🚀 go-pugleaf

>[Please note follow-ups]

In article <343D4045.A19012D3@qi3.com>, Steven Packham
<packhste@qi3.com> wrote:

>How revolutionary is this idea

  The revolution comes partially from the simplicity of the thesis.
It says that intelligence and learning are based on the temporal
relationships that exist between events and nothing else.  Although
work on temporal/causal networks are under way on several fronts, we
are postulating that this what is learning and intelligence are about.
Exclusively.  We believe that everything in intelligence emerges from
temporal relationships.  What are the temporal relationships one might
ask?  We have identified three:  before, after and contiguous.  So
what is needed at the fundamental level are simple processors whose
job it is to discover and manipulate these relationships.

> since we allready
>have know about the physical mapping of the world
>outside to a model inside our brain.
>
>speciffically area's 17 18 and 19 of striata
>cortex as a starter. Even in area 17 down to
>edge detection units.

  Even if this were true, I don't see any sharp relevance to temporal
neurons.  Please explain.

>obviously neighboring cells must deal exactly with
>contrast when dealing with vision, and vision
>is made up of patterns of contrast, contrast
>being most vivid in lines and edges, but used
>in a very analog sense in the brain where not
>just lines but contrast in shading and degrees
>of differing light give us our model of self.
>We are what we see, or sense for the blind.

  I partially agree with the above.  I would qualify it by saying that
we don't see contrast but changes in contrast.  It is known that our
eyes move in short saccades.  I think that this is so in order to
create the necessary temporal differences needed for neural
perception.  IOW, our neurons are driven by changes or events.
  I disagree that the nervous system processes analog signals, if that
is what you meant.  So called analog signals must be, at one point or
another, converted to discrete changes.  Preferably at an early
processing stage.  A change in a signal is an event.  A learning
system can use these discrete events to create its temporal order
expectations.

>I take it this must be a book.
>
>The Temporal Neuron Revolution? 

  No.  We don't intend to write one but you are welcome to work on it.
We are trying to bring attention to an aspect of AI research that we
think will lead to rapid progress and profound changes in the way we
view intelligence and the CNS.  Thanks for the interest.

Regards,

Louis Savain

President, Marengo Media, Inc.