[ExI] Brainport

[Bryan Bishop]

On Tue, Jun 2, 2009 at 7:47 AM, Nathan McCorkle <nmz787 at gmail.com> wrote:
> I only mentioned a clock cycle because I don't know how our vision system
> works... basically I figure if you could control the impulses at each
> individual "pixel" nerve ending (I know photoreceptors combine action
> potentials with each other so maybe the binding site would be up-nerve of
> the photoreceptors).
>
> The end-goal --- direct video signal transduction into the head, eliminating
> conversion to light (which requires a lens, a diffuser, a projector, etc...)

On that note, there was some research in 1999 by a guy now at Stanford
or Harvard who used giant electrodes directly (mechanically)
interfaced to the back of the cat visual cortex to grab videos from
the neurons. The cat was anesthetized, and the eyes were focused on a
television screen in a black box (to keep outside light, out). From
the visual cortex signals, they were able to recover the video images
with a lot of fuzziness that they cleaned up with some bayesian
filters. So, at least reading can be done. More recently there was a
study doing something like this except using fMRI to identify which
shape an observer looked at, and last year there was a paper where
fMRI was used to predict which word an individual was reading (the
news picked this up and called it "mind reading").

Mapping of the human visual cortex using image-guided fMRI
http://heybryan.org/books/papers/Mapping%20of%20the%20human%20visual%20cortex%20using%20image-guided%20transcranial%20magnetic%20stimulation.pdf

Visual image reconstruction from human brain activity using a
combination of multiscale local image decoders
http://heybryan.org/books/papers/Visual%20image%20reconstruction%20from%20human%20brain%20activity%20using%20a%20combination%20of%20multiscale%20local%20image%20decoders.pdf

I don't think I have a copy of the language-prediction fMRI study, nor
a copy of the 1999 study with the cat. Oops. Anyway, one other item of
interest might be repetitive transcranial magnetic stimulation studies
of the visual cortex and geometric hallucinations that arise as a
consequence, which could be used to inform you on the nature or
structure or wiring of the visual cortex. I suspect that's what the
image-based fMRI studies for mapping of the visual cortex is all
about. With big bulky coils for rTMS, it's harder to get that high
resolution that you're looking for.

> I'm thinking that a molecule which could be induced to resonate through the
> skin could be used to stimulate each individual "pixel" nerve ending. If the
> molecule is cleared from the body quickly (within an hour) I think I would
> definitely be a user... take a puff of metallic drugs hook into youtube or
> google earth... I think it would be a hit.

How would you make sure that those signals get routed to the visual
cortex without inducing (or engineering) something like a
tactile-to-visual synesthesia?

- Bryan
http://heybryan.org/
1 512 203 0507