From: doug@eris.berkeley.edu (Doug Merritt)
Subject: Ultimate input hardware
Date: Wed, 6 Feb 91 18:33:30 GMT
Message-ID: <1991Feb6.183330.8154@agate.berkeley.edu>
Organization: University of California, Berkeley



Has anyone been tracking recent research into EEG (or SQUID) analysis
that might be used as an effective computer input interface someday?

I'm aware that some crude work has been done long since on e.g. picking
the letter of the alphabet currently underneath the endlessly moving
cursor by biofeedback of alpha waves.

But what about higher bandwidth or more naturally targetted input?
I recall reading in the late sixties or early seventies that there
was an evoked potential (P40 or something) that had been determined
to be naturally produced whenever someone made a clear cut decision
to act. The example given was of a subject deciding to turn on a lamp.
They could reliably sense the potential and turn on the lamp in the
split second before the subject started to move his arm.

(They further speculated that it might be used someday as a fire control
in a fighter jet during high-gee turns, a concept picked up crudely in the
movie Firefox.  And one that the Air Force may actually use, for all I know.)

In virtual reality terms, this particular potential might be used in
conjunction with eye tracking in order to mentally command a selection,
or "move me there", or any similar single action.

The Feb. Scientific American article by Freeman about strange attractors,
and unique EEG amplitude maps in response to recognition of sensory
perceptions, makes me wonder if it might now be possible to sense a
broad enough array of natural data from the brain to use as an effective
input device. For example, if we could deduce from an EEG of the motor
cortex that the user was visualizing a clenched fist, that could be
used as a "grasp object" command, without even needing a data glove.

Even if things are a little cruder than that, I would think that there
are, at least, a variety of readily identifiable evoked potentials that
have been studied that might be usable for some useful set of commands,
even if it still needed a data glove and eye tracking to back it up.

Anyone know about this? Or know of a summary/survey source on evoked
potentials and EEG analysis that sums up recent research?

A similarly interesting possibility is that of using higher-temperature
superconductors to make SQUID's that could monitor magnetic fields
on the *interior* of the brain, potentially yielding even more information.
(EEG's pick up only surface currents, limiting the portions of the
brain that might be analyzed even in principle.) Any new info there?
        Doug
--
        Doug Merritt            doug@eris.berkeley.edu (ucbvax!eris!doug)
                        or      uunet.uu.net!crossck!dougm