Brute association—or “babbling”

by Suspended Reason

At the heart of snobbishness is a curation of associations—the idea of pathogen contamination metaphorically applied to the culture sphere.

Previously, I’ve connected the cargocult concept to William James’s idea of brute association.

Here, I want to entertain the idea that language itself is merely an adaptation on top of a built-in machinery for instituting associative bonds” (Katz & Chomsky 1975).

On the one hand, there is a way in which this may necessarily be the case: to connect sign and signified, we need some sort of associative hardware.

On the other hand, it may not be sufficient. Modern text transformers and neural nets run on pure association, and they don’t make a ton of sense. Robin Hanson notes that writing produced by GPT-2 looks a lot like a C-level college paper on a complex subject. He calls this sort of writing babbling.”

while I usually try to teach a deep structure of concepts, what the median student actually learns seems to mostly be a set of low order correlations. They know what words to use, which words tend to go together, which combinations tend to have positive associations, and so on.

Let me call styles of talking (or music, etc.) that rely mostly on low order correlations babbling”. Babbling isn’t meaningless, but to ignorant audiences it often appears to be based on a deeper understanding than is actually the case.

Which prompts the question: What are higher order” correlations? What is deep” structure, and what separates it from superficial” structure? The more statistically minded among us may be needed to formulate an answer. Causality is one approach to an answer, but is it sufficient, and how can it show up in data in the first place?