Cliques of Neurons Bound into Cavities

I’m reading this paper at the moment. I am vaguely reminded of error correcting codes.

Maybe it would be worth rereading some of the books by John L Casti.

I’m not heavily versed in algebraic topology, so I’m not completely familiar with much of the terminology here, but from what I gather, the paper suggests that based on computer models, neocortical networks should feature far more cases of relatively large, heavily interconnected networks of neurons than would be expected in a purely random network. Not terribly large (like 5-8 neurons), but where each neuron in the group is connected to each other.

This seems to me to perhaps have something to do with sequence memory. Perhaps this is a result of the cortex learning a lot of complex sequences? The paper also is suggesting this as a new way to mathematically analyze the connectivity in neural networks; perhaps it would be a good idea to run similar analysis on an HTM network? If we want to show that HTM is a good model of the cortex, more evidence to support it is always good.

Yeh, I only was only able to get a rough idea of what they are talking about. I would like to reread one of Casti’s books to see if it would help. I think there was some explanation of simplexes and reasoning with probability. Or are there crystalline aspects to the structure of the brain:
Or it could just be the signature of a common hardwired algorithm in the brain.
Anyway, I think I’d rather wait for someone else to figure it out.

Just read an article on this paper.

When working on a problem, neurons in the brain form into “a multi-dimensional sandcastle that materialises out of the sand and then disintegrates”

1 Like