>I have a question that's been burning in my heart, though: consider the
>idea of recording the signal from a sensory nerve (say, visual or
>auditory) in analog form, and then playing it back -- either immediately,
>restimulating the same patient, or even storing the signal and applying it
>to other patients. Does the technology exist for accurate detection of
>the signals generated in a neural pathway as a result of a stimulus, and
>for equally accurate restimulation there? Basically, how much information
>proceeds down one of these pathways and what level of accuracy would be
>necessary to recreate it (say, number of probes and amount of data storage
>per probe) to make the recreated sensory experience seem "real".
I'm not sure about the possible artificial reality implications of
this, but there are devices known as cochlear implants, which are
small arrays of stimulating electrodes (obviously) implanted in the
cochlea of totally deaf individuals, which provide stimulation based
on signals (i don't remember how they are transmitted) from a
microphone. People implanted with these devices can understand speech,
etc, and do remarkably well even in reasonably noisy environments.
They don't work, for fairly obvious reasons, in the congenitally deaf,
or those with central auditory disfunction. Up to now, while the
technology was still fairly experimental, they have only been
implanted in totally deaf ears (the implantation destroys any residual
hearing), and (I'm *not* sure about this) usually unilaterally. That
may change soon.
Unfortunately, I am not terribly familiar with the signal processing
or auditory psychophysics associated with these devices and patients,
but there may be a fair bit of stuff in the literature.
--diana smetters
dks at ai.mit.edu
