Non-motor Regions Creating Motor Commands? I'm Skeptical

I’m sure this will be a little controversial, but I’m not at all convinced that all cortical columns are actually generating motor commands. Jeff has yet to convince me on that. In motor regions perhaps, but not in all cortical columns. They may be projecting to motor-regulating subcortical regions, but that doesn’t mean they’re actually generating motor commands.

I have my own hypothesis about motor control in HTM, and the role I believe the Basal Ganglia plays relative to HTM theory. That thread is pretty old at this point and could maybe use some revisions, but I think all the main parts of my hypothesis are explained in the thread, even if they’re a bit unorganized. I should maybe redo my explanation of it sometime, and how I think it fits with Numenta’s current ideas. Some of the ideas I had there were kind of a primitive form of the Thousand Brains Model (though this was from before Numenta started discussing that theory, and my thread doesn’t explain it well). I’m not yet sure how grid/displacement cells play with it, but that definitely seems like a very interesting subject.

A summary of my hypothesis is this:

  • The striatum is doing a similar temporal memory algorithm to the cortex, however with a reinforcement-learning twist.
  • The dopaminergic receptors on the striatal cells use inhibition and disinhibition effects to group striatal cells into 3 subpopulations; those biased to respond to high-dopamine situations, those biased toward low-dopamine situations, and those with no significant bias one way or the other. These can be thought of as learning patterns/sequences that result in good/bad/neutral outcomes.
  • The extra circuitry in the basal ganglia (globus pallidus, etc.) seems perfect for detecting when “good” patterns are detected, and no “bad” patterns are detected (e.g, the action the cortex is proposing is more likely to result in positive outcomes than negative ones). When this occurs, signals are relayed back to the cortex.
  • The basal ganglia and cortex form many loops, that seem to have relatively small receptive fields. The cortex projects to the striatum, which projects to other areas of the basal ganglia, and those eventually project back to the same area of the cortex, likely within the same cortical column. These loops only feed back to the frontal lobe however. The frontal lobe is structured as a hierarchy, and almost all of it forms these loops with the BG, forming a hierarchy of reinforcement learners.
  • The feedback from the BG to the cortex appears to be selectively disinhibiting/exciting cortical columns that are producing “good” actions, and possibly inhibiting those creating “bad” actions.
  • Motor commands seem mostly driven by population coding. If the cortex is left to learn on its own, it will tend to assign neurons to patterns randomly, which doesn’t lend itself to population coding well. The BG in this model would be an extra force biasing neurons toward representing patterns strictly associated with useful commands. If the BG were to stop functioning properly, this force would then disappear, and neurons would fall into a new local minima and begin organizing themselves randomly again. It’s conceivable that this would result in erratic motor behavior, which is exactly what is seen in degenerative diseases of the BG (Parkinsons, Huntingtons, etc.).

A few things to note that probably are not mentioned or explained in depth in the original thread:

  • Every part of the cortex projects to the striatum (which is where I think Jeff is getting the idea that every part of the cortex has motor output), but only the frontal lobe gets feedback from it. I think this lines up well with the Thousand Brains Model; these other areas aren’t generating motor commands, but are providing context. If the motor cortex suggests reaching your hand out in front of you, that may be a good idea if there’s a friendly cat there, but not a good idea if there’s a hot stove. In that case, non-motor information may be necessary for the BG to make informed judgements on the outcomes of actions. This doesn’t require motor commands from these regions; just information on objects, locations, etc. Perhaps there’s something interesting with displacement cells going on here.
  • Also related to the Thousand Brains Model, the frontal lobe seems to project back to many sensory regions of the cortex. Even if the sensory regions aren’t creating their own motor commands, something with the appearance of motor commands could perhaps emerge from the interplay between these regions. I don’t have any concrete or well-developed ideas on this yet, however.

I don’t see any way in which the cortex could form any real motor commands on its own. Motor commands would imply that there’s some form of intention or goal that the cortex is attempting to achieve. Without reinforcement learning, I struggle to see what mechanism in the cortex could be driving that.

Like I said though, something interesting might be going on with displacement cells. I’m inclined to say that this is unlikely to be motor commands, but perhaps some kind of map of where features are relative to each other? That seems like it would be straightforward for the cortex to generate while being very useful for the striatum in judging the effectiveness of commands, as well as in path integration.

3 Likes

I picked up this bit in your post …

Also related to the Thousand Brains Model, the frontal lobe seems to project back to many sensory regions of the cortex. Even if the sensory regions aren’t creating their own motor commands, something with the appearance of motor commands could perhaps emerge from the interplay between these regions. I don’t have any concrete or well-developed ideas on this yet, however.

… and wanted to help with adding some meat on the bones of your idea.

How does this fit in your thoughts?

And this:

What makes all this to’ing and fro’ing in the cortex relavant is much of the feedback paths are considered motor output from the forebrain, radiating back towards the sensory cortex.

2 Likes

I had a similar idea. I think that an output of the cortex, layer 5, primarily responds to motion. It then projects the motion it detects to the motor neurons which drives that motion. There is a feedback loop between the muscles and the motor cortex, which makes body motions persist until the thalamus intervenes. There is evidence that input the the apical dendrites of L5 pyramidal neurons can cause spontaneous activity, which I think is the initiation of a new movement. Some cortical areas do not connect to any muscles, but they’re still detecting movement in their sensory inputs, and relaying that information throughout the brain.

2 Likes

Your points about the frontal lobe having feedback to the sensory regions seems to line up with my thoughts on that pretty well. I don’t have particularly good access to neuroscience journals (mostly working off some neuroscience textbooks, some free papers I’ve read, and some information I’ve pieced together from a variety of sources), so the specific details of my neuroscience knowledge may be lacking in a few areas.

One of the biggest problems I can’t figure out with this is how the frontal lobe drives the sensory regions. You need some way to feed back to proximal synapses, likely via the thalamus and L4. Most connections that aren’t strictly feed-forward seem to just project to apical dendrites in L1. Those are distal connections, and so would only depolarize neurons. I haven’t found any mechanisms by which the frontal lobe could cause real activity in sensory regions. Such a mechanism has to exist though, or else things like sensory deprivation tanks would probably be leaving people braindead. If you can point me to any specifics on a mechanism for that, that would answer a lot of questions I’ve had for a while now.

Ignoring that specific problem though, the frontal lobe is (according to my hypothesis) trained to perform actions by the basal ganglia. My main focus hasn’t been on how the instincts/motivations/etc. are generated, just how the cortex is made to conform to them. In a sense, the rear half of the brain can be thought of as a form of pure HTM that just converts sensory input into some giant SDR that describes the perceived state world. The front half of the brain (frontal lobe) then proceeds to generate a sequence of actions to accomplish a goal based on the SDRs streaming in.

It would make sense that if the frontal lobe can influence the sensory cortices, it could perhaps influence them in a way that they relay information to the striatum that’s more useful. As for the specifics there, I’m not too sure.


As in apical activity causing spikes rather than just depolarization? Got a paper about that? I’d love to read more. That might be an important mechanism that I’m looking for.

As for your other points, I’m not quite sure how they match up with my ideas, unless I’m just misunderstanding you. The part about the thalamus intervening might be where our ideas line up though, as the BG doesn’t actually project back to the cortex, but rather to cells in the thalamus that project back to the cortex. Motor-sensory loops that the BG can disrupt are an interesting idea I hadn’t considered before.

My mental model of action generation in the cortex is instead the sensory regions form a loose hierarchy going from low-level features to high-level models, while the frontal lobe forms a hierarchy going from high-level goals to low-level actions. The sensory regions are doing a form of pattern recognition and sequence detection, while the frontal lobe is doing pattern and sequence synthesis. That synthesis is influenced by sensory inputs, but the cortex cannot do this on its own. The BG is there to actually organize neurons into firing patterns that generate these patterns and sequences, and filter out neurons learning patterns that don’t fit.

Without the BG’s influence, the frontal lobe wouldn’t be generating motor commands; it would just be trying to learn higher-level sensory patterns, regardless of if they’re useful for action generation or not. Which neurons learn which patterns/sequences is mostly just up to random chance. Combine these two; neurons learning all patterns available regardless of utility, and patterns lacking any kind of orchestration or order, and you just wind up with random SDRs. Link that up to population-coded muscles, and you get random flailing.

I don’t think motor commands can be generated without some kind of mechanism to filter them based on utility. That seems to only exist in the BG, and not in the cortex. L5 might generate some data that’s relevant to action selection, but I find it hard to call that a motor command.

1 Like

Molinari etal 2008 “Cerebellum and Detection of Sequences, from Perception to Cognition” may be of interest? (short review of clinical experiments, proposing that cortex->cerebellum->cortex loop is important for action selection):

“Behavioral or script sequencing can be defined as the process that allows for the correct recognition of spatial and temporal relations among behaviorally relevant actions [46]. Script sequencing has been considered to be sustained by frontal lobe and basal ganglia circuits and it requires the ability to plan ahead [47]. The demonstration that script sequencing is impaired after cerebellar damage indicates that the cerebellum also has to be included in the script network…”

2 Likes

You are in luck! These fiber tracts are well documented:

1 Like

You might want to look in to somatostatin-expressing (SOM+) and VIP+ cells. I recall SOM+ cells inhibit distal apical dendrites a lot, like maybe one or a few can strongly inhibit all apical tufts of L5 thick tufted cells* in a cortical column (macrocolumn not minicolumn). M1 activates VIP+ cells which inhibit SOM+ cells, and it might be related to behavior.
*(L5 TT cells are the only L5 cells which project to subcortical structures, except slender tufted L5 cells project to the striatum).
https://www.nature.com/articles/nature11601

For higher levels of the hierarchy causing cells in lower levels to fire, you could also read about L6. M1 strongly activates some cells in S1. I don’t know whether M1 is higher in the hierarchy than S1 (it might be what/where pathway rather than hierarchy). I recall that it is based on the cortico-thalamo-cortical pathway but I’m not sure.
https://pdfs.semanticscholar.org/b8c8/0cf70ca7a0260a0419c980b2f87744e420c1.pdf

One possibility is that the cortex doesn’t need feedback to proximal dendrites. It might just nudge spontaneous activity a lot, by the biasing effect of distal dendrites.

Sources are in these google docs. I think ~1/2 don’t have a paywall.

1 Like

I don’t have a citation for that. Sorry, I wish I knew where I read that too.

Something to consider: Apical Dendrites have a special mechanism which basal dendrites do not have. The paper “Larkum 2012, Redirecting” has a decent overview. Here is a discussion thread regarding that paper: Larkum 2013 & A State of Attention

1 Like

No science backed input here just an hypothesis (or more).

  1. The non-motor regions have their motor outputs repurposed to either (or both)
  2. provide inputs for other regions.
  3. To “move” attention. Might not be a coincidence on how we speak of attention as of another … moveable “eye”. “Turn attention towards X” or “focus on this problem”. If we can not tell much about what attention is, at least it seems that it is “steerable” Maybe, just maybe attention is “moved” by (some of) these seemingly unconnected to any “muscle” motor outputs.
1 Like

I have posted about this several times in the past.
The same way that we can initiate external action using our musculature we can start internal action using the same basic structure of early stage forebrain and cerebellum.

I think we can do these internal actions faster as we are not limited by the physical reaction time of our bodies.

1 Like