Questions about Mini-Columns

A few questions about mini-columns:

  1. Are all mini-columns the same?
  2. Is a single mini-column considered a column of neurons in a single cortical column layer or is a single mini-column span all layers?
  3. Is there a fundamental role/function that all mini-columns do?

According to Numenta (unless I misunderstood):

  1. No, but the differences are minor.
  2. The minicolumn spans all layers.
  3. Here I’m not sure, but I think the question is important. This is what I think. All minicolumns do:
  • receive input
  • normalise input through a spacial pooler
  • detect sequences through a temporal memory
  • compare states through voting with neighboring minicolumns
  • project outputs to the thalamus
  • project outputs to neighboring columns in the same layer
  • project outputs to cells in other layers
  • project outputs to cells in other maps of the neocortex

If someone spots errors, please please correct me.

1 Like

They’re not sure, I think.

This might show different minicolumns in two sublayers of the same layer. From what I remember, I’m not 100% convinced.
Lattice system of functionally distinct cell types in the neocortex

Functionally, a minicolumn is just a group of cells which represent the same thing, and probably each have something else (context) so there’s a purpose having multiple cells. There could be things which aren’t physically organized like minicolumns but function the same.

1 Like