The problem is here: " (b) forms concepts (i.e., signs), ". This is absolutely not right. Concepts are big, dynamic, interacting processes. Signs, at best, can serve to reactivate concepts. One of my issues with representation is this feature of signs, that they miss the point of what is going on.
There also turns out to be a learning or inductive (call it information theoretic if you want to be fancy) problem with signs, in that you can learn which sign to choose in a situation, but it is never possible to finally and completely delineate the precise list of all things that a sign can and can't refer to. So the notion that signs "point" to meanings turns out to be fundamentally flawed. It's a weird result that I've only seen from some linguists at Stanford (I really need to dig up that paper), but it has really put the hook in me as to there really being fundamental flaws with what I'm calling the representational approaches. At some point, all symbols have to get back to the dynamic concepts. The "grounding problem" again. You might get something out of manipulating signs, but then what? But I say you can detect what signs to use. This is a pattern matching problem. The processing to combine those together is more complicated, but just picking signs--it's a basic brain operation. Many, non-human animals can do it. Parrots are super at it. Ain't language. andi On Apr 6, 2013, at 8:07 PM, Piaget Modeler <[email protected]> wrote: > > Symbolic interactionism looks interesting. > > So, the beginnings of the AGI's semiosis components: > > `1. Decoder component > > (a) receives percepts, > (b) forms concepts (i.e., signs), > (c) matches or constructs ???? what ???? > > > 2. Encoder component > > (a) starts with a semantic net > (b) decides upon a subset of the semantic net to encode > (c) chooses signs related to the concepts in the subset > (d) forms a plan to utter the signs > (e) executes the plan > > Is this enough. What is the name of 1.(c) ? > > Kindly advise. > > ~PM > AGI | Archives | Modify Your Subscription ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
