I read about prefrontal cortex and its executive function in brain. Does htm had any theory about it? Is the frontal area same as the sensory cortex in its internal working?
I haven’t seen HTM seeking to break the neocortex down into any functional sub-areas, rather it builds off the Mountcastle idea that the same basic algorithm exists across the neocortex.
The HTM guidance has been that the function of a area is determined by what it is hooked up to.
In the back the connections are to the senses.
In the front we are hooked to the sub-cortical structures that are the control structures.
The temporal lobes collect up the digested sensory areas and concentrate that for episodic memory and (bidirectional) connect it to the hippocampus one-day memory buffer.
At least some regions of prefrontal cortex have six layers, so it probably works mostly the same. There might be important specializations for things like working memory, though.
Yes, as long as I know, the new theory is still extend column’s function at all.
Yes, So, the more important thing is how to find a way to train such a six layers column-based structured artificial neural networks to memorize, think, inference, induction, form concepts, setup associative relationships between concepts. Another thing is the ancient brain, how it cooperates with the neocortex to make goal based mechanism, abnormal event discovery mechanism, and make world models and so on. There are a lot important concepts need to be implemented by a good agi theory and its framework. So if even we think Jeff’s theory is right, we must firstly explain all behaviors of human-being, secondly, is able to find a way to train the network to behave like human-being.