If we assume universal algorithm duplicated throughout cortex, same must happen with patterns made up of collections of edges, and the complementary alternatives in higher regions of cortex. Also I will mention that Neuroscientist Edmund T. Rolls has a slightly modified stdp rule, that seems to offer good computational models too. His latest book is available for free from his website in PDF form, and I believe covers the learning rule. edit(went over and found rule it is memory trace rule page 109 of latest book) I will say also, that both the recent nature paper on brain wiring, april edition of 2025, as well as prior wiring diagrams, hint that in some cases one neuron can make more than one synapse with another neuron. How rare such is or how frequent, I'm not sure, I know it happens, and some neuroscientists say multiple synapses between two neurons wouldn't be good but it does seem to happen.
No comments:
Post a Comment