Day 4
Day 4 (12/5)
Slept late & quite poorly again thanks to having to clean for hours due to this pantry moth infestation. Just a little bit of time to work on this. Let’s review lit notes & ask chatgpt/notebooklm what the core assumptions are that each paper makes. That might be a fruitful way forward. I do think I’ve been under-using LLMs as a way to interrogate papers and instead just skimming them myself. Should do both!
OK let’s go through it idea by idea
Neural population geometry
Is there an analog to Kolmogorov complexity for how neuron populations activate in response to stimuli? One of the difficulties is that these neural state space trajectories seem to be quite dynamical / time dependent
Kolmogorov complexity of an object is the length of the shortest computer program needed to produce that object. So what would the KC of a particular neural state space trajectory be?
How does a population of neurons start firing? Is it all simultaneous? Presumably not given there’s synapses that are crossed in some order. So you have to set some neuronal dominoes going to get a particular population level activation?
What might a useful KC analog definition look like?
- Minimum number of neurons that need to be activated to achieve a particular neural state space trajectoroy?