On Mon, Mar 14, 2022 at 9:18 PM Ben Goertzel <[email protected]> wrote:
> ... > Well I am working pragmatically with the notion that the meaning of > concept C to mind M is the set of patterns associated with C in M. I like your pattern based conception of meaning. Always have. It's a great improvement on symbolic meaning. But I came to conclude that patterns don't tell the whole story. The fuller story is that patterns can grow. Glider guns, and all that. In comparison with you, then, you might say I've moved to a notion that: The meaning of concept C to mind M is... the grouping a set of patterns in a certain way, which you can choose to call C if you wish, but it doesn't necessarily need a label, because you can generate it at will, and the exact set you get may differ according to the context M finds itself in. (That "way" of grouping typically in linguistics has been according to shared contexts. Transformers work basically the same way. But they try to label all the "C" too.) One way to see the contrast between the static set model and the actively grouping set model is to that made by Romain Brette, who contrasts "representational" vs. "structural" models. Roughly "representational" models are sets, while "structural" models are assembled according to grouping principles: http://romainbrette.fr/perceptual-invariants-representational-vs-structural-theories/ Brette's contrast is also in line with some views of the Active Inference line of thought, for whatever that's worth, e.g. Maxwell Ramstead: "The traditional view of cognitive processing ... is that the brain is essentially an aggregative, bottom up, feature detector ... the more sensory areas which are supposed to be lower down on the processing hierarchy ... essentially detect features ... " "What we think is that this is maybe just the wrong way to look at the problem. ... The predictive processing view or the active inference view flips this on its head. And says OK well maybe this top down thing is what the brain is mainly engaged in." http://www.youtube.com/watch?v=WzFQzFZiwzk&t=17m58s There are other versions of this approach elsewhere. How many do you want? Encapsulate Goedel incompleteness by creating an entirely new sense for logic and negation if you like. Sounds like a basis transformation to me. Sounds to me similar to the way Category Theory embraces Goedel's theorem by building a basis for mathematics in variability. Which is the line I see Coecke taking. So Coecke's "Togetherness" might equate to your "paraconsistent logic". Ought to be any number of ways you can formalize them top-down. But if the sets are generated, it will always be easier to generate them bottom up. We're already doing it. We just need to stop being surprised when we get an infinite number, and that they contradict. ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T0f3dcf7070b3a18e-Mb6da0dcf0b676d2eb2588841 Delivery options: https://agi.topicbox.com/groups/agi/subscription
