Wednesday, September 19, 2007

On dreams

Imagine there being no such thing as dreams. Everything that we experience would then be intimately tied to the real universe that we seem to be part of, and matters would have been a lot less complicated for the philosophy of consciousness. In fact, the argument for solipsism which relies heavily on the analogy of dreams would have been considerably weaker.

In the waking state, our sensory qualia seem to be virtually memoryless, providing an instantaneous representation of our own physical selves and the world around us. All qualia states which have memory like feelings, thoughts, etc. can be lumped together as emotional qualia, for lack of a better term. In digital electronics parlance, sensory qualia can be termed as "combinatorial".

Given the current world population of 6 billion or so, it is safe to assume that there are hundreds of millions of people, right now as you read this, who believe they are standing on solid grounds which we know don't even exist.

There obviously is no correspondence between the spatial coordinates in dreams and those of the waking state. However, it seems fairly certain to me that there is a close correlation between the flow of time in dreams and the waking state. Even if the scaling factor may not be exactly unity, I am pretty certain that a second of my dream time corresponds to some positive time of my bedroom clock. Certainly I don't believe anyone dreams in reverse play mode, nor in some other time dimension (which would correspond to a fixed time of my clock). This is because, our brains, which we assume to be the source of our dreams, still belong to the "real" world.

So it seems that my CBS can decouple itself from reality and stage a different set of consistent sensory qualia, and I wouldn't even know the fact during that time. (If I know it's being staged, it's called lucid dreaming.) Since our own consciousness is more faithful to our CBS (we always live in our CBS), we can make our CBS the reference universe. And why not? For a good portion of our time, our CBS locks onto the physical universe that we all seem to agree to as being the source of our consciousness in the first place, and we are led to believe that our experiences during these times have direct relevance to happenings in this universe. The concept of spacetime itself is tied to our CBS. It is possible that the "real" universe is nothing but a bunch of computer bits arranged in a one-dimensional array in some dusty basement of the Creator's abode. This is not too far-fetched from reality anyway, since when we see perfect circles or straight lines, in dreams or the waking state, the brain maps corresponding to such experiences are anything but these. Certainly even if there were a final physical theory, it wouldn't be able to explain the magic of why we are aware of our 3-D universe as such, let alone the qualia that we experience in it.

I had already discussed backward binding in dreams in this post. In the waking state, where the sensory qualia are more or less combinatorial outcomes of the sensory inputs from the various sense organs, we can think of the brain as a giant neural network that extracts patterns and features. Thus it is common to talk of a "grandmother cell" that fires when one sees one's grandmother. Of course, the "grandmother cell" theory is now discredited, since it is overly simplistic. But if we assume that information flows one way from the sense organs to the portions of the brain that extracts higher and more abstract features (like one's grandmother), then if one sees one's grandmother in a dream, is it enough if the last stage (corresponding to grandma) alone fires? Or does the entire chain have to fire just like how it would in the presence of the real-life grandma, to maintain consistency? If the latter is true, at least in principle, we should be able to tell what someone is seeing in his dreams by scanning his retinal cells (assuming back-propagation goes all the way).

As a chip designer, I know that combinatorial circuits are a lot more easier than recursive networks. If dreams occur because earlier stages take part, how do those stages "know" the main theme behind the dream in order to fire consistently? For example, when I am hearing someone talk in my dream, the audio has to get synchronized to the visual lip movements and has to emanate from the same direction as the speaker. Given that the lower level stages of the visual and auditory cortices don't seem to have any kind of link between them, how can participation of these lower level stages (if it's the case) be synchronized? I think that it is probably only the mid and high level stages that participate in dreams. The lower level stages are still coupled to the sense organs, but are shut out from the rest of the brain during dreams. Only when there is some abnormal external stimulus (such as the alarm going off), do the lower level areas interrupt the dream and force the consciousness back to the waking state. Sometimes, the output from the lower level areas gets incorporated into the dream, and the whole theme of the dream might change direction to accommodate this new intruding stimulus, in a consistent manner. This is quite astonishing, since how does the part of the brain that orchestrates the entire dream immediately come up with a sequence that would fit the external stimulus, all in real-time? For example, if you hear a baby cry, the dream would immediately summon a baby and fit it in with the rest of the characters and events.

A related question is, what is the evolutionary purpose of dreams anyway? Most of the answers given by psychologists have to do more with the emotional aspects than the sensory ones. I feel the latter are probably as important, if not more.

Also, it seems like dreams are capable of reproducing all kinds of physical and mental states that a person can possibly experience in his waking state. These are not just limited to sight and sound, but also smell, vestibular, etc. Strangely enough, I cannot recall any dream where I have experienced significant physical pain of any sort. I am not sure if others have either.

Also, in this post, I had referred to being drunk in a dream. But, after posting it, I realized that I didn't recall any dream where I'd been drunk either. I assumed that, unlike other senses, the state of being drunk is not derived from any sensory organ(s), but the large scale compromising of the function of the brain itself. So I assumed that it may be theoretically impossible for us to dream being drunk, because it is an unnatural state of the brain.

But less than a month ago, that is exactly what happened. I had a dream where I was quite drunk (even though I went to sleep sober). I am now wondering if the reverse can be true - can someone with a blood alcohol level that corresponds to being drunk, have dreams where he is completely sober? It should be possible to do a controlled experiment where alcohol is administered through IV through the entire period of the subject's sleep to keep it at some level, and ask him to recall his dreams. If in his dreams, he is completely sober and drives vehicles normally, then there is something quite mysterious. This would imply that a blunt-force compromising effect of alcohol can be completely negated in the CBS, if the brain decides to.

This, and the overall meta nature of dreams leads me to believe that there is more to it than meets the eye, when it comes to the brain. Even if we leave the hard problem aside, and assume that other beings are zombies, we can still assume they dream. (Since when asked what dream he had, a zombie would answer no differently from a conscious human.) And the fact that the brain seems to have seemingly infinite power when it comes to staging the entire set in real time, as well as negating the effect of blunt acting drugs, I am wondering whether there is an infinite-complexity engine powering the brain. With such an engine, a small part of it still equals the whole. Maybe that's the secret behind consciousness too. In terms of information theory, the entropy of the brain can be infinite. The entropy of purely random phenomena is also infinite. Quantum mechanics, which reconciles to the randomness of nature, and which led Einstein to proclaim "God doesn't play dice with the universe", may quite well be the key to the hard problem as well as the seemingly infinite capacity of the brain, even a zombie one.