Hypothesis for distal apical dendrite (also sensory-motor inference)

Hello HTM experimenters,

I have come up with a new hypothesis for sensory-motor inference. This hypothesis has two parts: one of apical dendrites and the other of the large scale structure/organization of the neural network. For both parts I will first state it and then explain it in detail. I think that this theory is biologically plausible, but I should say that my background is in computer algorithms. I am currently working to implement and test this hypothesis.

Apical Dendrites
My theory is that apical dendrites are used to control HTMs. Apical dendrites have a small multiplicative effect on the proximal excitement, which max’s out at something like %2. Apical dendrites have a different learning rule than proximal and basal dendrites; instead of learning when the input predicts the post-synaptic neurons activation, the post-synaptic neurons activation predicts the input’s activation. This is the difference between P->Q and Q->P.

The pre-synaptic apical input is called the focus set (and it is of course an SDR). The goal is for the HTM to output one of the things in the focus set if one is present. Otherwise ignore the focus mechanism and output the most prominent features (which is what HTM’s normally do). This should not cause columns to activate if originally they were not near activating; this is why the effect of apical dendrites is capped at 2%. In a sentence, apical dendrites select between multiple interpretations of the data.

Example effect of apical excitement:
proximal-excitement <= proximal-excitement * min(1.02, 1 + alpha * apical-excitement)
This exact formula is guesswork, this is just a hypothesis.

Rational
The issue which this solves is that the things we are looking for might get buried in irrelevant information. Specifically, this mechanism is for the case where one area of the cortex is making basal predictions using another area’s output and wants to tell that other area to help out by looking for specific things. By this reasoning, I would expect to neurons to receive apical input from distant areas which receive basal input from the neuron. I hope this also explains why the learning rule is different.

HTMs without these apical dendrites are not controlled in any manner. They output a representation of what they see. So what happens when you show the HTM multiple things at once? For example a colored shape, which has both a shape and a color. The HTM should output a joint/combined SDR which has bits representing both shape and color, which isn’t useful. To bias it towards outputting a color (and not a shape), the focus set could be the union of all the colors.

An interesting experiment is to stare at a single point and, without moving your eyes, think of something to look for (such as the color red, or all circles, or the letter ‘E’) and observe how the things which you notice change. It helps to look an image with many objects in it. I think that this shows the extent to which your focus affects your perception.

A possible result of training an HTM using these apical dendrites is that it changes what the HTM learns. By causing certain patterns to win out in the spatial pooler, it learns more about those patterns. This could be the basis for ‘what’ and ‘where’ paths; the ‘where’ path receiving apical input from the motor cortex and the ‘what’ path receiving apical input from the control/reward areas of the brain. This explains how ‘what’ and ‘where’ paths could receive similar feed-forward proximal input but do very different things.

Organization

Figure 1, Block diagram of agent

Rational
Notice the feedback loop consisting of: processing the proprioceptive input (labeled Motor System), passing it through a spatial pooler, and using the result of the spatial pooler to directly control the muscles. In theory, this feedback loop should cause the motor HTM to associate the sensation of a movement with doing the movement. The other areas of the brain then use apical connections to control which movements it considers. User JRowe47 proposes a similar feedback loop for learning motor control.

The three main HTMs (control, sensory, motor) are connected each to all by basal and apical connections (the red arrows). This isn’t a nuanced design but illustrates that these components can and should interact with each other.

The hardwired control is where the theory runs out. As best I can tell this is where things like emotions should enter the system.

Sensory-Motor Inference
In this theory, sensory motor inference is not an explicit step. I think it will be the natural result of basal predictions and apical focus in the visual and motor areas.

Future Work
I plan to test this hypothesis with an eye saccade mechanism, which searches large images for specific objects.

Thank you for reading,
dmac

4 Likes

You write above, “The hardwired control is where the theory runs out. As best I can tell this is where things like emotions should enter the system.

On this idea of where emotions should enter a system, please let me slip in a few details about the neuro-based AI Minds I have purportedly been creating, in Perl for webservers and in Forth for robots. A few months ago I switched from having the MainLoop module directly call the Think module, to having Mainloop call the FreeWill Volition, which in turn calls first Emotion, then Think, and then Motorium. During this current week I have been coding the Forth AI, and I came up with a weird way to (hopefully) implement the voluntary selection of a motor option. Since MindForth currently can only sound a beep, yesterday I inserted the idea 'Robots want a beep" into the MindBoot sequence. The plan is that, if the AI outputs the thought, “I am a robot,” it will also propose an inference expressed as a question, “Do I want a beep?” Accretion of positive reasons for a beep, such as the need to summon a human user to answer a question, may eventually tip the volition mechanism into sounding the beep. Meanwhile, I enjoy reading the various ideas here and I am glad to have found fellow AI enthusiasts who think in terms of neuroscience. -Arthur

1 Like

With apical dendrites, you seem to be talking about a type of attention. Something to consider is that the apical dendrite also contributes to burst firing (typically 3-6 spikes at 100+ hz), which acts as a more reliable signal. It’s also a stronger signal if the synapse does short term facilitation, because each spike in the burst produces a stronger signal.
Your interpretation of spike timing-dependent plasticity is interesting. The typical interpretation of positive timing STDP is that inputs which help make the cell fire undergo LTP, whereas those which do not contribute undergo LTD (1), which is basically hebbian learning.
I don’t know which interpretation is right, but if you decide to use the second one, it’s likely that the apical dendrite learning rule (negative timing STDP) is functionally equivalent to positive timing STDP, except it reinforces inputs which cause bursting (1). This results from a delay for the signal that the cell fired to reach the distal apical dendrite, which is the part of the apical dendrite which causes bursting.

  1. Learning Rules for Spike Timing-Dependent Plasticity Depend on Dendritic Synapse Location (Letzkus, Kampa, and Stuart, 2006)