Need references on cortical mini-columns

Science General Neuroscience
1

Hi,

I’ve been reading the paper by @fergalbyrne Encoding Reality: Prediction-Assisted Cortical Learning Algorithm in Hierarchical Temporal Memory ([1509.08255v2] Encoding Reality: Prediction-Assisted Cortical Learning Algorithm in Hierarchical Temporal Memory) and on page 3 he says:

Real layers of cortex have been found to organise their neurons in mini- columns of about 30 cells, which have strongly correlated feedforward response.

But there’s no source given for this claim. Is there any paper or textbook where this is discussed. I ask because I’m writing up how HTM works for my thesis and I feel like I should give a source when I say that HTM is based on how the brain is actually organised.

1 Like
2

You might look at Jeff and Subutai’s latest paper, Why Neurons Have Thousands of Synapses, a Theory of Sequence Memory in Neocortex. It has a section on mini-columns with 3 references, although I have not read them.

3

This blog post gives some references on cortical columns and regions:

1 Like
7

Changed title from “Need a Citation Regarding a claim in a HTM Paper” and moved to #htm-theory:neuroscience

9

Thanks for the links guys! I have another related question. Is there a reference that explains (at a high level) the model of the mammalian neocortex used for HTM? I understand HTM deviates from the operation of a real brains for technical reasons, but I’d like the reader to understand what the algorithms are inspired by, and how they got to where they are in the current nupic implementation.