From: Doug Sutherland (wearable_at_earthlink.net)
Date: 2002-02-05 05:11:50
Hi Moritz ...
>> 3) The "training protocol" (thresholds etc)
> Maybe we could use Fuzzy Logic for the training protocol?
> I think it would be ideal for this. Does anybody know if
> this has been tried before ?
Engineers are an interesting personality type, they are
some of the most creative and innovative people around.
Their "art" is engineering, and they like to paint very
vivid pictures (see my jacket for example).
As it turns out though, it seems that engineers are
actually the WORST people to prepare functional specs
for any given system. Engineers like to try stuff
"just because you can" and "because it's cool". Left
unattended they will keep designing forever. Years
ago I worked in financial services IT. We had this
big project called 'project gold'. This thing would
have gone forever into complexity, eventually one of
the executives said it was a philosophy not a project.
So let's go back the the PROBLEM. What is an EEG
training protocol? Anyone? Anyone? Bueller? Bueller?
(Ever seen Ferris Bueller's Day Off?) <grin>
http://home.earthlink.net/~wearable/bueller.wav
We need to get OUT of engineering mentality and take
a step back, look at the big picture. There is no
way in hell we can build good EEG software unless we
understand what the user and/or practitoner actually
does with such software.
I wish we had neurofeedback practitioners among us,
who could help us get this right from ground zero.
Rob Kall, if you're listening, this is where we
could really use some help. I have played around
with Pentiston (sp?) protocol. There is very little
info available on EEG training protocols. So the
challenge is, how the heck do we write good software
if we don't even know how it shoud be used, from a
user and/or practitioner perspective. I will do some
searching for info on EEG protocols, but I already
know that there is almost no info on the web, as I
have already searched. This is precisely why both
me and Rob Sachs agree that functional specs are
the right way to proceed. What are we building?
Waves are not the end of the puzzle! In fact they
are not good at all for user feedback, better are
stacked bar graphs, pie charts, etc. And even then
we need to understand CONFIGURATION; ie what is
the TARGET of the excerise? Relaxation? Increase
of alpha waves? Or what? It will depend on the
person, session, etc. So we need to come up with
some scheme that facilitates a good "user experience".
This is why I said before that we would be wise to
analyze some existing EEG software and figure out
from there what we need. If we think we are the
right people to design EEG software (features) I'd
say we are the blind leading the blind ...
Regarding fuzzy logic, I love these kinds of ideas,
but before we build the "eeg of the future" let's
build the best-of-breed "eeg of today". Then we can
go wild and explore advanced usages. Some really
interesting stuff is being done with neural networks
and EEG. In fact, I can buy PC/104 neural coprocessors
that will plug right into my little mobile system.
Check out some of these BCI ideas:
http://home.earthlink.net/~wearable/biopsy/#symbiosis-eeg
Namaste,
Doug
Here's a little Marvin Martian from the cartoons
http://home.earthlink.net/~wearable/Marvin04.wav
This archive was generated by hypermail 2.1.4 : 2002-07-27 12:28:38 BST