves will fire and accept the chemical impulses from neighboring neurons. (Before anyone starts talking about electrical impuses, those are only conducted within the nerve cell not between nerve cells and its not an electrical impulse as much as a chemical shift within the nerve that changes the electrical potential of the local region while the signal travels down the length of the nerve -- thats why you can have your hand off a hot stove before you actually feel its hot)
So a memory is generally a tangles mess of restimulations of fragments of what happened. Thats why with few exceptions, most people can't really remember details very well, and everyone is prone to manipulating memories. (ie, you read an interesting tale when you're young, later in life you're sure it happened to you or that someone TOLD you it happened to them, and not that you read it) Things like that happen a disturbingly large amount of the time, with everyone. Luckily such errors don't often affect anything serious.. I mean who cares where you heard a story?
That's why things like memory and attention span and personality can be manipulated chemically -- because you can control the way those experiences link up with each other and how the brain reacts to those experiences.
One of the most interesting things I think people find when they really start studying how the brain learns, and stores its experiences is how little actually comes from the senses or memory. (For example, how the brain can only distinguish general colors and shapes beyond a half-dozen degrees off center in your field of view, but you're constantly fooled into thinking you can see more than you really can)
The question with the brain then is how discret these fragments of memories and experiences are, how many times they can crossconnect with others to produce memories without those crossconnects getting so blurred that you get confused about the truth of what you're remembering, and the number of different fragments that make up a given memory.
Most likely no one will have any idea about the answers to those questions until there is a better understanding how a "neural network" arrangement can store and rerecognize patterns of nerve impulses when the "matrix" used is numbering in the millions of cells at a time...
Actually, human vision is the aspect of the brain that I find most fascinating, partially because my professional specialty is artificial vision. I find it fascinating how much more advanced our biological vision processing is that anything we can achieve artificially.
Several fascinating aspects of human vision processing:
The "raw" "pixel" resolution of the human eye is actually flaberghastingly low, on the order of 200 x 200 pixels (the effective resolving ability of the rods and cones).
However, the human eye "snaps" about 10 - 12 "frames" per second (maximum, in good light) and the brain integrates subsequent frames, each with very subtle positional differences, and compares adjacent "pixels" from frame to frame to assemble an image of dramatically higher resolution. Thousands by thousands of "pixels" when required by the task being performed (e. g. threading a needle or intricate soldering). This is why staring at a small object for some length of time is necessary before we perceive all the most subtle details.
An additional "weakness" of the eye for which the brain performs some amazing processing to compensate for is this: The rods and cones of the eye are "recharged" by flushing a fluid containing rhodopsin across the retina. Rhodopsin is a protein that breaks down and emits a tiny electro-chemical current when struck by photons of light. The speed of the rhodopsin decay is proportional to the intensity of the light hitting it. The retina "recharges" when the previous charge of rhodopsin is nearly depleted. This works out to 10 - 12 times per
second in bright light, much less (down to a minimum of perhaps once per second) in very low light.
Anyway, the flaw in the above scheme is that the "dose" of rhodopsin that each rod or cone receives in any given "recharge" is very poorly controlled. It varies all over the place. This means that the electrical current emitted by any given rod or cone for any given intensity of light from frame to frame is not consistent! So, the brain has to analyze the average current emitted by each rod and cone, over the surface of the retina and over time, and integrate this information to produce and accurate and detailed internal picture inside the brain!
The analogy is this (for all you artificial vision programmers):
Imagine that your boss gave you this task:
We are going to give you a CCD camera with an array of 200 x 200 pixels. We will rapidly vibrate the camera so that by integrating the subtle changes between adjacent pixels you will, after storing 30 frames, interpolate a picture with a resolution of 5000 x 5000 pixels. Furthermore, the brightness value digitized by each camera pixel is going to randomly vary by 200% for any given actual light intensity. Your system has to output a real time image flow at the above resolution and a brightness accuracy of +- 0.01%.
Yeah right!
But this is analogous to what the brain does!
I've always been more impressed by the brain's processing power than by its storage capacity.
Disclaimer: I'm not a neuroscientist, but I've been reading a lot of books on this recently, and there are huge differences between computer and brain storage that make this kind of measure meaningless.
:-)
First, the "write" operation is highly dependent on how you experience an event. You can't be fully, simultaneously aware of every input- the brain is an excellent signal filter, and only processes those aspects of the environment you are focused on. But there's also an "interest" component- even if you are really paying attention an input, the aspects of it that you find important will be what you remember. Example- there was a study where a researcher asked workers in a museum about a particular painting they all saw on a regular basis. No two people described it the same way- some described the colors, others the emotions they felt as a result of the content, still others the execution of the painting and the specific stylistic elements. And what they remembered correlated closely to what it was about the painting they were "interested" in as part of their job- the curator's recollection (style, context, etc..) was very different from the guy who cleaned it (complicated, hard-to-clean frame).
Secondly, a very important aspect of remembering is uniqueness- something distinctive about a memory that allows you to get a "handle" on it later. It's also thought that multiple, similar experiences tend to blur each other and reinforce the common elements between experiences. For example, I can tell you exactly how I get to work and what lanes I prefer to use, but I can't tell you the exact sequence of lane changes I made on any specific trip.
Third, the brain has a very powerful reconstruction mechanism. It's kinda like dinosaur skeleton reconstruction. Just as a paleontologist can fairly accurately reconstruct an entire skeleton from a relatively small number of bones (or bone fragments) your brain pulls together and reconstructs the few bits of a specific experience that were stored and synthesizes a more detailed rememberance from the fuzzier "generalized" remembrances to give you the impression of remembering much more detail than you actually stored.
This all contributes to explaining why it is so difficult for humans to remember "digital" data. For most of us, there's very little that's interesting, unique, or distinctive about the numbers in a sequence. Mnemonists with apparently infinite abilities to recall details generally have a learned or innate mechanism by which they create unique, distinctive symbols for number sequences which make it possible for them to remember them. In the most highly-developed cases, these symbols encompass every sense- sight, sound, taste, texture, smell.
But such people are often cognitively lost in details... they can't deal with concepts easily, and can't abstract over information they have taken in, since they are so overwhelmed by the distinctiveness and richness of the details. As are computers, which know nothing except detail. So the "lossiness" of the human memory actually serves a useful purpose, and is a large part of what makes us "intelligent" relative to a piece of silicon.
Of course, I'd like to have it both ways...
I nead a defragger. ;)
+--
Given infinite time, 100 monkeys could type out the complete works of Shakespeare.
+-- (Score:-1, Moderator on Power Trip)