Background thoughts, mental reminders and HTM

Hi all,

Since earlier this year when I was first introduced to HTM, I have spent a lot of time day to day gradually comprehending how it can help explain different aspects of thought and intelligence.

I have a list of these aspects that I’m still not sure about, in other words I don’t have an HTM-oriented explanation or analogy.

Instead of dumping them all in one topic, I’ll ask them one at a time and see if the HTM community can help by:
a) directing me to some good reading
b) offering your own ideas or explanation
c) telling me that nobody knows the answer yet

This first question relates to how “background” thoughts work in the neocortex. By this I mean that the thought doesn’t feel as though it occupies my attention, but there is either some level of effort being invested into it, or it is idle but will spontaneously re-enter my attention over the coming minutes, hours or sometimes days.


  • Someone proposes a question to me of a memory recall nature (“what was the name of the main actor in movie X”, or “which band sang song Y”). I can’t recall it, but then a few hours later it spontaneously pops into my head.
  • Someone proposes a question to me of a technical nature (“what is the best approach to solving problem X”), and I’ll tell them I need to think about it and come back to them. But without really spending much time consciously pondering it, the next day I am usually feel much better placed to propose a solution.
  • I tell myself I need to remember to get milk before going home, or similarly I tell myself that I have to remember to talk to a certain person about a certain topic. Later on, often hours later, I spontaneously remember to do it (hopefully, anyway…it doesn’t always happen).

With HTM, neurons are activated up and down the layers of columns, with the representations becoming more invariant further up the hierarchy. But to borrow a Comp Sci term, are there many “threads” of neuron activation? And if so:

  • How many could there be at any one time?
  • Is there a “main” one, or what is it that would make one of them feel “active” in our attention?
  • How do they interact or interrupt each other?
  • Would they all generally involve the full hierarchy? It doesn’t seem to make sense that the lower levels would be involved in these sorts of tasks, so does that mean the activations become predominately Distal for some period until a downward path is available?

I have a feeling that perhaps I’m making a too simplistic connection between the basic HTM theory and the topic of attention.


I’m not an authority on this. But the way I think about it, the fraction of your mind that you’re consciously aware of is very small. So different regions that are not involved in consciousness (we don’t know which these are, although there’s a lot of work toward identifying all of these correlates) can be working in the background all the time, undergoing spontaneous activity and transitions around different attractor states.

And then my assumption is that when something particularly salient happens in these background regions, the degree of overlap with the previous foreground activity (trying to recall something or whatever) causes a bunch of distal segments in the conscious regions to fire, and the conscious regions take on those thoughts.

What I think is not at all controversial is that there’s a lot of unconscious background activity going on all the time, and due to the heavy interconnectivity of the brain, this can bias the foreground regions when the overlap is significant.

Oh, and keep in mind that the brain is not a strict hierarchy. There are all sorts of branches and parallel hierarchies that connect in weird ways.


@jakebruce thanks for the reply, that was fast! Was still pedantically editing my post :stuck_out_tongue:

I hadn’t considered that conscious vs subconscious could mean a different region is responsible. My invariant representation of the actor from that movie could be activated in either scenario, couldn’t it?

Could be. I would be surprised if coherent background activity could be sustained in the same region for that long without affecting the main state.

The other thing is just that small coincidental SDR overlaps are likely to happen with increasing probability with time, so it may just be a matter of random activity accidentally co-activating the target memory with the recent episodic memory in the hippocampus that represents your searching for the info.