4 ms·

> There have been cases studies of people with a fully ablated cerebellum, and they can still walk, just a bit unsteadily. ? I had heard the cerebellum is respo

> There have been cases studies of people with a fully ablated cerebellum, and they can still walk, just a bit unsteadily. ? I had heard the cerebellum is responsible for predicting the next token in a sentence (reminiscent of language models) but you're saying language is not impaired without a cerebellum?

kaba0 5y ago
The axons are just as important as the neurons. While neural networks are not good analogues of the real thing, one can imagine that no matter the neurons, without weights between them it would be utterly useless. The cerebrum is chock full of axons.
- SemanticStrengh 5y ago
  > The axons are just as important as the neurons We don't know about that. > without weights between them it We don't know where are located the weights but yes probably. We know for sure that individual neurons and axons do not reduce to a single floating point though. They have action potential memory and many dynamic modulators. Basically I believe the idea that neurons are mostly doing linear regression (which is what neural networks are) is very broken. I have read papers about neural coding and neurons are able to encode information in the spike timing frequency and in the strength (weight) but non of those two mechanisms even begin to explain how the system works. Anyway to answer your comment you have a tentative point but mostly unknown. It is unknown how much percent of human axons are in the cerebrum vs the cerebellum. I would be surprised if the cerebellum would have less since I expect some kind of neuron/axon proportionality. Especially since the main neuron type in the cerebellum are highly branchial.
- andbberger 5y ago
  > The axons are just as important as the neurons. what. an axon is a component of a neuron are you referring to glia?
  - kaba0 5y ago
    No, I meant to highlight that “most neurons are here” may not be the important question to ask, but their interconnectedness which is achieved through axons, and which gives the most volume of the brainz
    - andbberger 5y ago
      indeed the focus of most research is the neuropils

kaba0 5y ago

The axons are just as important as the neurons. While neural networks are not good analogues of the real thing, one can imagine that no matter the neurons, without weights between them it would be utterly useless. The cerebrum is chock full of axons.

SemanticStrengh 5y ago
> The axons are just as important as the neurons We don't know about that. > without weights between them it We don't know where are located the weights but yes probably. We know for sure that individual neurons and axons do not reduce to a single floating point though. They have action potential memory and many dynamic modulators. Basically I believe the idea that neurons are mostly doing linear regression (which is what neural networks are) is very broken. I have read papers about neural coding and neurons are able to encode information in the spike timing frequency and in the strength (weight) but non of those two mechanisms even begin to explain how the system works. Anyway to answer your comment you have a tentative point but mostly unknown. It is unknown how much percent of human axons are in the cerebrum vs the cerebellum. I would be surprised if the cerebellum would have less since I expect some kind of neuron/axon proportionality. Especially since the main neuron type in the cerebellum are highly branchial.
andbberger 5y ago
> The axons are just as important as the neurons. what. an axon is a component of a neuron are you referring to glia?
- kaba0 5y ago
  No, I meant to highlight that “most neurons are here” may not be the important question to ask, but their interconnectedness which is achieved through axons, and which gives the most volume of the brainz
  - andbberger 5y ago
    indeed the focus of most research is the neuropils