Last night, I watched a show called
Robochick And The Bionic Boy.
The Bionic Boy was a guy with an artificial arm that read muscle
tension on the surface of the remainder of his arm. Wonderful for
him, but not what I'm covering here. I'm going to talk about the
vision system created by
Dr. William Dobelle
and implanted in at least two volunteers.
I am not a neurologist, MD of any kind, or even computer hardware
person; I'm a sysadmin and programmer.
I have, however, been paying attention to neural regeneration and
brain interface research over the last decade or so.
It is probably best for me that Dr. Dobelle is dead, because, I have
to say to him: YOU'RE DOING IT WRONG.
You can see pictures of parts of the system at
this video. In
particular, note the size of the plug in her head at 0:38. She has
two of those.
The system consists of a giant battery, and two huge ruggedized
computer systems. These then attach to two thick cables
(substantially thicker than rounded IDE cables, for those of you
who like to take your computers apart), which then plug into
giant plugs in her skull, that feed something on the order of
400 electrodes lying against the surface of her brain.
I hardly know where to start.
Let's start with the hardware pack itself.
Unless the system is ruggedized to survived being run over by a
steamroller, there is
absolutely no reason
for the system to be that big! I mean, come on! The battery is
something like 4 times the size of a modern laptop battery, and the
rest of the system is no better. The only reason laptops are as big
as they are is the monitor, and there isn't one in this system.
It's a travesty.
Age of the system isn't an excuse either: the woman whose story I
watched had this implanted in
common era year two thousand and four.
By 2001 we had subnotebooks, like this one:
One of the most notable Sony models was the Transmeta-based Vaio
PCG-C1VE or PictureBook (2001), which measured only 1x6x9.8
inches. It had a digital camera built into the lid, which could
be used for video conferencing or swivelled round to photograph
a scene.
So, she's walking around with an extra many pounds of hardware.
That's the least of her problems.
Let's talk about the plugs into her brain, shall we?
The plugs are about as thick around as my thumb. I do not have
small thumbs. Not surprisingly, there are substantial issues with
healing: the wounds don't ever seem to quite want to heal.
Why, one might ask, are the plugs so big? Because the plug
has a hole for each electrode. Seriously.
USB was invented in 1996, but the
plug has a hole for each electrode. The mind, it boggles.
For those of you who are not computer geeks, let me explain the
problem: just because you have a few hundred electrodes, does not
mean that you need a few hundred strands of cable. Human neurons
are very slow; the very fastest
transmit at 120 m/s. Light speed, which is what computers wires
work at, is 299,792,458 m/s. Human neurons, as I understand it,
fire at
about 200 Hz
(times per second). Modern computers perform operations at
2-3 billion Hz (times per second). You can drive an awful lot
of 200 Hz electrodes with a modern CPU.
All you need is a bit of hardware in the brain to switch the
incoming inputs between the different neurons, similarily to how the
electron beam sweeps across the screen in a
CRT monitor. You'd want to wrap
the hardware in plastic or something, but it really doesn't need
to be very big; think fingernail or smaller.
USB could easily drive this sort
of thing... It uses 4 small wires. That's it.
Imagine how much nicer it would be to have 4 small wires sticking
out of your head than that giant plug. It just makes my angry,
which is why I'm writing this in the first place.
Next on the list is the electrode arrangement itself. The
electrodes are all on the surface of the brain. That strikes me as
immensely stupid: most of the brain is not at the surface, so it
seems highly improbably that one could have enough communication
with the visual cortex to be effective that way. I'm not a
neurologist, though.
Now, having said that, electrodes
do seem to fail after a while
no matter what, but still... Surface only
doesn't even seem like trying.
That's not the worst part of the internal hardware design, though.
Not by a long shot. The ground of the system
doesn't go out through the wires. I'm serious. The ground of
the system is actually inside her brain, but nowhere near the
stimulating electrodes. The ground (you have no idea how
desperately I wish I was making this up) is a piece of foil
resting on top of the brain.
Whimper.
The show included an incident where the system was having problems,
so the user ended up getting a shock to her brain from the
grounding foil.
Five minutes of electrical engineering class could have told you
that was a bad idea.
None of that is the worst of it, though. The worst of it is that
they apparently completely ignored
brain plasticity.
We know from
a variety of research
that the brain is quite happy to rewire itself
given the chance. Your brain actually flips what you see from your
eyes upside down, and I can't find it, but I've heard several times
about experiments where people flipped their vision upside down with
mirror goggles and after a while (days? weeks?) it became normal for
them: their brain stopped flipping things for them, so everything
looked right-side up.
However, as far as I can tell, this vision system involves the
scientists experimenting with which electrodes to activate by hand.
This really doesn't seem to be necessary: just turn them all on,
associated with different parts of the visual field, and let the
brain figure it out!
I'd love to see a system that used thousands or millions of
electrods (or whatever the state of the art for this sort of thing
is these days) at various depths, turned them all on, and just let
it run until the brain figured it out. With only a few wires
leading out of the head, of course. Hell, I'd be willing to try it
out myself as an augmented reality experiment.
I just feel sorry for the people having to live with the giant plug
version.
If I'm full of crap, do
feel free to let me know.
-Robin
;){kind=link}