A Thousand Brains Q&A with Jeff for the HTM Community

A couple months ago, I asked for input on doing a special event with this community for Jeff to talk about his new book, A Thousand Brains.

Thank you for the replies and the feedback. The consensus seemed to be that a Q&A format where people could post questions ahead of time would work, so I want to check back in now that the book has been out for a few weeks.

I hope that by now everyone who was waiting for a book has gotten it. I know many of you waited a long time, especially for international deliveries. If you are still waiting on an order, let me know.

We are planning to announce our next Brains@Bay Meetup soon, which will also be with Jeff talking about the book, in a similar format, and is open to anyone to attend. But we still want to give the opportunity to do an event with this Community if there’s interest.

We’d be looking at something mid-April, likely in the morning California time to accommodate other time zones. We can collect questions here on this thread ahead of time.

8 Likes

Mine arrived yesterday! (NSW, Australia). Very excited to read it over the next week.

4 Likes

Adding questions here that have been previously collected on other threads:

Thank you very much for this opportunity. I received my copy in Germany and as a participant of this forum for over 10 years, I want to pass on my highest compliments to Jeff for this great accomplishment with this book, which recaptures the long path to this great milestone. I fully agree with his assessment that Numenta has now found all the puzzle pieces on the edges, corners and some inner portions of the brains mystery. The framework is now identified. I have two questions I would like to pass on to Jeff:
1.) How is a multi-CC consensus achieved after voting begins? In other words, how is the “invariant” stability reached amongst the voting neurons from each CC? (There is no guarantee they all will always agree, right? They may encounter some cognitive dissonance).
2.) Pertaining to learning: If a multiple set of CCs is involved in learning a new object, and each CC is creating an individual model of the same object (i.e. coffee cup), how are all these individually unique models (city maps) in each CC interconnected with each other in such a way, that the same set will be used during inference (perception cycle) so that the voting is based on the same set of models (city maps)? Could this involve some sort of timestamp in episodic memory? (This would allow joint selection during recall). Or is it possible that all the models in the superset of CCs are stored in some form of temporal-spatial associative super-structure, encompassing the Thalamus? I thank you dearly in advance. I want to express my certainty that you have achieved a phenomenal breakthrough, worthy of Dr. Vernon Mountcastle’s admiration.
Your highly motivated follower and community participant,
Joe Anthony Perez

1 Like

I have one more question: For me, I see an analogy of the CC maps (object models) in our ability of 10-finger typing skills on a keyboard (especially for us dinosaurs that learned it in school). Each hand has a clear map of a segment of the keyboard. So each hand could represent a CC in our neocortex. (I understand, that the maps in CCs are not adjacent segments of the object, but normally over-lapping with some variance in feature fidelity. So both hands know the keyboard, but some letters are more clear to the left hand and others more clear to the right hand). And we agree that the motor-sensory discovery process (when, for example, exploring an object in order to identify it) requires movement. How do the CCs in the active set of CC (say 1000 CCs) agree on motor-control direction during exploration? Each CC would like to determine which direction to explore next in order to disambiguate. Does some preliminary form of voting already take place during the exploration-disambiguation cycle to agree on, for example, the direction of the next saccade of an eye? Thanks a lot in advance! Joe

I received mine a few weeks ago. Well worth the wait!












4 Likes

I have just completed a book review for “A Thousand Brains” which I would like to share with this community. (See posting above). I have written this review from the perspective of one of you, who has been an avid follower and occasional participant to this forum and community events for about a decade. I wanted also to both pay tribute to Jeff and his valuable contributions which he so openly shares with us. And also to reflect on my own thoughts which reading this book has evoked. Perhaps my thoughts are also of interest or may stimulate further discussion. I certainly recommend anyone who has not yet had the pleasure, to make certain you read “A Thousand Brains”.

1 Like

Hello @BrainConstellation! Thanks for such an extensive and thoughtful review on A Thousand Brains! I shared it with Jeff at a team meeting and everyone thought what you did was great. I’ve linked them to your review and they’re looking forward to reading it :slight_smile:

1 Like

Thanks a lot for the feedback. I am very honored that it receives attention. I may be far off on some assumptions, because I do not have the same level of awareness and knowledge regarding all the neuroscience publications. My thoughts hinge to a large extent, on how many pathways from CC-to-CC exist, that can play a significant role in the consensus-building (voting) process. We also have an insightful exchange of views going on right now in the parallel thread: My “Thousand Brains” book review

I’m very interested, Jeff tell me how HTM describes vision and hearing, how working memory is modeled in the neocortex.