Does a column ever connect to itself?

question

#1

Are there examples in the neocortex of cells within the same column that connect to each other?


#2

Yes, that is exactly what happens during “Temporal Memory” within a layer. Watch:


#3

4:22

“The output of a layer can project to the input of other layers including itself.”

Assume a sensor with all the letters of the alphabet S = [“a”, “b”, “c” … “z”].

I’d like to teach the system the word “rootkit” that has the letter “o” appearing twice in succession.

The sensory input should be:

t = 0: “r”
t = 1: “o”
t = 2: “o”
t = 3: “t”
t = 4: “k”
t = 5: “i”
t = 6: “t”

At t = 1 and t = 2 the same columns will be active because they both represent “o” but different cells within the columns will be active because the first “o” came after an “r” and the second “o” came after the first “o”.

Do cells at t = 2 connect to the cells that were active at t = 1 even though they are located in the same column(s)?

Is this the correct interpretation?


#4

Yep, that’s correct