Kharakov

However it can be reduced to an extremely complex cause and effect relationship.

In addition, that the system has to be in place for the behavior to emerge implies that the system is causally linked to the behavior (even if it is not a direct cause- perhaps the system was caused by whatever caused the behavior).

Kim o' the Concrete Jungle

Um, no. It most certainly does not assert that there is some supernatural phenomenon at work. In fact, emergence is what makes naturalism a viable point of view, philosophically, for me. Emergent properties are not supernaturally caused, they are naturally caused, by the same set of natural laws that govern each component part of a complex system. But you cannot predict an emergent property simply by looking at all those natural laws. You will only discover emergent properties by allowing the system to run and seeing what happens. (Hence the term, emergence.)

Emergence is the reason why meteorologists can't just predict the weather by applying a few simple formulas. Instead, they have to use some serious computer power to create models of weather patterns to see what they are going to do.

Emergence is an important concept for the philosophical naturalist, because it shows us why scientists can't explain everything, even though everything has natural causes. Galiel gave the taste of salt as an example. While you can naturalistically explain the taste of salt by saying that molecules of salt react with certain tastebuds and not others, thus sending a distinct pattern to the brain, you cannot explain why we don't perceive the taste of salt that way. The personal, human experience of tasting salt is an emergent property of the system.

So why is all this a problem for a deterministic world view?

Up until the nineteenth century, it was supposed by some deterministic philosophers that if you had a perfect, god-like knowledge of the universe, down to the position of the smallest particle and every natural law that effects it, then you would be able to perfectly predict everything that happened in the universe from then on. If this is the case, then logically, the future must necessarily be pre-determined, so there can be no such thing as free will. Ierellus's question "Is structure anti-possibility?" would have to be answered in the affirmative. But on the positive side, one would only need to get a better understanding of the natural laws of the universe to be able to predict the future.

This comfortable deterministic hypothesis was nice while it lasted, but then the wheels started to fall off. Scientists began to discover things that are fundamentally not predictable.

The first such thing was the three body problem. Using Newtons laws, it is possible to predict the future position of the planets -- but only up to a point. Because while you can easily predict the gravitational relationship between the sun and the planet, you cannot introduce a third gravitational effect (such as might be caused by another planet) into the equation. It is now mathematically proven that there is no solution to the three body problem. After a certain point, we cannot accurately predict the position of a planet, because the accumulated gravitational effects of other planets will have altered its position. This is not just a matter of lacking information, it is truly unpredictable, because the three body problem is mathematically insoluble -- not just practically insoluble, but intrinsically insoluble.

Heisenberg also undermined the deterministic hypothesis with the uncertainty principle, which states it is impossible to know both the position and the velocity of a sub-atomic particle. And again, not just practically impossible, but intrinsically impossible. Then along came chaos theory, and the study of complex systems, and the realization that there are emergent properties that -- even though they are still naturalistic -- cannot be predicted by looking at components of a complex system in isolation.

Since such areas of unpredictability exist, it is not, therefore, possible to predict the future course of the universe. Nor would it be any more possible if we had perfect knowledge. So the deterministic hypothesis, as I described it above, is undermined.

Perfect predictability is not a property of the universe we live in. Even perfect knowledge would not bring perfect predictability about. So you cannot logically conclude that the entire history of the universe was predetermined by the initial starting conditions. It could just as easily be that, if you froze the universe at one point in time and made five exact duplicates, then each of the five might be entirely different from each other at the end of time. After all, why -- apart from wishful thinking -- can we entirely rule out randomness from our universe (especially at the quantum level)? It is already clear that naturalistic rules only work up to a point.

Feather

That's fine--this still does not mean the system in question is not deterministic. To make such an assertion one must propose that "predictable" is the same as "determined." Semantically this can be accomplished by introducing a new word to cover what is meant by "deterministic" (or "predictable", depending on which way you force the equality of "predictable" with "determined").

The last sentence in the quote is, again, merely a way of restating that "we experiment." It doesn't introduce any new concept, and I would argue that it in fact confuses a perfectly well defined existing concept and further compounds the error by attaching extraneous significance to the new definition.


Um, "human experience" is "just" a bunch of chemicals reacting and electrical impulses firing. The term "experience" is a label humans have supplied to the process. I don't see how caling "experience" an "emergent property" assists in understanding the system any further.

Okay, I think I've missed something else, here. Are there formal definitions of "anti-possibility" and "structure?" It seems to me that what is certainly is possible. If, on the other hand, "anti-possibility" is equivalent in some way to "undetermined," then I can understand the requirement to answer in the affirmative given the rest. But this seems to me to restrict the definition of "possibility."

This is incorrect. There is no analytic solution to the n > 2 body problem. This mean it is not (currently) possible to provide the trajectories of the planets in a closed functional form. However, the solution can be expressed as an infinite series of terms, and therefore be found to any desired accuracy.


But the state that describes the sub-atomic particle can be known to any degree of accuracy, as above, when it cannot be expressed analytically. Moreover, any number of null points are determined by the state of a given system on the quantum level.

Quantum Randomness does not equate to indeterminate.

Kim o' the Concrete Jungle

Feather:

Here are two related definitions of determinism offered by Brian L. Silver in his book, The Ascent of Science.

And the related definition:

This is sense in which I use determinism here, and the way I've seen it expressed many times by other people. So it's not just me perversely equating predictability with determinism. Predictability has always been a central part of the deterministic hypothesis as conceived by Laplace and friends.

If you have some other definition of determinism, please share it, otherwise we're going to be talking at cross purposes in this discussion.

I've been addressing an argument that looks like this:

(1) A system governed by natural laws is predictable.

(2) The universe is a system governed by natural laws.

(3) Therefore the universe is predictable.

The basic point I've been trying to make is that the first premise of this argument doesn't hold up, because there are things like emergent properties that cannot be predicted, that are fundamentally unpredictable, even though a system is governed by natural laws.

Kim o' the Concrete Jungle

Emergence isn't some crackpot theory I'm putting forward, it is a fact that theory needs to explain. It is a fact that simple cause/effect models have failed to explain, which has led some people studying complexity to speculate that complex systems are irreducible. That there is such a thing as an emergent property is, I think, broadly accepted (to the extent that "emergence" has become something of a buzz word).

If you call it an emergent property, it at least stops people from poking around in people's brains trying to find the "personal experience" node. When the experiences are understood to be emergent properties, one no longer expects to find particular nodes or neurones associated with creating the personal perception (or qualia) of an experience. These are properties of the whole system -- not of any one part of it.

And again I will say that, philosophically, the idea of emergence is crucial, because it explains something that would otherwise be inexplicable. Namely, if the universe is really just made of wave/particles bumping around according to a few simple physical rules, what is the source of this perverse way that human beings experience the universe? Why do we see trees and buildings and other people when these things are actually nothing more than collections of wave/particles? And when we can say that the source of our perception is the sum of all the physical processes going on in our brains and bodies, and the myriad of subtle ways that these interract with each other -- in other words, emergent properties of the whole system -- then we don't have to resort to supernatural answers.

I don't know how else to put it. An emergent property is the sum of all processes within a system, and the way that these processes subtly -- and sometimes surprisingly -- interract with each other. It is just a feature of these emergent properties that they cannot be predicted by looking at the processes in isolation. If you want to know why that it is the case, you're going to have to ask someone more knowledgeable in complexity theory than I am.

Maybe I interpreted Ierellus's original question wrong, but I thought he was asking whether structure is restrictive -- whether, for example, the structural form of his piano would restrict him to a very narrow range of expressive possibilities, as compared to the universal set of everything that is musically possible.

I think you've failed to grasp the enormity of the problem. Your proposal of repeating the two body calculation for each successive body is not a solution to the three body problem, because all you've done is set up a feedback loop.

Imagine that you want to measure the position of the earth in relation to the sun very far into the future, when you know that the gravitational pull of Venus upon the Earth is going to have a significant effect on the Earth's position. Here is what happens:

(1) You do the standard Newtonian equations between the Earth and the Sun, and calculate that the earth will be in a certain position on your given date.
(2) You do a second calculation between the Earth and Venus, and correct the final position of the Earth by the required amount.
(3) However, you have altered the position of the Earth (to take Venus into account), so the numbers you fed into the first equation are now incorrect.
(4) You must repeat the first calculation between the Earth and the Sun, to take the new data into account.
(5) But now you have changed the predicted position of the Earth for a second time, the numbers you used for the equation between the Earth and Venus are wrong, so you must redo that calculation.
(6) But then, hang on, the first calculation is wrong again...
(7) Repeat steps one to six forever. Each time you will come closer to the "correct" answer, but you will never reach it. You're stuck in a negative feedback loop.

There is no solution to the three body problem. Not even a hypothetical solution. The three body problem is mathematically impossible. That has been proven, mathematically.

[ October 31, 2002: Message edited by: Kim o' the Concrete Jungle ]</p>

Feather

Those definitions are exactly what I have in mind when I think of "deterministic." But notice that the word "predictable" is qualified by "in principle." All the laws of nature so far discovered fit this description quite well.

I understand the argument you are addressing. Moreover, I'd agree that as a matter of practice (1) is not in general true (however, it is in principle true--otherwise our models would be grossly in error and they are not). For example, every time you use any mechanical object, you engage in a product of predictability.

Where I strongly disagree is this concept of "emergent properties." Again, as you've used it in your post, it seems to me to be little more than the application of new terminology to a rathe mundane concept--namely "stuff happens."

Would you consider the pressure and temperature of a fluid to be an emergent property? No individual molecule of a fluid has either. Yet, applying very basic Newtonian mechanics (which are observationally true and entirely deterministic in nature) to an ensemble of particles, one can "predict" the properties we label "pressure" and "temperature." I'd venture a guess that "pressure" and "temperature" are two properties you'd not call emergent.

But another argument occurs to me. It seems to me that the entire universe is itself an "emergent property," for there is no reason to suppose that the universe should exist in the form it does given the laws as we understand them. In my mind this makes the term "emergent property" tautological at best, and trivial at worst. It would be like describing a triangle as "a geometric figure composed of only three angles" and then claiming one could not possibly have predicted the "existence" of the one from the other.

So to say a system has "emergent properties" is to say the system "does stuff."

Kim o' the Concrete Jungle

[continued]

Now let us extend the example of the negative feedback loop I gave above, concerning the position of the earth very far into the future.

It is not just gravity from Venus that will interfere with the future position of the earth. All the other planets will have an effect. So we can add them one by one to our equation. Now, logically, you would expect that your equation would still be a negative feedback loop -- that no matter how many planets you add to your infinite round of calculation, that it would remain reasonably well behaved and that each time you would get just that little bit closer to the real answer (without ever actually reaching).

I wouldn't bet money on it though.

You see, whether a complex system -- like your never ending calculation -- is stable or chaotic is one of those nasty, unpredictable emergent properties. It could be that for some given starting values and numbers of planets, your calculation will be a nice, stable, negative feedback loop, such as I described. But it could also be that other starting values and numbers of planets will form a positive feedback loop. In this scenario, each round of calculation will take you further and further away from the truth.

The point is, that you will never be able to tell, just from looking at your starting numbers, whether your endless calculation will be a stable negative feedback loop, or a chaotic positive feedback loop. In complex systems, chaos is an emergent property. Emergent properties are unpredictable.

Kim o' the Concrete Jungle

I think this debate is starting to get a bit confused, so I will try and state things as simply as possible.

An emergent property is any property that arises from the interraction between components of a complex system.

A complex system is any system where the component parts interract with each other.

Not all systems are complex. Rust is not complex. Free oxygen atoms combine with free iron atoms, and the ass end falls out of your car. This is not a complex system because rust molecules don't interract with each other and change each other's behavior.

The emergent properties of a complex system are not predictable, because when you look at each component part, there is nothing in that part that's going to tell you how it will interract with the other parts.

The reason why I'm banging on about all this is that human beings and human societies are complex systems. So there is emergent behavior in these systems. There are complex relationships between the components of these systems.

For example, you might get the impression that someone is an idiot, so you treat him like an idiot. And because you treat him like an idiot, he starts to behave like an idiot, just to spite you. So your impression that this person is an idiot is reinforced. This is a complex relationship between your idea of someone, your behavior towards them, and their behavior towards you, with all sorts of interractions and feedback loops forming.

My basic argument is that human behavior (and other complex systems), are not predictable -- even in principle -- because the have emergent behaviors that are not, in principle, predictable. You can point to any number of psychological principles that form components of this complex system of human behavior, but you cannot predict how these components will interract with each other. You cannot predict what one particular person will do in a particular situation until you put them in it. So you cannot say that behavior is deterministic.

Kim o' the Concrete Jungle

Well, here's the thing. Wherever you can isolated simple cause/effect relationships you can expect our scientific principles to work very well. And if you think about it, so much of experimental design is aimed at excluding factors other than the ones you want to look at.

But, if you look at things in this way, you would also expect that our scientific principles and formulas would work only within certain limits. If, for example, you try apply evolution theories to human social behavior, you end up with a wrong-headed theory like Eugenics. It's like our three body problem. Once you start introducing extraneous factors, things don't add up any more.

The limitedness and context sensitivity of theory is something I have noticed before, and it is a central platform of the philosophical treatise I'm going to get around to finishing one of these days.

Feather

KCJ, that was a lot to digest; please forgive the delayed response.

Allow me to illustrate, in some detail, why your notion that even in principle systems of particles are undetermined.

In a given system of n particles there will be n + (n(n-3))/2 "interaction terms" for each different kind of interaction that applies (you might note this is the formula for the number of line segments required to draw a polygon of n sides and all its diagonals). So for example if you have four charged point mass particles, there will be 12 interaction terms--6 for the charge interactions and 6 for the gravitational ones.

The acceleration of a particle is the net force on it divided by its mass. The velocity is the first integral, and the position is the second integral. So by numerical integration, just knowing the force law applicable, we can determine to any desired accuracy the motion of the system if we know the initial conditions.

In practice we may never know the initial conditions precisely, but in principle it is possible.

If we can calculate the positions of the system of particles to any accuracy and see chaotic properties, then I'd say you're wrong. Since chaotic properties can only be identified by construction or inspection anyway, I'd argue that chaos is rather predictable, in the sense that we either propose a chaotic solution that matches the observations, or the system evolves (according to our numerical predictions, mind) and at some points exhibits chaotic behaviour.

The fact that we know and can apply theories of chaos to physical systems implies they must serve some useful process of prediction. Otherwise they'd be useless as scientific tools.

This is absurd. One doesn't just stare at his initial conditions to see how a system will evolve, even if he has the exact analytic solution to a given system. Your statement is equivalent to claiming that the plot of a function on a cartesian coordinate manifold is an "emergent property" of the function and its first derivative at a given initial value.

If I know that the solution to a system is f(x, y, z, t) and I know f(x0, y0,z0,t0) and f'(x0, y0, z0, t0), then I can indeed determine every aspect of the system from t0 onward. If I don't know f(x, y, z, t), but I know the interactions involved and the given initial conditions, I can do the same thing, except numerically.

And we're back to the fact that "emergent property" is another way of saying "stuff happens" and then trying to couch significance in the new terminology that is accepted trivially in the old. Two particles interact or they don't. We know the law of interaction or we don't. If we know it, any system of the particles is determined if we know the initial conditions of the system.

Then that would imply a new force law is necessary. We know of four interactions possible between particles--Strong; Weak; Electromagnetic; Gravitational. Depending on the distance scale, one of these are the only ways known that particles can interact. That's it. We don't know of any more. If you insist that an "emergent property" is significant beyond a merely human label, then you are going to have to supply evidence that you have discovered another interaction.

There is a legitimate alternative. Toss causality right out the window. This is a scary thought, but a scientist must consider its possibility. However, why should causality be tossed when it works so wonderfully in every other aspect of the natural world, without exception? Perhaps a sort of hybrid does exist. Where is the boundary between the causality regime and the non-causality regime? How would you even tell? And, just to make it known, the concept of "emergent property" does not answer either of those questions at all. It doesn't even serve to adequately pose the questions.

As for "limitedness and context sensitivity" of theory, I just have to laugh. Every mechanical and electrical device you use is the result of predictable, deterministic natural law. Your car moves because the combustion of chemicals and movement of mechanical levers and arms is predictable. Your computer works because electron behaviour is predictable. And so on. The theory may be "context sensitive," but it is by no means "limited." Unless you mean "limited to the universe." Which is trivially true.