10/26/2025

I'd recently thought a lot about the idea of a mind space, i.e., a mathematical representation of every thing we can experience and name.

Main idea

I have already presented in Sustantivos, adjetivos y conjuntos (in Spanish) my theory of interaction between human language, knowledge and experience. Essentially, I hold that there is a space $E$, which I'll refer to as mind space, which represents every "thing" we can experience, for example, thinking about it. Thus, language is a deictic mean to analytically refer to various subsets of the space $E$. I use the word "analytical" as this space allows the notion of "word taxonomies", where each noun, such as "moon", can be understood as the intersection of several adjectives, such as "aereal", "bright-over-dark", "round". Here, we think of "round" as a word pointing to everything round we can think of. Similarly, we can go further and consider smaller subsets which do not correspond to any existing word, e.g., "the moon over the horizon, which because of an optical illusion seems bigger".

I like to think in this space as the encyclopedia of everything that exists. Therefore, it provides a bigger insight of how humans think that regular dictionaries. Obviously, mapping the whole mind space is an almost impossible task. Some approximations to this mapping could be encyclopedias, or even Wikipedia. Relating to word embeddings, I'm curious as if there's any kind of embeddings with this nested sets structure, and, in that case, how an embedding of the encyclopedia of everything that exists would be like.