[Neuroscience] Re: Memory formation via a machine

r norman via neur-sci%40net.bio.net (by r_s_norman from _comcast.net)
Sat Sep 22 20:38:17 EST 2007


On Sun, 23 Sep 2007 00:09:32 -0000, chadmaester
<chad.d.johnson from gmail.com> wrote:

>Could memories theoretically be created in a person's brain if precise
>parts of the brain could be stimulated with electric impulses?
>
>If yes, then suppose eventually through research we came to an
>understanding of how memories are physically encoded in the brain (how
>close are we, by the way?). Also, let's suppose we developed a method
>of looking into any part of the brain and decode memories in order to
>determine what memories exist already for particular concepts. Would
>it be possible to then create the necessary memories and associations
>in a person's brain which would allow them to then, say, understand
>and speak a foreign language (which they had no prior knowledge of)?
>How about knowledge to perform new physical activities, for instance,
>martial arts (I suppose neurons in the spinal cord and other parts of
>the body would need to be trained as well, no?)?
>
>What kind of timeframe would be necessary to create a sufficient
>amount of memories for the person to understand and speak the target
>language sufficiently? A few hours? A few days (across multiple
>sessions)? A month?
>
>I realize memory formation depends on raw materials such as chemicals
>as well, so wouldn't the person's body need to be provided with a
>dosage of "raw materials" required for the massive amount of memory
>formation that would take place?
>
>(I am a CS major who is interested in studying neuroscience, so please
>forgive my lack of knowledge, and please go easy on me if I sound
>stupid :p ).
>

Suppose I reverse the problem on you.  Suppose I am a neuroscientist
studying CS and I want to make a computer speak a foreign language.  I
don't want to 'train' it or even program it using techniques and
strategies derived from artificial intelligence and programming areas.
No, I am thinking of a completely hardware solution on an embedded
system.  I want an electrical engineer to go into the masks used to
created the circuit boards and the masks used to created the chips
plugged into the boards to do the job.  Of course, a lot of the
solution is stored in the ROM chips that hold  the program so I want
the engineers to create mask-programmable ROM's and specify the 0's
and 1's from scratch.  Incidentally, whoever does this must be totally
ignorant of assembler language programming for the CPU.  All she/he
must do is just generate the proper 0's and 1's to do the job.

So what do you think of that project?  Frankly I believe my  problem
would be far easier to solve than yours because there do exist
techniques of seeing and modifying the hardware at the microcircuit
level for electronics while the corresponding techniques do  not exist
to see and modify more than a few hundred or, perhaps someday a few
thousand or tens of  thousands or the cells involved.

we  are able to monitor and control activity in only a handful of
cells.  We can see general patterns in specific areas of the brain and
similarly control them but that averages out the function of tens and
hundreds of thousands of cells.  We do not understand what 'memory' is
or how it works or where it might be located.  We do not understand
what 'language' is or how it works or where it might be located.  Yes,
you can say there are "memory centers" or "language centers" but that
is no more specific than saying there is "computer memory" or
"computer arithmetic processing unit" and inside each of these "units"
is incredible complexity totally unknown.  Aside from that, the
problem sounds quite simple!





More information about the Neur-sci mailing list