Re: Invariant Recognition, Grandmother Cell, and Memory Hierarchy


"dan michaels" <feedbackdroids@xxxxxxxxx> wrote in message
news:1122759025.323666.322730@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

<snip>
> This filtering and abstraction is probably mainly a feedforward
> phenomenon. Furthermore, it is all dictated by genetics, and
> probably has nothing whatsoever to do with memory, and little to
> do with feedabck.
<snip>
> Then, memory comes into play more and more towards the higher
> levels of the hierarchy. After the signals have gone through
> many levels of filtering and abstraction, they hit the memory
> banks, and produce an output [I won't go into how associative
> memory works], and this output provides the ultimate feedback
> from the highest levels, which goes back down the hierarchical
> chain as "prediction" information.
>
> So, this is it. The hierarchy is anatomical, and is feedforward,
> and is dictated by genetics and *nothing* else. It's a result
> of evolution. The existence of feedback is also anatomical and
> dictated by genetics. It's role is to carry predictions and not
> to create abstractions/invariances. Memory comes into play
> mainly at the higher most levels, and carries the experiential
> information of the organism. Eg, in IT, you have face cells
> and grandmother cells. IT is maybe 8 or 9 or 10 levels up in
> the hierarchy of abstractions. It's high enuf that the memory
> areas directly impinge upon it.
<snip)

Holy Cow Dan, it's really exciting to see you describing this! This is the
stuff of which AI progress is being made. This is it, this is it exactly!
The brain goes from inputting raw sensed input to more abstractness, and
this is first routed through the "static" pathways in the brain. Its job is
to extract the *meaning* from these inputs, whether it be objects in an
image or concepts derived from simpler concepts. There is a part of the
brain that stays the same after birth, or after it becomes fully developed
after birth. The end product of those structures creates the touch points
for memory, which is dynamically created and modified throughout life.

I have to say, it is really exciting to read discussions of this sort that
are on the right track. Expert systems and primitive neural networks
developed long ago are dead ends, meant to be used as examples for further
development. Only AI more closely based on the real architecture and
principles of the brain will bear fruit.

Dan, have you designed and created any AI programs based on this knowledge?
What are your plans?
Gary Frank


.

Relevant Pages

  • Re: Invariant Recognition, Grandmother Cell, and Memory Hierarchy
    ... >> This filtering and abstraction is probably mainly a feedforward ... >> probably has nothing whatsoever to do with memory, ... The hierarchy is anatomical, and is feedforward, ... > this is first routed through the "static" pathways in the brain. ...
    (comp.ai.philosophy)
  • Re: CORBA IDL versioning, evolution, backward compatibility
    ... Real applications need to do things such as adding a field to struct, changing the signature of an operation, adding or removing a parameter of an operation, etc. ... And, quite often, the versioning requires a change in *behavior* of an operation, but not its signature. ... It allows us to program at a level of abstraction that creates the illusion of an unlimited amount of memory, and allows us to ignore things such as physical addresses and how data is mapped onto physical memory pages. ...
    (comp.object.corba)
  • Re: What is different with Python ?
    ... What makes bytes of memory, data busses, ... >and CPUs the right level of abstraction? ... From "counting" you can go ... Also lower than memory and data bus there is ...
    (comp.lang.python)
  • Re: force
    ... You seem to be saying that you think charges, ... |> | Time is an abstraction that relates such anhilation to ...
    (sci.physics.relativity)
  • Re: Relational model versus object model
    ... >> want to boost performance, you often have to introduce redundancy, ... These tables would be located in memory, ... > using OOPLs than it is implemented in an RDB. ... > abstraction so that the same underlying problem space entity can be ...
    (comp.object)