Numenta Office Hours (April 14, 2020)

[vpuente]

Literally a map of the cortex, i.e. the place where the SWR should happens

[fcred]

@Bitking

It was aimed at you but I wanted to flesh out the question a little.
So - the grid cells could be located remotely and the evolutionary innovation is a connecting fiber tract?
One of the “lowest cost” evolutionary sports is adding connections between areas.
I put this in the same “cost” as duplicating layers, areas/maps, or increasing the area of a region.

That would be a good way of looking at it. I have a picture in my mind of the extra connections being added by multiplexing to the grid cell module modules and have some clock cycle control the switching.
For this to work, as I noted in my question to Jeff, each sensory area would have to have its own grid cell modules in order to have the scaling be invariant over the object being sensed.
Maybe grid cells originally developed for eye positioning and then the cells were reused for navigation rather than navigation coming first. It works either way.

[rhyolight]

For those of you who’ve RSVPed, you should have just gotten an email from Meetup with the Zoom URL and password to join at 5PM.

[Falco]

@rhyolight

If you are joining the Zoom meeting, I expect you will have a comment or question for Jeff and/or Subutai. If so, please post a comment below (or DM me) so I can let them read it beforehand.

If it’s not too late, I’d like to ask a question to Jeff and one to Subutai, but I don’t need to ask them personally:

@@jhawkins

In the “ice-cube” representation of the minicolumns in V1, we see orientation markers. These are, I suppose, the orientations to which the individual columns are responsive. When you talked about this during the last research meeting, you explained that these orientations are in three dimensions. “… for instance when you move forward and down-left”. Would that mean the minicolumn receives orientation input in 3D?

@@subutai

This new idea to organise a minicolumn containing about 120 cells, spanning all neocortical layers, and containing one (or very few) grid cell(s) and one (or very few) orientation cell(s) etc. and receiving static and flow inputs, how would this impact the Numenta model of Spacial Pooler and Temporal Memory?

I know this is a bit unfair as questions, since the material is so recent. But this new idea sounds really interesting. Thanks in advance.

[rhyolight]

Starting now at https://www.twitch.tv/rhyolight_

[rhyolight]

[fcred]

In the context of my discussion with Jeff, I just came across this paper: A map of visual space in the primate entorhinal cortex that seems to support the idea of common grid cells used by the visual cortex.