From: diego@minerva.st.dsi.unimi.it (Diego Montefusco)
Subject: Re: TECH-THEORY: Brain-machine interface
Date: Fri, 5 Mar 93 23:04:36 CET


In <9302260219.AA17458@ghost.dsi.unimi.it>, on Feb 25, you wrote:

> corresponded to existing senses?  If one were to graft a large-bandwidth
> bundle of "artificial neurons" (I'm ignoring the massive engineering
> problems here for the sake of argument) onto the cortex, configuring
> some as inputs to neurons, some as responding to the electrical outputs
> of neurons, couldn't we create a *new* mode of communication between
> the human and a computer hooked up to the other end of these artificial
> neurons?  From what I know about the flexibility of the brain, both
> with regards to individual variation and compensatory responses to brain
> damage, it doesn't seem implausible that, with plenty of practice,
> one might learn to understand the new form of input, and acquire precise
> control over the output.  In fact, to me it seems more plausible to
> teach the subject this new form of communication than to try to diagnose
> the incredibly complex workings of the sensory pathways.  In this method,
> the *brain* does all the work!  Of course, during the learning process,
> you'd probably have to provide some sort of redundancy between the new
> "sense" and the already-interpretable ones.

But since the brain is ever-evolving, the learning process should never end.
I don't know much about neural nets, but can you USE them while they are
still learning?
 
Diego

    -----------------------------------------------------------------
   | Diego Montefusco |         Computer Science Department          | 
   |     Via Pirano 4 |         University of Milan - Italy          |
   |     20127 Milano |                                              |
   |           Italia |     P L E A S E   R E P L Y   M E   T O :    |
   |+39 (0)2 27001467 |         montefus@ghost.dsi.unimi.it          |
    -----------------------------------------------------------------