From: kai@cyklop.nada.kth.se-dd-dd  (Kai-Mikael J\d\d-Aro)
Subject: Re: HARDWARE: What about SQUID's?
Date: Thu, 27 May 1993 21:02:30 GMT
Organization: Royal Institute of Technology, Stockholm, Sweden


In article <1trr82INNhvu@news.u.washington.edu> cyberoid@u.washington.edu (Bob Jacobson) writes:
   In article <1tp74sINNpvr@news.u.washington.edu>
   robman@pinn.nacjack.gen.nz (Rob Manson (Robman)) writes:

   >Does anyone know anything about the use of Supeconductor Quantum
   >Interference Devices (SQUID) as a form of interface?
   >
   >I read an article about some Japanese researchers using them to
   >dirrectly interface with the brain.

   As I understand them, SQUIDs are incredibly sensitive devices, built
   around near-zero-degree superconductors, for measuring electrical
   waves.  I have heard that a SQUID can be used to detect brainwaves,
   but only in total isolation from the surrounding environment (like in
   a submerged, shielded vault)...  and even then their outputs are
   unreliable.  How a SQUID could be interfaced to the brain is beyond my
   meager knowledge.

In an interesting coincidence, I'm currently working less than 50 m
away from one of these monsters at Karolinska Hospital in Stockholm
and a couple of weeks ago had a chat with the people running it.  The
SQUIDs are used for MEG, magnetic encephalography, that is, as Bob
says, you can pick up the electrical activity of neurons in the brain,
with much greater precision than with surface-field sampling EEG
electrodes.  (The method used is very much like in the Polhemi we all
use and hate :-), only much, much more sensitive.)

In conjunction with PET (Positron Emission Tomography)  and MRI
(Magnetic Resonance Imaging) you can get a pretty good idea of what
(groups of) neurons are active during given tasks, and then you can
make lots of fancy testing and so on to try to figure out what makes the
brain tick.  (Did you know that your brain constantly moves within
your skull?  Not much, but it moves.)

The basic problem with interfacing with the brain is of course that we
still have only a *very* approximate idea of what the brain is doing
when it's doing what it's doing - we can see the activity, but it's
not clear what it's for etc.  Lots of research left.  Of course one
can use coarse knowledge, such as letting the subject twiddle the
fingers on her right hand, which will cause activity *here*, and then
twiddle the left fingers, which causes activity *there* and let that
mean something, but that can certainly be solved a lot easier and
cheaper, such as by using a keypad to twiddle your fingers on.  Then
you don't have to lie still with closed eyes in order not to confuse
the readings either.

Anyway, if you think DataGloves and Reality Engines are expensive and
cumbersome, don't even think of using SQUIDs - we're talking
multimegabucks here.  The MEG unit is an entire room in a Faraday
cage, and of course the SQUIDs are cooled with liquid helium which
costs outrages sums as a certain amount constantly boils off.


-- 
Kai-Mikael J{{-Aro	email:  kai@nada.kth.se	    "Don't win, don't lose."
IPLab	 		voice:  +46 8 790 91 05		-- Yoshigasaki-sensei
NADA, KTH		fax:    +46 8 790 09 30
S-100 44  Stockholm	telex:  812 6154 7052 SICS
SWEDEN			teletex:2401-812 6154 7052=SICS