Or so say D. E. Dijksterhuis, M. W. Self, J. K. Possel, et al., in their “Pronouns reactivate conceptual representations in human hippocampal neurons” (Science 385.6716 [26 Sep 2024]:1478-1484; DOI: 10.1126/science.adr2813). I don’t have access to the full article, but here’s the Editor’s summary:
Languages use pronouns to refer to nouns or concepts that were introduced earlier in a conversation. Do these pronouns activate the same neuronal representations in the brain as the previously introduced words? Using human intracranial recordings, Dijksterhuis et al. found that during reading, single cells in the medial temporal lobes that respond selectively to specific individuals also respond to pronouns that later in sentences refer to previously read nouns. These results indicate how memory and language are linked at the single-cell level. —Peter Stern
The abstract is available at the link. Interesting, if true!
Goodness! The brain is involved in thought!
I can’t read the paper either, but “specific individuals” suggests some confusion about grammar on the part of the writers, or perhaps on the part of the editor.
Specifically, that they suppose pronoun anaphora is basically the same process as referring to the proper name of a person. Judging (possibly wrongly) by the exact words of the summary, all that they have shown (if they have shown it) is that a pronoun referring to a specific person known to the reader “activates” the same part of the brain as the person’s name.
This would only say something about actual grammar if, when reading
“A cat sat on the mat. It was purple.”
your electric phrenology (or whatever: here they seem to be actually sticking things into people’s brains) zoomed in on the same area for “cat” and “it.” (Or, what would be really interesting, the same area as “mat”, and subsequent enquiry showed that that was how the reader had interpreted the sentence.)
Possibly they did something just like that, and the summary is just careless.
But if not, they have not demonstrated anything about language at all: only that certain specific bits of your brain light up (apparently) when thinking of a particular person. (Quite romantic, really …)
Preprint.
Quotable: “Shrek and Courtney Love walked into a bar. He/She sat at a table.”
“A cat sat on the mat. It was purple.”
“A cat shat on the mat. It was purple” is for me much less ambiguous. Here the color is attributed to the catshit.
Thanks, Y.
EDIT: Apparently this is about pictures of celebrities, so it applies only to proper nouns (so far).
@Stu:
Truly, pronoun reference is a complex matter.
@Y:
Thanks!
They seem to have done exactly what I suspected. They are not actually investigating language at all, much less grammar. At most, perhaps, “mentalese.”
The references are all to neurobiology or computational linguistics. Nothing on syntax (though, to be fair, their methodology makes all that irrelevant anyway.)
They are not actually investigating language at all
Oh hell. Not that I’m surprised in the least.
I think so. But studying what and where happens in your brain as you’re processign speech still contributes into understading of speech even if the specific part chosen for study has more to do with referent concepts/the signified.
You’ll need this information.
True; I suppose that it is a “linguistic” point (on some level) that a third-person pronoun used anaphorically may summon up the same activity in the brain as its antecedent did shortly before.
I don’t think many linguists would take this as a epoch-making discovery, however.
I think the authors (who seem to be innocent of any specialist syntactic knowledge) are presenting their technique as a sort of proof of concept, that might lead to actual fruitful discoveries. But I think they’ve fallen prey to a confusion of levels: words refer (in some way) to the real world, but what their method is investigating is the physical correlates in the brain of thinking about the referents, not the words themselves. I don’t think this approach can even in principle shed any light on the $64,000 question of just how language can refer to the extralinguistic world.*
(Of course, in a world in which many apparently sensible people suppose that APEs/LLMs are “intelligent”, despite having no actual means of referring to the physical world except via plagiarism, many people are somewhat confused on these points.)
* Apart from anything else, this is a complete non-starter if you try to divorce language from its social and cultural context. (The latest Hiphilangsci podcast makes an interesting, if not altogether convincing, case that Chomsky was driven to separate language from social interaction as he does by his bad conscience about his linguistic work being funded by the military.)
You saw noone? And at that distance. You must have excitable neurons.