Re: AIs on drugs
- From: casey <jgkjcasey@xxxxxxxxxxxx>
- Date: Fri, 11 Apr 2008 14:50:00 -0700 (PDT)
On Apr 12, 3:26 am, c...@xxxxxxxx (Curt Welch) wrote:
On this issue of the person seeing freeze-frames when
filling the glass of water.
It just occurred to me why it might be described that way.
The brain tends to lock on to an interpretation and tries
to stick with. Whatever the brain believes it sees, is
what you continue to to see until enough evidence makes it
clear it's something else. So even if the brain can
internally represent the concept of the water in the glass
being at 100 different levels, when it sees the water at
one level, it will tend to lock onto that interpretation
and not let it change until it's very clear the
interpretation is wrong. This could keep the brain from
seeing it move from level to level like from 50%, to 51%,
to 52% and make it jump from 50% to 60% to 70%.
For a person with a working motion detection section of
the visual system, the brain would be wired up to use
that as the driving force for changing interpretation
quickly from 50, to 51, to 52, and not "sticking at" the
50 value. Without the motion detection signal active,
the 50 detector will try to "hold on" to the current
interpretation until it was very obvious it was wrong
causing it to jump.
My theory is motion is added to higher level processes that
really do only occur at the fixed frame rates. Normally they
would have the motion attribute stamped on them giving us
the _illusion_ of seeing smooth motion as we do with eye
saccades as we read text. Motion velocity and direction are
easily and quickly extracted.
This can be simulated using a strobe light that is mostly
off and flashes at some optimum rate that the motion detection
cannot work but the higher levels can use to see objects
changing positions at each flash without the motion data.
JC
.
- References:
- Re: Google AI: well at least it's not not evil
- From: Tim Tyler
- Re: Google AI: well at least it's not not evil
- From: Curt Welch
- Re: Google AI: well at least it's not not evil
- From: Tim Tyler
- AIs on drugs
- From: Tim Tyler
- Re: AIs on drugs
- From: Curt Welch
- Re: AIs on drugs
- From: Tim Tyler
- Re: AIs on drugs
- From: Curt Welch
- Re: AIs on drugs
- From: casey
- Re: AIs on drugs
- From: Curt Welch
- Re: AIs on drugs
- From: casey
- Re: AIs on drugs
- From: Curt Welch
- Re: AIs on drugs
- From: casey
- Re: AIs on drugs
- From: feedbackdroid
- Re: AIs on drugs
- From: casey
- Re: AIs on drugs
- From: feedbackdroid
- Re: AIs on drugs
- From: Curt Welch
- Re: Google AI: well at least it's not not evil
- Prev by Date: Re: Are neural networks worth the pain.
- Next by Date: Re: Fermi paradox
- Previous by thread: Re: AIs on drugs
- Next by thread: Re: AIs on drugs
- Index(es):
Relevant Pages
|