Interfacing the Optic Nerve
0428870 at tucson.princeton.edu
Mon Jul 4 23:20:11 EST 1994
I am thinking of an experiment to measure electrical impulses in optic
neurons and to correlate them to visual stimuli using computational
neural nets. e.g. a white field of vision produces a particular pattern
of impulses, a green field of vision produces another, bright, dark, objects,
etc, with the idea of eventually building something that could tap, or read,
the optic nerve and SAY something about what it means, or build some
vague representation of what is in the field of vision that produces those
THEN, I want to try bypassing the biological interface (e.g. the eye, or
whatever more primitive sensor) and artifically jacking in impulses with
the hope of producing the 'vision' in whatever neural system normally
interprets those electrical impulses (e.g. the brain).
I want to focus on the first part of the experiment for now, e.g. the
reading of optic nerves, and the attempt to get computational neural nets
to identify the characteristics of impulse patterns in the neurons, and
hopefully to correlate them to distinguishable images, or types of images.
Does anyone know:
(a) current work related to this?
(b) a good subject to use -- e.g. an organism that uses simple
neural framework for vision, and uses visual stimuli to survive
in an obvious way. For example, a simple light/dark perception
ability might be a good place to start.
(c) the mechanics of
(i) measuring neural currents at the microscopic level,
in the trunk or in and around the optic nerve, without
damaging the organism at hand, in real time, etc.
(ii) how one might possibly INJECT electrical current
into a functioning neural framework, reliably, measurably, and
without critically damaging the organism at hand?
I am really just looking for a good subject organism at this point.
I would appreciate any ideas, criticism, guffaws, whatever.
More information about the Neur-sci