I have been thinking of this problem for a while now. How human's memory is trigged? For example, you barely think of an event (your friend and you had a great conversation in a club on a rainy night) which happened 20 years ago. However, when you pick up a pen, which was given to you by your friend. Then you suddenly remember some content of the club conversation which took place 20 years ago.
How does a neural network model such brain activity? Here I am proposing a concept of super neuron. This super neuron is a loosely connected neuron in the NN with both a low probability and an associated event (or event summary) stored deeply inside the network.
The event associated with the super neuron has to be trigged by the event. Even more, the probability after the event registered with the system and then the memory flashing up is extremely low.
No comments:
Post a Comment