Real-world implications of the TBT?

Research and Theory
1

None of the coverage I’ve seen of the Thousand Brains Theory (TBT) addresses real-world implications, such as:

  • How might TBT inform Orientation and Mobility training for the blind? For example, what could the concept of Reference Frames suggest in terms of the design of training exercises?
  • More generally, what might TBT suggest regarding practices in education?
  • It seems like “earworms” could be caused by sets of cortical columns which are getting triggered “for no good reason”. Might TBT suggest ways of dealing with these?
  • etc.

Would anyone be interested in discussing this sort of thing? (ducks)

-r

4 Likes
OT: "Can Machines Think" (Richard Feynman video)
2

I’d be thoroughly interested! At some point in the future I’d like to have a podcast that address these tangential-to-the-project but important for humanity topics.

2 Likes
3

These are fascinating questions. I have long felt that brain theory should lead to better practices for education. The ultimate goal would be to have a complete picture of how memories are formed in the brain and from that derive better practices.

The use of reference frames is already evident in pedagogy, for example, learn by doing, teaching math with visuals and history with timelines and maps. I can imagine doing more detailed experiments to test the TBT related to education, but this is beyond my ability.

6 Likes
4

I would be particularly interested in doing the best practices in education part. I have been trying to do the same with LLMs by taking the research backed pedagogical techniques and building an AI tutor guided by those.

3 Likes
5

Designing “more detailed experiments to test the TBT related to education” seems like a very worthwhile direction to explore. I wouldn’t expect nice crisp results, given that this is a “soft science”, but useful insights and practices might well come out of it.

In the direction of “teaching math with visuals and history with timelines and maps” (etc), I’d strongly recommend watching Bret Victor’s tour de force presentation, Media for Thinking the Unthinkable. His companion essay, An Ill-Advised Personal Note about "Media for Thinking the Unthinkable", is also worth reading…

-r

1 Like
6

Peter Diamandis and Neal de Grasse Tyson just put out an interesting YouTube video: Neil deGrasse Tyson on Understanding Our Current Reality …. In it, they discuss the idea that human brains are set up to handle linear processes, but our current reality is filled with examples of exponential growth. Hmmmmmm.

2 Likes
7

I just watched another StarTalk program by Neal de Grasse Tyson and friends: The Science of Illusions with Teller. In it, Neal discusses the neurophysiology of magic with Teller (of Penn & Teller) with co-host Gary O’Reilly and neuroscientist Susana Martinez-Conde.

The program alternates video clips from an interview of Teller by Dr. Tyson with a wide-ranging discussion of attention, magic, prediction, sensory and cognitive limitations, etc. It’s at a much higher level than Monty’s current work (AFAIK), but I think it has a lot of longer-term relevance…

8

TBT is much more compelling than the vast-GPU-farm paradigm that has presently gripped the AI world. TBT offers a more direct path to Intelligence Amplification (IA), in the tradition of Hawkins’ insights that inspired the Palm Pilot three decades ago.

Penn & Teller did a documentary years ago that briefly explored this terrain by asking a stunning question: “Was Vermeer a machine?”

1 Like
9

Although the LLMs are great at explaining their topic matter (albeit with occasional glitches), they aren’t so great at explaining their own processes. Also, the generated models tend to be rather monolithic, making it hard to simply “pour in” sub-models. I’m hopeful that TBT will be able to address both of these concerns.