Re:Brain Maps Out Ideas and Memories Like Spaces
by beaconstudios on Friday January 18, 2019 @08:19AM
(#18939074)
I think concept spaces acquire a "shape" when we become familiar enough (I associate this with the idea of a subject "clicking"), and that creativity often involves realising the shape of different concept spaces are similar and thus ideas from one can apply to another. Furthermore, I believe these "shapes" exist as a metaphysical "truth", akin to platonic forms. I think it ties into how real-world problems can often be reduced to pure mathematical problems. I experience ideasthesia very strongly (often having strong, immediate instinctual reactions to e.g. how well new software features fit within the "shape" of the overall domain as well as having affinity or aversion to concepts for reasons I have to deduce later).
I'm not sure exactly what I'm trying to say, beyond "I think so, yeah". It's a super interesting area of exploration and either it hasn't been looked into enough, or I don't know the right words to look for. I can feel a thread linking concepts like mental models, systems theory, cybernetics, Christopher Alexander's work on pattern languages, and so on.
Re:Brain Maps Out Ideas and Memories Like Spaces
by thecupisblue on Friday January 18, 2019 @12:59PM
(#18941231)
You're on a good path, abstractions and concepts overlap a lot between seemingly different areas of reality.
Re:Brain Maps Out Ideas and Memories Like Spaces
by beaconstudios on Friday January 18, 2019 @01:53PM
(#18941797)
Yeah they do, and I'm interested in what we can do with that. I believe that accessible modelling could change the world in a lot of fundamental ways.
Re:Brain Maps Out Ideas and Memories Like Spaces
by pizza on Friday January 18, 2019 @09:00AM
(#18939310)
Could be like a fourier transform of a semantic concept (pointer) (multi-)graph.. And as a behavior loop, attention/perception = solving some class of sparse problem in a very high dimensional space..
So... I wonder then if there is duality in this mapping like between frequency/time domains with a Fourier transform of signal over time, just, instead, over a concept domain
<->