I suspect that there is a distribution of constancy across parameters. The molecular parameters (i.e. channels) are probably the most constant. More emergent parameters such as resting potential and reversal potentials are probably also mostly constant, maybe varying a bit?
@KordingLab @tyrell_turing @TonyZador @neuro_data @bradpwyble If the params of neurons themselves are just params in the computation graph, what is the consensus that they are largely held fixed during an organisms lifetime vs the notion that they too change with learning?
-
-
-
Counterpoint: I believe that all parameters, including channel densities etc are learned. Reason? Learning a channel is computationally identical to learning a synapse with constant input.
- Pokaż odpowiedzi
Nowa rozmowa -
-
-
Some params are pretty fixed (e.g. biophysical channel kinetics), but many params under regulatory control by the cell -- like channel densities.
@Timothy0Leary and@MarderLab have done beautiful modeling and experimental work on this in the context of homeostasis. -
@bradpwyble@johndmurray please keep in mind im not a trained neuroscientist. A more specific question would be: if the behaviors of neurons are changing (i.e. the params are changing), does that mean they too are following a gradient? I suppose thats what it must mean. - Pokaż odpowiedzi
Nowa rozmowa -
-
-
I think the consensus is that mainly synaptic strengths move, and everything else changes via homeostatic control loops that try to deliberately keep specific things fixed (average firing rate etc.) This is somewhat outdated e.g.: https://www.nature.com/articles/nature06725 …https://www.nature.com/articles/nn.2428 …
-
but do you think our techniques would allow us to know if the channel densities were learned?
- Pokaż odpowiedzi
Nowa rozmowa -
-
-
You might find this review interesting: https://www.nature.com/articles/s41539-019-0048-y … (Though the cell-intrinsic parameters it focuses on might be different from those you have in mind)
-
All of this is dependent on what kind of circuit you're studying. Eve Marder's work showed that ion channel composition varies a ton but the output remains constant. This is an essential circuit, though, lobster dies if it stops working.
Koniec rozmowy
Nowa rozmowa -
-
-
Just wanted to point out Albert Lee showed excitability of a hippocampal cell before exposure to a novel environment can predict it's recruitment during learning. Plastic excitability could select/allocate cells for future learning or protect stored memories.
Dziękujemy. Twitter skorzysta z tych informacji, aby Twoja oś czasu bardziej Ci odpowiadała. CofnijCofnij
-
-
-
Network's topology is probly more stable than synaptic weights. Yet, there is more information in topology
Dziękujemy. Twitter skorzysta z tych informacji, aby Twoja oś czasu bardziej Ci odpowiadała. CofnijCofnij
-
Wydaje się, że ładowanie zajmuje dużo czasu.
Twitter jest przeciążony lub wystąpił chwilowy problem. Spróbuj ponownie lub sprawdź status Twittera, aby uzyskać więcej informacji.