[ExI] Meaningless Symbols.

Eric Messick eric at m056832107.syzygy.com
Wed Jan 13 20:35:37 UTC 2010


Gordon asks:
>
>The Englishman stands naked in a field. He represents the entire
> system. He and his neurons (trillions upon trillions upon trillions
> of them if you like) process Chinese symbols according to the
> *syntactic rules specified in a program* which he and his neurons
> have memorized. Show me who or what understands the meanings of the
> symbols.

That's easy: the data state of the system is where the understanding
is.  Syntactic processing involves keeping machine state.  In this
case, that state might represent interconnections between neurons, and
the strengths of those interconnections.  Those connections and
strengths change over time, and are data which can be syntactically
manipulated to model those changes.

The changes represent learning.

The semantics is learned based on experience.

The semantics is encoded in that data.

If you take that same data representing a system which understands
something and use it to drive another computational process based on a
different substrate, the resulting system will still understand the
same things.

If neural interconnections in a human brain have achieved some
understanding, we can (theoretically) extract that understanding and
move it to another substrate, like computationally based neurons.

Oh, and symbol grounding is learned based on interactions with
external entities.  Why is this such a mystery?

-eric



More information about the extropy-chat mailing list