Numenta Research Meeting - May 22, 2019

1 Like

Concerning the 22/5 research meeting, about the local increased firing rate of certain grid cell locations compared to others, I was thinking perhaps the same grid cell module might produce multiple grid patterns on top of each other.

Compare this with a ruler with indentations at every 1/16th inch, larger indentation at quarter inch and so on.

Perhaps as a resonance pattern, where certain spots on the grid show an increased rate due to addition of multiple sources (and perhaps even others lesser through inhibition). After all we don’t know what the source for this firing is. If there are multiple sources, it would make sense that those influence each other.

Also, in a limited area, the larger pattern(s) might be invisible due to a limited number of points. As in low resolution.

@Bitking you said you know of some papers on this subject. Any that could corroborate this?

@chuston also had a good closing remark: increased activity would possibly have an effect upstream in case sensory input from a feature is combined with the increased firing grid location. The synapses would get reinforced faster at that location.

This is so fascinating! :-).‬

‪@Numenta your 2019-05-29 research meeting sounded like you were almost talking about “gravity well root locus perception” (is that even a thing ?)‬

1 Like

Hello Rajinder, welcome to the forum.

Yes, @mrcslws is basing an hypothesis on the innate ability to detect a gravity vector.

Unfortunately that paper you linked to seems to indicate that the mainstream hypothesis for gravity perception in plants is not correct. (I am surprised that biology science hasn’t figured out this principle yet. It’s humbling sometimes, to find out how little we really know).

I’m also enclined to think (speculatively) that if there is a neuronal perception of gravity, it should be based on a much faster system than whatever plants use.

I found this study using fMRI to detect what regions of the human brain are involved in gravity perception.

They seem to conclude that it’s based on motor resonance (I think that is what Numenta refers to as sensimotor feedback, but I’m not sure) and visual familiarity (what I compare to HTMs sequential memory with visual information).

They’re not talking about a sensor for gravity though. I thought that came from the vestibular system, as explained somewhere in this document.