Voting in the Thousand Brains Theory (Part One) - August 25, 2021

Subutai goes over voting in the Thousand Brains Theory.

In the first of two research meetings, he lays the groundwork for understanding how columns vote in the theory by unpacking the ideas in our “Columns” paper. First, he presents the hypothesis of the paper on how cortical columns learn predictive models of sensorimotor sequences. Then, he explains the mechanisms behind a single cortical column and how it learns complete objects by sensing different locations and integrating inputs over time. In the next research meeting, he will review voting across multiple columns.

Columns paper: “A Theory of How Columns in the Neocortex Enable Learning the Structure of the World”

Other paper mentioned: “The columnar organization of the neocortex”


One thing I always find confusing … in the diagrams you use ARROWS from one block to another.

It seems that the INPUT (feat@loc) block is connecting COLUMN output-signal to the OUTPUT (pooling) layer i.e. if there are N-columns then N-bits SDR is passed to the next layer.

But that can not be, because your step diagrams dont match by column.

The only possible way is if the connections between IN<->OUT are direct neuron-to-neuron connections from IN layer (eventually neuron of the same column can be connected to different OUT-neuron) to the OUT layer. m’I correct ??

Second :

  • The Input layer has only External distal connections (LOC). FF is Feat
  • The Output layer has only Lateral distal connections. FF is from Input layer.