Recently, I have been thinking about “what it feels like” (roughly, Heidegger’s Dasein) and sensory perception. I feel like I made a big break-through when I realized that standard emotions (love, lust, loneliness, jealousy…) are kind-of uninvited guests that show up whenever they feel like, and wreak havoc with your life. So my thoughts turned to the question: why would something as primal as this also be so far outside of rational control? The answer, it shot into my mind, is that I *do* have a co-inhabitant, an uninvited guest, in my house: the cerebellum.
So, conscious, verbal, talkative me (and also rational me) is the frontal cortex. That’s the “me” that is writing this. It’s attached somewhat loosely to the cerebellum, which is much much older — ancient — and was already highly functional before the cortex evolved. And what does the cerebellum do? Well, sex drive, for one. No sex means no species means extinction, so evolutionarily, those neural circuits are extremely highly conserved: evolution doesn’t fiddle with them, doesn’t override them; they need to work flawlessly for the species to survive.
So, by the time the cortex shows up, all the basic functions work well and are fine-tuned. The cortex provides additional survival benefits (e.g. reasoning!! Its a big deal!!) but these extra circuits get wired in such a way that they don’t wreck the already-works-great cerebellum.
So what does this “feel like”? The cerebellum falls in love, and gets really busy with that. The cortex only “senses” or “perceives” what the cerebellum is doing. The bandwidth between the two is relatively low. What comes across it are sensory perceptions. We give these sensations names, like love, heartache. Much like we give names to colors, or to sounds.
That is to say, the cerebellum is doing whatever it feels like, and we have little control over it, because evolution made sure we couldn’t control it. The cerebellum is wired in loosely: we can sense what it’s doing, but little more: we sense emotions, and that is why they feel like uninvited guests. We have some control: the idea that we “ride” our emotions much like one rides a horse, by cooperating and getting along. This is an ancient mystical idea, found in all world religions. And it’s explainable via wiring diagrams.
So I’m claiming here that there is a “sensory boundary” between these two distinct brain structures.
Now, the nature of “sensory boundaries” is that there’s a finite, limited-size inside, an unbounded outside, and a finite, limited bandwidth pipe between the two. The inside maintains a “world-model” of the outside; the world-model is constructed from whatever came across via sensory perception at the boundary. The inside makes all decisions of what to do next. And, to do this, it is necessarily “conscious”.
So here, I **define** consciousness to be “that thing that makes decisions based on the current state of it’s world model”. Note that this is an extremely broad definition, but I think it is correct.
This system need not be “strongly self-conscious“, i.e. it does not need to be able to express “I exist” (or even think about that), but I think it is necessarily (tautologically) “weakly self-conscious“, in that, in order to make decisions about what do do next, it has to have some idea of self-in-the-world. For example, it’s thinking “I need to do this to get food, because I am hungry”, and it is this “I” that is the “me” of consciousness. The system acts of itself, of it’s own interest, to satisfy it’s own demands, to achieve desired goals, and I think that it is this “reasoning with respect to self” that gives rise to the feeling of “me-ness”.
“I’m doing this for me”. I think that is the whole and entire set of ingredients for consciousness. To recap: this requires:
- a boundary, an finite inside, an unbounded outside,
- a “sensory” system to “perceive” the outside.
- a “world model” that represents the outside
- a (finite) menu of motor actions or “things that could be done”
- a “decision maker” that picks from this menu, by applying some kind of “reasoning” on the world model.
I think any system that has these properties is necessarily “weakly self-conscious”. This is not just a claim, but a tautology: to make a decision to act, that decision is necessarily with respect to the current state of self-in-the-world, or rather, self-in-the-world-model. The actions are necessarily in reference to the self. The decisions are always necessarily with respect to “self”. The decision–making mechanism has to be very tightly integrated with the world-model, and necessarily has some locator of “self” in relation to that model.
I think this will always give rise to the *subjective* feeling of *me* within that decision-making (sub-)system. This is how “subjective reality” arises, and any agency will necessarily have this “subjective feeling”. I think this even allows the location of “subjective me” to be located: it’s that corner of the algorithm that is tracking self-ness and it’s relation to the world.
So, to recap, again:
- an “agent” is defined to be anything that satisfies the five bullet points above,
- an agent necessarily acts with respect to self
- the subjective feeling/sensation of self arises because of the self-centeredness of decision and action.
In other words, **all agents are conscious**. Or more precisely, are **subjectively conscious** or “have internal feelings of being-ness“.
Again, this is “weakly self-conscious”. Agents, as defined above, don’t have the ability to think “I am me” in the strong sense. They’re not recursively conscious. To be recursively conscious, the world model has to include a strict, explicit model of self. Agents, as defined above, only have an implicit self-model. This is enough for the agent to get on in the world; an explicit self-model is not needed.
Anyway, I think that solves the “grounding problem”. The “grounding” is nothing more (and nothing less) than the world model. All decisions made, and all actions taken are with respect to this ground. That is, a “grounding” is exactly the same thing as a “world model”.
Please note that there was no mention of linguistic-anything in the above. No symbols, no words, no logical inference, no weight matrices, no neural nets.
Please note that you can stick in lots of additional layers and abstractions into the agent design above.
Please note that agents can have sub-agents can have sub-sub-agents.
Please note that in humans, the verbal subjective self is an agent, which can perceive the cerebellum as a sub-agent. However, that sub-agent has no language control. Thus, it is conscious, but it cannot speak. However, it can act out: it can make you heart-sick, and ill, and self-mutilate. For example, in the (ancient, Arab story) “Layla and Majun”, Majun literally goes crazy-mad out of love. His mute cerebellum went on a shit-fit that rendered his cortex inoperative. The sub-agent fucked up the agent but good.
There are other obvious sub-agents: the optic cortex, the enteric brain. Again, these are sub-agents in that we “feel” them, but they are a bit detached from (verbally expressive) “me”. For example, it is not rational, verbal me that feels hungry; it is my stomach that feels hungry. Verbal me don’t give a shit, until seperately-conscious-but-mute stomach-enteric brain say “no, you do give a shit” and “oh BTW, I control your sphincter”.
FWIW, I think my muscle cells are conscious too; just that verbal-me really has almost zero bandwidth to what is going on down there. Verbal me has no idea if the mitochondria are having a party or not; there’s only a few nerve/pain cells to monitor what is going on down there. Whatever consciousness my muscle cells or mitochondria have, I do not have any access to that. I can’t communicate with them, except in very minimal ways.
FWIW, there seem to be more than a few botanists who seem to be studying “plant consciousness”. I skimmed something about bean plants trying to grab a pole, and being aware of other bean plants that have already grabbed it. So this idea that “lots of things are conscious” is pretty hip and popular. My focus is to try to go deep, rather than try to prove that it exists.
Read more: Open this link: Linas Diary Part Ten-H with the LyX editor.
Leave a Reply