I recall that they trained a part of an insect's head like the nasal passage and it was able to be used for language better than the model at that time chatGPT2. So there's something innate in nature that can learn human languages.

Can you find that study?

Took me a few hours but it was fascinating. I couldn't find it again easily.

https://www.discovermagazine.com/the-sciences/fruit-fly-brai...

https://oadoi.org/10.1371/journal.pgen.1007430

Cool, and thanks for search!

I think you may have slightly over-sold the study, though -- or at least what you remembered from it.

My reading of the first study was that they took a simulated version of a relatively large (2000 node) neutral network that makes up part of a fruitfly brain, and were able to do standard neutral network training on it to do some language prediction.

I'm not sure that this says anything about fruitfly noses being wired for language though. I expect that they could have taken that same simulated architecture and trained it to do anything that regular neutral networks could do -- detect faces, make stock market predictions, play a (poor) game of Go, or learn a homeowner's thermostat patterns.

I think it's just more a statement about the power of neutral networks in general.

I did misremember, it seems they used a network of algorithms based on how flies use 2000 neurons and the original article I read it from a while ago may have oversold it. The fact it was able to predict well is them looking for inspiration in insects, and that they'll be using more insect inspired behavior.

> It sounds like the researchers are saying there's something special about learning spoken language. But it seems to me that there can't have been many cases similar to his.

We've learned a lot from people who have received cochlear implants at different ages. Earlier implantation is strongly associated with functional spoken language use and fluent speech. There's a big benefit before age 5; a large proportion of those implanted before 24 months basically have normal language skills, while few after age 5 ever fully "catch up."

edit: Here's a study of prelingually deafened adult outcomes with CIs https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5720870/ All of those studied had acquired spoken language before implantation and had some degree of effective hearing earlier in life, so were not fully deafened before the language acquisition window.

The implants provide an improvement of quality of life but do not allow most of even this population to e.g. understand spoken language on TV without subtitles.

Listening to TV conversations is a high bar. In conversations people can ask for clarification or to slow down etc.

Which explains: Before implantation, 7% of the patients were able to have telephone conversations. vs After implantation, 60% of patients are able to have telephone conversations.

Also, the technology dramatically improved over time so we don’t have long term data on high quality implants.

Still, pay attention to context. This was people who were successfully using hearing aids and oral language before implantation. Even in this subpopulation, they did not do nearly as well as children do, even though this subpopulation was less deafened than most deaf children.

> Also, the technology dramatically improved over time so we don’t have long term data on high quality implants.

We do have enough series to know that 5 year olds receiving treatment have (on average) significantly worse outcomes than 18-24mos.

For the timescales we have tested there’s significant differences, and in terms of quality of life it’s clear early intervention is a significant benefit.

However, slower adaptation isn’t zero adaptation. The limits for people implanted at 5 when they are 50 is still an open and IMO interesting question.

> The limits for people implanted at 5 when they are 50 is still an open and IMO interesting question.

Slower development usually means a lower plateau, and I think we pretty much have to assume as such (and can be prepared to be pleasantly surprised).

Else, we get to wishful thinking: older people on older devices developed more slowly and plateaued at a lesser value of hearing. Now, we have implanted older people on newer devices, and they're developing more slowly, but hey, maybe they'll eventually develop fully normal hearing.

Do you have a source for that plateau? I’ve read that early success predicts future success on age adjusted tests. But the children were still improving in absolute terms over a decade post surgery.

I was arguing mostly from the standpoint of delays in development are almost always correlated with lower ultimate attainment, no matter what measure you're looking at.

But if you want cochlear implant specific data, here-- https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10760633/

You're right that time narrows the gap between early implantation and later implantation, but the slope of that narrowing is pretty small by the 20 year mark (and barely statistically significant in this moderately-sized study) and the gap is relatively big.

The difference of time of implantation between the two groups was relatively small (mean implantation at 45 months vs. 34 months) and produces a gap that's durable for decades. >130 months is way, way, out from 45 months.

Very interesting paper!

Still, lots of work to do, to quote "However, it was not possible to control other factors, such as the socio-economic environment of the participants.".

In my view this could affect the study quite a lot (or not, but unknown for now). They mention the initial intervention was 3 months on-site, but after "the patient returned to his area of residence, where he/she would have speech therapy and special education".

Thank you, interesting read.

Looks like scores are still improving at 25 vs 20 years. Though excluding tests which maxing out the gaps never closed.

I was surprised at the amount of continuing improvement, too.

Never fully catching up sounds a lot different than "window for learning spoken language is permanently shut". Am I missing something?

"Permanently shut" for everyone is probably an exaggeration.

A better description "very, very few of those [with hearing restored] after age 5, who had never had any hearing before, develop anything close to normal spoken language skills."