Interesting research that suggests correlation between neural activities related to language processing and predictive language models (but not models that are optimized for other language tasks).
According to Evelina Fedorenko, one of the authors:
“… [this research] suggests that perhaps optimizing for predictive linguistic representations is a shared objective of both biological and artificial language models.”
The neural architecture of language: Integrative modeling converges on predictive processing
Thank you for posting this paper, unfortunately we have read too many of these in the past with much excitement and yet no results.
The first line of the paper reads:
"Language is a quintessentially human ability. "
This is completely false, I am shocked reading these words in PNAS?! We are kept pushed back to Galileo’s time where human being on earth was considered the centre of cosmos! These absolute views in science by now have fortunately proven to be false one after the other by science itself And yet they persist!
There is plethora of non-Human and non-Neuronal organisms with exceedingly complex languages, in ways similar to human languages and in some cases perhaps more advanced.
Best example to recall:
These trees have no known brains similar to ours and yet have memory, learning and yes communications. Where there is communication there is language. Because the tokens of the language e.g. molecules (see the next article) are in need of interpretation(Semantics?) in order for the organism to function to remain extant.
“We also assumed that they lacked the ability to communicate, a crucial function for organizing group activities. Our view has changed. Bacteria can organize into groups, they can communicate, and these abilities are important factors in the development of many diseases.”
Basically Ants possess symbolic discrete language(s) for communications and yes arithmetic and yes! even compression algorithms to reduce tapping and other energy consuming tokens of their language.
This not a debate about a PNAS paper which I couldn’t care less about, this reply is hoping for a leap of imagination for those of us who are badly limited by our own college education and the current human-supremacist neuronal views of human brain and language.
Currently we have very little to go to understand the human language in any of its forms.