Mapping neural connections

[michaelseung]

We've discussed various methods of defining brain regions in class (spatial, anatomical, functional). Is there any advantage in defining the areas of the brain purely on neural connections? Scientists might map the network for extremely small and simple organisms, but for large organisms like humans, tracking the billions of interconnected neurons obviously poses an incredible challenge. If such a challenge can somehow be overcome, how useful would this information actually be? Or would it still be more effective to distinguish brain regions with functional or anatomical partitioning?

[adjordan]

Different functional areas of the brain usually send many neurons to several targets, so instead of tracking every individual neuron that provides input from the amygdala to the hypothalamus, it's simpler (and probably just as useful) to measure the whole signal from every neuron. Measuring individual neurons might even be worse than measuring many, as a single neuron can stray pretty far from the mean of the rest of the neuronal signals in the same pathway.

[DSP137]

Is this perhaps where clustering becomes useful? The brain is complicated to look at as a network, especially if we look at each neuron and its connections independently, but if there is a way to group neurons together and track the connections between the groups we get a general idea of what is going on.