[extropy-chat] Can a single brain cell recognize Bill Clinton?

[Neil Halelamien]

On 6/30/05, Terry W. Colvin <fortean1 at mindspring.com> wrote:
> (1) How can single, ostensibly rather simple, neurons process the flood
> of bits
> arriving from subject's optical system?

These neurons are at or near the top of the visual hierarchy. If I
recall correctly, it's something like retina -> LGN -> V1 -> V2 -> IT
-> stuff -> MTL (where these neurons were found). As you go up the
hierarchy, neuron responses get more invariant to things like pose and
illumination changes, while becoming more specific to particular
objects.

> (2) Can Darwinian evolution account for single-cell pattern recognition?  Of
> course it can; it *must*!

Sure. Highly sparse and invariant representations are easier to use as
inputs to other learning systems.

> (3) Has the data stream from the subject's optical system been preprocessed
> by the optical system itself, leaving little for the neurons to do?

I wouldn't necessarily say that it leaves "little" for the neurons to
do, but yes, the information has been preprocessed by earlier visual
layers.

> (4) Are the Bill Clinton cells only the output terminals of holographic
> (whole-brain) image processing.

I'm fairly certain they serve as inputs to other areas, particularly
memory-related regions.