Hacker News Reader

Comment by mrshadowgoose | Hacker News Reader

That argument is utterly unconvincing, as you've conveniently left out the millions of years of "training" and "fine tuning" that has occurred through the evolutionary process.

View on HN

Which implies the brain has more than a structural bias for language; ie language and/or knowledge is somehow passed genetically.

There's some reasons to suspect this is at least partially true, but to what extent is unknown and contraversial.

It could be argued based on the extreme difficulty we're having in teaching animals human language, even primates. We can create a semiotic system and get a chimpanzee to communicate through it, but it won't learn English no matter how hard we try.

I think that's generally accepted to be that humans have a very large language center in the brain; IE: and inductive bias towards learning language (and I think there is something about our mouth and tongue being able to make specific sounds?)

It doesn't say anything about that knowledge being precoded.

Isn't a language center by itself precoded human toward the acquisition of human language (which has co-evolved to make use of our language center)? I don't understand how it could be any other way.

Exactly. But it is unclear if there is intrinsic language knowledge somehow encoded too (which is what the OP is implying: https://news.ycombinator.com/item?id=35590091)

Your argument is utterly unconvincing too, a human left alone with no parents or way of learning can basically do nothing at all, i.e. what we start with isn't some trained neural net at all, it's an empty framework with no data. The OP is correct in that we are several orders of magnitude more efficient at learning than ANNs.

Empty framework my ass.

The fact that human brains are capable of learning things that animal brains cannot, is a clear indicator that it is already "trained" to a significant degree.

Human "learning" seems to be much more analogous to "fine tuning and memory storage/retrieval", than actual training.

Why can't a macaque teach a calculus class? Why can't its "empty" brain be taught to do something like that?

Try raising a child with no inputs and see what it ends up doing. We don’t start “trained” at all. We just start with a neutral network architecture with no data. The difference between us and other animals is the number of neurons and the network complexity, but it doesn’t start trained to any degree. Humans do not start with any ability to talk, reason, identify objects etc.

Reply on news.ycombinator.com