[Noisebridge-discuss] openeeg

[Jean Rintoul]

Hiya,

I'd just like to add that I think you can get some good stuff out of EEG,
and it isn't all just noise. Noise is a big problem and I agree that it is
hard to detect anything reliably. This is why you can use learning
algorithms.
The Emotiv headset is an example of this, and you will find some believers
out there who tried it at the 3 GDC's it was exhibited at, as random people
could come up and try it. There are many complex technical hurdles to
overcome to get something to work reliably on all people, cheaply. Also, you
can most definitely see alpha, some muscle movements and eye movements
plainly and simply with the naked eye using a BioSemi EEG headset... no
mystery there.

It's true, EEG won't read your mind or see your dreams currently, but it's
limitations are due to different types of noise, spacial smearing/absorption
over your skull, and your abilities to isolate Event Related
Potential<http://en.wikipedia.org/wiki/Event-related_potential>'s
through machine learning algorithms. You can most definitely recognise and
distinguish different types of muscle movements, again limited by how well
you can repeat the same action, and how good your algorithm is. It's an open
area of research, and there needs to be more EEG based development. Here is
some:
http://ida.first.fhg.de/projects/bci/competitions/

This one is MRI, but still, impressive to think about what we can see non
invasively? MRI is much lower sample rate but higher res than EEG.
http://www.nature.com/nature/journal/v452/n7185/full/nature06713.html

At Emotiv, a research engineer/electronic music producer guy called Glen
mapped the arousal algorithm through to music, as well as playing out the
raw EEG in matlab. Turns out raw EEG multiplied up into the audible range
didn't sound so sweet. Glen basically changed the music tempo, adding layers
of compelxity to the track based on your arousal score. This sounded much
nicer than raw EEG! Again, it wasn't some sort of psychic music creation
system... but depended on how creatively you could interpret the signals.

Other sensors can also be intelligently integrated, like GSR and pulse
oximeters. Making a basic EEG(just a couple of electrodes) and integrating
with these sensors could tell you a huge amount about a person's mental and
physical state. Seems like this could be used for lots of interesting
projects.

Jean

On Tue, Mar 17, 2009 at 4:51 PM, Jacob Appelbaum <jacob at appelbaum.net>wrote:

> Jonathan Foote wrote:
> > Judging by some of this traffic, I think people have misconsceptions
> > regarding EEGs and would do well to read up on it a bit. EEGs are not
> > a "brain jack." You are not going to control anything reliably, and
> > even detecting anything is hard. (Yes, you've seen the monkey move the
> > robot arm video. That's not EEG: that measured specific neural regions
> > using electrodes inserted through a hole in the skull.)
> >
>
> I'm sure there are some misconceptions about the limitations of EEG
> hardware. I'm also pretty sure that people are expressing that they want
> brain output of *any* sort. To play with the data that is easily
> available and not to build some kind of commercial whiz-bang
> application. I personally want to explore and to experiment. I'm not
> expecting to control gdb or vim with a headset. I like the idea of the
> Emotiv headset because it seems to be industrially designed for many
> heads and not just one. It sure seems to beat a flimsy headband too.
>
> > At best, you can measure the energy in the various bands (alpha, beta,
> > etc.) with pro EEG gear and (lots of) good electrodes. How and whether
> > this correlates with particular mental activities or emotions is still
> > an open research question. With practice, you can likely change your
> > alpha/beta ratio but it's going to take the better part of a minute if
> > you can do it at all.  And noise and muscle movements (e.g. blinking)
> > will give you 10x the signal of anything going on in the brain.
> >
>
> That sounds great for the authentication system I'm vaguely interested
> in building. It sounds pretty solid for some kinds of music related
> interaction too.
>
> > I don't mean to say that this isn't an excellent thing to be
> > interested in or to hack on, just that it's much easier to get your
> > expectations up than it is to get anything actually working -- see
> > Emotiv as an example of hype vs. reality.
> >
>
> I totally see where you're coming from. However, I think there's no need
> to be a wet blanket about excitement from the group. We'll all learn how
> we're wrong ten ways from Sunday soon enough. That's part of the
> excitement of learning and growing while undertaking such a project. :-)
>
> > Art projects using the EEG like Kal's robot or the Monochrom drink
> > machine are cool and fun, but they basically use EEG as a glorified
> > random noise source. It would be excellent to do something better.
>
> I think those machines are interesting. They're limited but fun.
> Specifically they seem like a good quick inspirational step on the way
> to something better. I currently cannot serve cocktails with a robot by
> thinking about my desired drink. That seems like a fun first project and
> a personal shortcoming. I admit though, it could just be that soft spot
> in my heart for robots and cocktails. :-)
>
> I think many people (myself included!) interested in EEG stuff probably
> don't have a good handle on all of involved concepts. Nor does everyone
> have a way to capture the data that needs to be acquired, nor do we all
> have the circuits involved in that acquisition and I'm certain that we
> haven't all done the programming for processing that data. A simple
> project may serve as a good peer learning project where we can cross
> pollinate. It's all part of some kind of boot strapping process that
> will take us to bigger things. Who knows where and what?
>
> Perhaps as you seem to know lots more than many involved, you can help?
>
> Perhaps give a workshop on the general overview of EEG stuff? I'll show
> up and I'm surely not the only one!
>
> Best,
> Jake
> _______________________________________________
> Noisebridge-discuss mailing list
> Noisebridge-discuss at lists.noisebridge.net
> https://www.noisebridge.net/mailman/listinfo/noisebridge-discuss
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.noisebridge.net/pipermail/noisebridge-discuss/attachments/20090317/00589d6b/attachment-0003.html>