Comment by ozy | Hacker News Reader

Comment by ozy | Hacker News Reader

Consciousness is the way your brain self corrects. That is my personal theory.

On a few time scales you brain makes plans. It uses beliefs and a model of the world to make those plans. It stores those plans and the actions it has taken. And then checks if actions had their intended effects, and if plans reached their intended goals. Updating its model of the world, its beliefs, and what useful actions are.

This is a constant process reflecting upon the current situation, and upon the recent past of actions and plans. It is like an observer observing itself. This is the root of consciousness, I think.

Probably abstract thinking and language, enough intelligence, and a theory of mind, leads to self awareness.

If this is correct, we can formulate some necessary parts for a system to be conscious:

* recognition of the environment * a model of the environment * ability to reason about actions, their consequences, and how to achieve goals * memory of past plans and goals and outcomes * a process of executing actions, making new plans, and learning from the outcomes

Self-awareness and consciousness are different things. Sometimes when you wake up you don't know who you are, where you are, what you are, but there is a conscious experience. Consciousness is much more fundamental than your level of reasoning allows to explain.

Here's a list of things that you are adressing and that are not dealing with the hard problem of consciousness ('easy problems'): https://en.wikipedia.org/wiki/Hard_problem_of_consciousness#...

An observer observing itself, would that not feel like something to the observer?

What you describe is when the observer is disconnected from longer term memory.

OK, let me rephrase my objection.

"This is a constant process reflecting upon the current situation, and upon the recent past of actions and plans. It is like an observer observing itself. This is the root of consciousness, I think."

Your description seems to fall under "the ability of a system to access its own internal states" and/or "the reportability of mental states" from the 'easy problems' section of the wiki pages I linked to.

Both could be possible without consciousness at all. We could imagine (or even program) a robot that in response to external stimuli modifies its state. It could also have second order algorithms, eg. it could check what state it's in and if it remains in the same state for too long, the state changes randomly. It would be the same as 'observer observing itself'.

This functional explanation in any way doesn't lead to the phenomena of consciousness, which would require the robot to have some specific first-hand 'experience' of the world and itself. That's the difference between the easy and the hard problem of consciousness.

But what if the consciousness is just a notion for the systems that reach certain level of complexity, i.e. can solve the same easy problems as human brain can. I mean there is no qualitative difference between programmed robot and us, but only quantitative. So physicalism accepts your objection and go on with programming robot until it can solve Turing Test. Will you accept that the robot is conscious now? Why can’t “first-hand 'experience' of the world and itself” automatically emerge if the system is complex enough and aimed to survive?

Hard problem solving is not required to create consciousness. Humans can’t solve hard problems but they make children conscious meanwhile human children grown by animals become animals.

May be hard problems of consciousness is like describing feeling or summary of all feelings. We can feel but can’t explain. Meanwhile every our feeling has underneath chemical reactions.

> Hard problem solving is not required to create consciousness. Humans can’t solve hard problems but they make children conscious meanwhile human children grown by animals become animals.

You're serious? :) Humans can make children conscious? They barely can control their instincts, most of the time they are not aware what and why they are doing. We have no clue how are we working, how reality is created.

To say that consciousness "just" appears automatically when some complexity is reached is just magical thinking or, at the very least, explaining the problem away. You would have to answer why, how, when is the critical point, what's the difference between conscious and unconscious robot etc.

Maybe this thread is not yet dead, seeing a reply a few hours ago. I did intent to reply, but life happened I suppose.

I am curious about your thoughts. Here is why I think it does address the so called hard problem:

The goals of these plans are evaluated by simulating the world and the body and future mental states. So the brain can measure, or "feel", the response, and thus judge the plans and compare them.

This constant self reflection and self prediction is what feels like something to the thing doing it. Why would it not?

---

But in general I have some objection to the "hard problem".

We cannot proof something to be conscious. Probably we never can. The best we can do is find good proxies indicators of consciousness and good lower bounds for what can potentially be conscious, and what cannot.

But the opposite is also true, we cannot proof there is such a thing as the hard problem of consciousness. For all we know every program we have every written has had some kind of experience as it was running.

I agree it is reasonable to assume this is not the case. But I don't think you can proof this is not the case. Therefor the hard problem might, or might not, actually, be a problem.

To predict no 'feeling' need to be involved. You can have prediction without consciousness (any program that simulates some environment to check whether some condition will be met, eg. collision detection in 3d games) and consciousness without prediction (there probably are people with brain damages that are conscious, but have limited imagination capacities).

You can see color of red and don't predict anything at all, while still being consciouss. Predicting is just a narrow part of our conscious activities, however all of them are conscious.

So cosciousness is something much much more basic and fundamental than any high order cognitive function.

> We cannot proof something to be conscious. Probably we never can. The best we can do is find good proxies indicators of consciousness and good lower bounds for what can potentially be conscious, and what cannot.

We can't prove it because we have no idea what consciousness is, even on a philosophical level. Moreover if consciousness is basic property of reality, in a way parallel and coexisting with another property - matter, as panpsychists suggest, than it indeed would be impossible, as this dimension would be completely 'invisible' from the matter point of view.

When you're talking about 'proving' most likely you assume existence of some reproducible causal chain in the material, 'intersubjective' space. So you can't 'prove' consciousness in this paradigm by its very definition.

Of course one can insist to throw away any theory that's not provable in our paradigm, but does it bring us any closer to the understanding of consciousness? Also, is there anything inherently wrong with neutral monism or naturalistic dualism? These theories are not internally inconsistent, moreover they are also not inconsistent with anything current physics states - they just provide a wider view that at least somehow addresses the notion of consciousness.

It think you took predicting a bit too far and singular. It is not just predicting that is conscious, and it is more simulating or considering or even pattern matching.

This is what I think: Consciousness is continuously self observing past actions, past plans, actual outcomes, planning new actions and predicting future outcomes. That feels like something to the thing doing such a process, because practically all of those steps involve simulating/predicting/checking what something means to itself. Our brains do this because it is a self correcting mechanism.

> You can see color of red and don't predict anything at all, while still being consciouss.

I don't think so, when you see that color, you cannot help observing its context, what it means, check if something needs to be done, etc. All that is what I meant with predicting. Maybe the color is of no significance, but the only way you figured that out was by "predicting".

---

Even if consciousness is 100% material (as I suggest), we can not prove it exists in others. We don't need consciousness as another dimension for it to be improvable.

As a matter of fact, if it is another dimension, then our brains are "interacting" with that dimension. So the hypothesis predicts that certain configuration of matter can have some kind of interaction with that dimension. That should not be hard to proof.

But since we have no such proof, and more mundane hypotheses, we must conclude that the other dimension hypothesis is extremely unlikely.

Another line of thinking on the dimension hypothesis is this: we have brains as small as from C. Elegance with 300 neurons, to ants, 0.25 million neurons, to mice, 70 million neurons, to humans, 86 billion neurons. At which neuron count does consciousness come in? What if we simulate the full neurons of any of these, will it behave the same? Will it be conscious the same?

Reply on news.ycombinator.com