Tuesday, August 02, 2005

The Multiple Self

If you've spent any amount of time trying to "improve yourself" in any way, you've undoubtedly experienced some frustration, in that you discover your "self" isn't as much of a part of you as it would appear. It seems bizarre - "you" want "yourself" to do something (or refrain from it), but "yourself" goes ahead and does whatever it wants. Often, "you" may rationalize your behavior in context, only to be later exasperated by your lack of "willpower".

If that is the case, I have some good news for you. It may sound a bit snake-oily at first, like I've got a product to sell or something to gain from convincing you of the idea. And I've been tempted not to try to explain this here, because why should I explain something that could give me a tremendous advantage over people who don't understand it?

On the other hand, since I've come to understand the secret, I find that virtually every form of self-help (that has any competence at all) reflects at least some portion of this secret, and many of the teachings of Zen and other mystic traditions point to it as well. So, in a sense it's not really a secret at all. Everybody's been practically shouting it from the rooftops for centuries, and yet you can't give this secret away. This is mainly because whatever you say about it, people think they already understand it, and so never really dig down to the real truth of it, and the profound implications it has for every aspect of life and the human condition.

So, I probably could turn this idea into yet another school of self-help of my own variety, but I don't really see any reason to do that at the moment. So here it is, a freebie - the ultimate answer, not to life, the universe, and everything, but the ultimate answer, I think, to the nature of the human condition: "You" are not "yourself" today.

For that matter, you were not yourself yesterday, and you will not be tomorrow. You never have been, and never will be, because "you" and "yourself" are distinct neural subsystems which do not overlap.

What does that mean? Why should you care? And how is this a deep and mystical revelation, exactly?

Who are you, anyway?

Wave "your" hand at the screen. Do it now.

Who waved your hand? Was it you? Could you have? Do "you" know how to wave your hand? Of course not. "You" don't know how to do anything, but "yourself" does.

Now, if you are determined not to learn anything new, you will have already conveniently reclassified these ideas as being about conscious/unconscious mind, or left brain/right brain, and if you want to do that, by all means do so, but you won't actually end up with any new knowledge that way.

In fact, that's precisely how I managed to read so very much about these subjects for so many years without actually ever understanding what this really means about life and living. Abstract ideas like conscious and unconscious mind don't help you do anything differently than you already do. But if I succeed in actually conveying the real experiential truth that lies beneath concepts like "unconscious mind", then you can begin to change your life in the same way that I'm already changing mine.

To really understand, you need to first understand that you are an animal. Most of us humans pretend our entire lives that we are something other than animals, and as a result we think our "animal nature" is something you can just ignore or somehow transcend -- preferably while ignoring it. We enter the false dichotomy of "man or beast", when the truth is actually "man and beast." We are not one - we are two. And the one of us who thinks he's running things is really just a recent software upgrade that runs atop a highly sophisticated operating system that's already had millions of years of performance tuning -- and can run just fine without you.

That's right. "You" are just a subroutine, and a recently-added one at that. You're like a user-mode driver that gets access to certain kernel data, but you only see and control what the kernel lets you. You have no direct access to the kernel's process space, but you can make calls into it, and you get notifications from it. The bulk of your nature as a human lies entirely outside your process space, outside your ability to directly perceive or control.

(If you find yourself thinking here about the famed "10% of your brain", well, you're not wrong, but it's important to understand that the truth of this and lots of other ideas (like conscious vs. unconscious mind) is quite a bit deeper than the sound bites.)

Now, when I say you're just a subroutine and that your animal nature is the kernel, this doesn't mean that we are robots or machines or that we don't control our actions. Far from it. I mean, however, that we are deluded when we think we directly control our actions, and therefore ascribe intention to our actions that doesn't exist.

Why does anyone do anything?

In fact, we frequently do things for reasons that are entirely opaque to us, and then make up reasons later to explain them, because nobody wants to admit that they don't know why they did something. Nonetheless, none of us know, because it's not in our process space to know why the kernel switches in this process at this time, and that process at another time. We can reverse engineer things, and we can use our "supervisor calls" to inject new programs into kernel space, sure, but we don't run in kernel space and we never will.

And yet, we all mostly go around pretending as if we did run things in our mind and body, which then leads to all sorts of screwed-up thinking - "delusion and ignorance" as the Zen Buddhists call it. We mistake kernel notifications for our own thoughts. We think our actions somehow reflect on us, when in fact they may reflect nothing more than a poorly-written script that the kernel is running. This is like trying to eat pictures of food: it might fill you up, but it's ultimately unsatisfying.

The thing you need to understand is that it's not a question of "using only 10% of your brain". The point is, you are only 10% of your brain. The rest of your brain is bigger, smarter, and better-educated than you, because it can learn things you don't even know you're learning, faster and better than you. It can actually do things, for one. You can't. "You" are really just an I/O filter, in a way. You can control everything because you are a software hook that controls what meaning things have. By defining the meanings of things, you can get the kernel to do whatever you want. However, if you are confused about any of this, you will feed the kernel garbage meanings, and, well, you will get garbage results.

For that matter, if you are confused about this state of affairs, you will try to "do" things yourself, and because you're only 10% of the processor, you will only screw things up by getting in the kernel's way. Don't do that, it doesn't work. The hidden meaning of "Just do it" is, "establish the kernel's operating parameters and then get the hell out of its way."

Shortly before I figured this out, I read about an interesting computer game called NERO. The idea of this game is that you get a team of robots and you have to train them to fight, by designing training exercises in which you establish goals like maintaining a particular distance from the enemy, and lay out a battleground for the robots to learn in. The interesting thing about this is that the robots basically evolve neural networks that seek to meet the goals you set, and they learn how to do this from experience, not by being programmed. The neural nets that evolve have no self-consciousness, no awareness of how they accomplish the task. Wiring just forms and improves. It would be a mistake to look at the robots' behavior and conclude that there is intelligence or intention there; it is just formation of neural nets in response to environmental stimuli coupled with goal-oriented feedback.

It's not "you" that learns...

And that is precisely what the other 90% of our brains are up to. Formation of neural logic networks in response to environmental stimuli and goal-based feedback. If you are not actively participating in this process, what you get is well, whatever you get. Pretty much, you're going to have random feedback loops, because whatever you initially train the networks on (e.g. childhood) is going to get used as the basis for evaluating later input and as the basis of your goals. Your subconscious is like a random number generator whose output is fed back into its input, only there's goal seeking involved. It's more like it's a random number generator that's striving to repeatedly reproduce whatever numbers it generated before.

At least, it works that way without your input. Your subconscious needs you to establish meaning, and goals, and values, because without those things, all it knows how to do is maintain the status quo, or respond to what other people put into it. If you're not doing your I/O interception duties, it's like the captain is asleep and the entire ship's crew fights to the death to carry out the captain's last orders, whatever they were, even if he was half-asleep (or a child) when he gave them and they no longer make any sense. Conversely, if you over-manage yourself, then you're like the captain going around and trying to do everybody's job, and that just isn't workable either.

After stopping and reading over what I've written so far, I'm a little frustrated, because I still seem to be slipping away from the heart of what I want to convey. It's too easy to come away from what I've said without realizing just how incredibly small "you" are, in relation to your own brain. It's like that old system administrator joke, "Go away, or I shall replace you with a very small shell script." Only it's not a joke, and you are a very small shell script that thinks it's the data being piped through it.

Thoughts Are Data – And The Numbers Always Lie

You see, your thoughts aren't you either. They're just data being piped through "you". What you "think" about things is mostly just regurgitation of patterns captured by the kernel as part of its massive imitate-and-evolve subsystem. This is not a bad thing, but a good one. Have you ever realized how little control you have over your thoughts? Can you imagine how bad life would be if those were really your thoughts?

Imagine a person listening to the radio. Now imagine that this person has never seen a radio before, and thinks that the radio is talking to him or her directly. That is, he or she thinks the announcer is talking to him, all the singers are singing to her, etc. For example, when the radio plays the song "You're So Vain", our imaginary person thinks the song really is about him!

This is almost as bad as the situation the rest of us are in, but not quite. At least our imaginary listener thinks someone else is talking to him. Most of us, on the other hand, think we are the announcer on the radio in our heads, and that we're announcing live, when in fact most of it is previously-recorded, and all of it is being piped to you straight from the kernel.

So, don't be frustrated by your thoughts, because they' re not "you" either. Just because the kernel sends you a heap of worry, fear, anger, or other crap on sys.stdin doesn't mean you have to send it on to sys.stdout. Until I understood that, I was under the mistaken impression that fear and worry, hurt and anxiety, disappointment and regret were all real things. But they're not real! They're just data. Heck, they're not even data about real things. They're data about previous conclusions drawn about similar things! Sometimes, they're even data about erroneous conclusions previously drawn about similar things.

It's not enough to know; you must also rewire!

Don't just understand this intellectually. I've "understood" all these things intellectually for many years and it was useless. The question to ask is, "How do I implement things with this knowledge?" How do you drop input data on the floor? How do you insert new data into the output stream? How do I make supervisor calls to the kernel? How do I edit the scripts the kernel is running? Until I started asking these questions, I wasn't able to do much besides dance to the tunes on the radio in my head.

And don't get me wrong, after weeks of playing around with this stuff, I'm still no superman or Zen master. But I have managed a few very interesting hacks. For example, a few weeks ago a certain situation led to me feeling very bad. Intellectually, I totally knew there was no reason to feel bad, because what happened had nothing to do with me. Emotionally, though, I was a wreck.

Suddenly, I had a flash of insight: these are two different neural networks. The intellectual understanding and the emotional response were networks that evolved at different times in my life, under different circumstances. They were therefore not connected, except through their mutual activation in the current circumstance. Therefore, I experienced each network's output as a full and distinct input, but the "emotional" net had no way to receive data from the "intellectual" net, in order to moderate its output. This led to an experience of conflict, in which I could try to suppress the output of the emotional net, given the data from the intellectual one, but this couldn't and wouldn't stop the emotional input from coming in my input pipe.

As soon as I could see that, it was obvious what I needed to do: pipe the output from the "intellectual" net into the "emotional" net, instead of trying to integrate the data downstream in the "consciousness" process. And literally, as soon as I imagined this, the two upstream networks integrated, and the need to feel bad went away. I still felt bad physically, in my body, so I "shook it out" and it went away. (It appears that shifts in glandular output and neurotransmitter states are used as a crude system-wide state machine to aid in sorting input and output, so even after you adjust an upstream source, you may retain some kinesthetic "pollution" downstream until you garbage collect it.)

Many Circuits, Loosely Joined

Now, before I go further, I want to explain that the "emotional" and "intellectual" networks I just mentioned were not my entire emotional or intellectual being. That's precisely the sort of large-scale behavioral integration that our brains do not have by default. I integrated two isolated "understandings", each of which was a simple script to assign meaning to a certain class of events. In programming terms, each of these nets could be considered a "business rule"; just pattern recognizers that fired off to send "me" their analysis of the situation. It's just that one of those rules fired off a "knowing" and the other fired off a "feeling".

So, the fact that I did this one particular edit of my brain's rule system does not now mean that intellectual understanding is now integrated with all my emotional impulses. During early life, we write a lot of scripts in our brains that are not abstracted or reused in any significant way. Later scripts may abstract or absorb chunks of previous scripts, but they often do so in a downstream way; that is, they take their input from older scripts and output commentary on them, but this commentary doesn't necessarily have any effect on our behavior or feelings, and therefore leads to the experience of inner conflict. So, we inherit a lot of "legacy" code that desperately needs refactoring.

I'm still experiencing these conflicts from time to time, and it's not always easy to integrate the processes. Sometimes, an emotional network offers some kind of input or is linked to a goal that's important, so even piping the intellectual data into that network's input doesn't modulate its output much, and I have to do more extensive refactoring. This can be a pain to try to do while the net is still running and making you feel bad! It's even worse when the net uses bad feelings as an input indicating that you're in a bad situation; once you get a loop like that started it's pretty tough to get out of without a reboot (like going to sleep, or a surprise interruption that forces the whole process to swap out, as it were).

However, I'm starting to get a kind of catalog of "design patterns" together to make the process a little easier, like establishing state machines to evaluate rules over longer time periods, rather than needing to immediately feel a certain way at the first evidence of a circumstance that potentially matches a pattern.

Command Mode

So how do you do this? How do you edit rules, pipe one net to another, make a supervisor call? In the same way you waved your hand at the screen, several paragraphs ago. You imagine it, in command mode.

There are a lot of books out there about creative visualization and imagining what you want and all that, but there are very very few that even hint at the need for the command mode. You see, imagination is like a scratchpad; it's working memory. If everything you did with that working memory were a command, you'd say and do everything that came into your head. The command mode is like a modifier that says "actually do this", or "make it so". It's quite literally metadata that describes what to do with what you imagine.

Point your finger at the screen. How did you do that? Do it again. Try something else. Make various motions with your body. Now just think about making the motions. What's the difference between thinking it, and doing it? That's command mode.

If you play with it for a bit, you'll discover that command mode is easiest to use with a destination. It's likely that when you decide to point at the screen, you're visualizing the endpoint, where your finger is pointing at the screen, rather than visualizing all the motions in-between. Similarly, when I integrated those two networks I spoke of, I just visualized them integrated, and I happened to engage command mode even though I didn't really know what it was yet.

This is what I mean about not getting in your own way. Your operating system has enormous parallel processing power, whereas "you" are a serial processing filter. If "you" try to get involved in the "how" of things, you will just interfere, because you're a bottleneck on how much "yourself" can do! Really, saying that your conscious mind is 10% of your brain is probably grossly exaggerated, because we mistake a lot of the things the kernel does as being part of our consciousness. 5% might be a better estimate.

But even that doesn't truly show just how bad it is to try to shove the entire system's I/O load through the consciousness filter. Even if "you" had a whole 10% of the brain to play with, that 10% is set up for simultaneous use in serial processing of experience. "You" just don't have the pipes and peripheral processors to handle that load, but the operating system does.

Those pipes and processors will never be directly accessible to you, for much the same reason that intellectual understanding of your behavior that comes after those behaviors is ingrained usually has no direct effect on the behavior. The newer, more sophisticated abstractions are there to process outputs from the older, more "primitive" subsystems. We can reason about our inputs, but our inputs are not driven by our reasoning.

Of course, within the scope of networks wired after we were born, we have a lot more flexibility. Among those networks, we can rewire older nets to include input from newer ones. But at the periphery of the brain, these networks are simpler, more hardwired, with less "meta" wiring capability.

This isn't a problem as such, because the brain has plenty of "meta" wiring to go around, and you can do a lot up there that can't be done directly. However, that meta wiring is also part of an older layer than "you", so you still can't access it directly. But you can access it via emulation mode, which is another variation on command mode.

In fact, I've been finding that there are a lot of command mode variations. I'm rather reminded of a novel I once read, where a computer programmer traveled into a dimension where magic worked, and he ended up writing a magic compiler. Unlike other magicians who worked out their spells in somewhat random fashion, he developed abstractions, a subroutine library, and an orthogonal command system.

For example, he would say things like "backslash light enter" to cast a light spell, where other magicians might chant some kind of short poem about light. As a result, he became one of the most proficient wizards in the land, because his powers were extensible and composable in ways that other wizards' were not.

In the same way, most of us don't really know how to use our own brains in a systematic way. We give them commands like we were a cat walking across a keyboard: every now and then we end up with something syntactically valid, but semantically... questionable.

So I'm beginning to understand now why Bandler and Grinder speak so disparagingly of the conscious mind, and talk about the need to engage the unconscious mind when training people, meanwhile distracting their conscious minds so as to keep them out of the way. I suppose that beats trying to actually get across to someone all the information that I just wrote in this article, and further get them to actually believe it, just in order to teach them something else.

WYFIWYG: What You Feel Is What You Get

And on that note, I'm going to tell you one more secret-that's-not-secret before I go. A lot of self-help books mention this, but I'm going to try to explain how and why it's so important. They usually tell you to suspend judgment, to just "try the ideas" without getting caught up in your ideas about whether the book's approach will or won't work. This is not, as it happens, because of some sort of faith-healing type thing, so that they have an excuse in case their stuff doesn't work. (Well, maybe with some of them it is.)

The actual issue is that if you think you're the announcer on the radio in your head, you'll probably believe whatever it says as the gospel truth, and act on it, regardless of whether it has anything to do with your goals or what you're trying to accomplish. The next thing you know, you're believing stuff like "This won't work," and you're suddenly adding metadata tags like "doesn't work" and "don't believe his lies" to the audio and video streams you pipe back to the kernel. And the kernel, since it has no reason to doubt you, will then act according to the metadata you give it.

You see, the real secret about command mode is that there is no command mode. It's really more like a tagging system, where the kernel acts according to the tags you put on stuff. You can tag things as "just pretending", or "not important", or any number of other things, and the kernel does whatever you've set it up to do for those tags. That means you really do need to watch what you think, and learn how to browse your own thoughts at -3. Otherwise, you can clog your kernel with an awful lot of crap.

There is so very much more to all this, more that I've seen and discovered and experienced and want to tell you about, but many books have been written about it already, and there's no way for a single article to compete. I'm not going to promise to elaborate further in future posts, although I'm sure I'll want to try from time to time as new insights or possibilities occur to me. But I'd certainly be interested to hear if any of you pull off any cool or interesting hacks using the information in this article. There is an entire field of knowledge waiting to be discoveredout there... or should I say, in here? (he said, gently tapping his forehead.)