To better understand your point, what's the difference between an idea-variable, real world examples, and learned/derived data. I get the sense there's an unexpressed assumption you're making about the architecture which is key to making the fullest sense of what you're saying.
-- Sent from my Palm Pre On Oct 22, 2012 7:01 AM, Jim Bromer <[email protected]> wrote: Because thought can use ideas that act like variables in some ways, and because ideas can be applied to other ideas and have an effect on them and on how they are used, that means that metaphors can be mapped onto meta rules or meta systems of ideas. (In particular an operation can be applied to the meta systems of a *class* of ideas. It is important for me to note this because I am not just talking about the effects on a meta-system of all ideas or a preconceived subclass of relations. I am referring to a notion of a undefined class of ideas that will be defined by the run time experiences of the program as it reacts to novelty in the IO data environment.) Many people in this group have thought of things like this before so they don't react to my comments about this sort of thing. However, they also do not talk about how their models would enable or react to this kind of thing either. So while this is not totally radical I believe that people have to think about it in a way to program the capabilities of the system into their AGI programs. In particular, many people who are interested in *low* level or bottom up programming of AGI do not feel that they have to write explicit code to enable things like the use of a metaphor that is to be applied to a idea-variable (or a partially defined idea-variable) because as long as their low level plan will allow derived or learned data to be mapped to other derived data then they can leave it at that. They think that when their model is activated by real world examples the details of these special cases will just be there (or something that is not all that far from this.) The idea that I am trying to spread is that this sort of thing must occur at the lower levels of reasoning because it is such a basic idea. Of course you have to be able to figure which idea I was talking about in order to understand what I am trying to say. Jim Bromer On Sun, Oct 21, 2012 at 10:22 PM, [email protected] <[email protected]> wrote: It's not just attributes, but relationships, as well. This is especially true when dealing with a multiple-role metaphor, where each object in a set of interacting objects is mapped to one of another st of objects. I think you hit pretty close to the mark with your attribute transference rule, but I think it's more of a suggestion to be examined and potentially incorporated than a hard assertion. There's also the consideration that it's possible to use multiple projections (or mapping of parts) to relate two concepts when determining their similarity, which can lead to entirely different metaphors connecting the same two concepts. Perhaps it's difficult to identify the projections that make things sufficiently similar, which would explain why being asked to identify what a fog and a cat have in common is difficult to answer until someone spoon feeds you an effective projection, such as the style of movement. I'm sure everyone has experienced the spreading effect of metaphors. Once one concept has been mapped to another, the first's neighbors are easily mapped to the second's. If dating is fishing, dates become fishes and charms become lures, etc. This tells me that the key to a good metaphor is a comparable set of structures related to each other in similar ways for both the source and the target. If I were to do this in a semantic net, I'd be looking for subgraphs which are isomorphic. Finding extensions of such a subgraph that also can be mapped isomorphically without altering the original mapping ought to be a much easier operation than finding the initial mapping. Elements in the source that are excluded from the isomorphic mapping would provide hints as to what candidates ought to be considered for addition to the sink. -- Sent from my Palm Pre On Oct 21, 2012 8:22 PM, Piaget Modeler <[email protected]> wrote: Oops, should be Similar(<source>, <target>) , hence Similar(Cat, Fog). From: [email protected] To: [email protected] Subject: [agi] Similar(Fog, Cat) Date: Sun, 21 Oct 2012 18:07:27 -0700 Bipin Indurkhya in his book Metaphor and Cognition stated that the purpose of metaphor is to transfer attributes from a source concept to a target concept. Before reading any further, do this thought experiment. Name as many relationships as you can between between the fog and a housecat? Think for a moment before reading the next paragraph... .......... . Indurkhya's classic example "The fog came in on little cat feet, waited awhile, then moved on" creates a new concept in one's mind "fog-as-cat" and with this new relationship facilitates the transfer of attributes pertaining to a cat, to the the concept of fog. Now there are a whole host of relationships you can transfer between the two concepts which did not exist before. There is probably a mental inference rule that says IF Similar(x, y) and Attribute(x, a) then THEN assert Attribute(y, a) Perhaps. ~PM.------------------------------------------------------------------------------------------------------------------------------------------------ Confidential - This message is meant solely for the intended recipient. Please do not copy or forward this message without the consent of the sender. If you have received this message in error, please delete the message and notify the sender. AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription AGI | Archives | Modify Your Subscription 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-c97d2393 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-2484a968 Powered by Listbox: http://www.listbox.com
