Quantum myth 3: nature is fundamentally random

[Ken G]

Consider, for example classical mechanics. Imagine, you have a ideal gas, consisting of tiny balls in the box, which behave according to Newton's laws. If you want to do any measurement of the state of the gas inside, you also have to interact with it somehow, introducing some contact with outer world, which is in thermal state, thus affecting the gas in random manner. Thus,the situation is IMHO pretty much the same.

Again we are thinking similarly, I've made that point myself, and again it says that too much is made of the "measurement problem" in an expressly quantum mechanical situation (as a similar problem exists classically).

However, in this case, I personally would consider such a system deterministic. The quantum mechanics is for me just different, more sophisticated way to establish the equations of motion.

I agree that we are led to the same conclusion both ways, but where we disagree may be our conclusion about reality. We both have used the word "deterministic" to describe a theory, not necessarily a reality, because determinism is itself an aspect of a model. As with any model, the task of comparing it to reality falls to us, and we determine how we will make that connection. The concept of "determinism" never survives that connection, it is lost in how we do science.

So neither theory "tells us" that reality is deterministic, though both theories are themselves deterministic until they make a testable prediction. The testable prediction, by virtue of the testability, requires that contact is made with elements that are outside the theory, and those elements will introduce the concept of randomness. We agree there. Where we may disagree is that you seem to view that as a kind of side effect of testing theories (as when you said "everything is deterministic", it's not clear what you meant there), but since the whole point of a theory is to be tested, I do not distinguish the expectation that the theory must be testable from the theory itself. In that sense, no theory is truly deterministic once its encounter with reality is included in the grand picture-- it is only the theory as idealization (i.e., as a mathematically structured object) that is deterministic.

Hence, everything is not deterministic, but key elements of the models are. We recognize that the wave function will evolve deterministically, but when we go to test it, we will encounter an incomplete ability to predict the testable outcome, and that incompleteness will obey an uncertainty relation (as alluded to by peter0302 as well). Some would count that as a lack of determinism, not so much in the theory itself, but in its point of contact with the rest of the scientific exercise. Others see that apparent breakdown in determinism as so severe that they feel the need to outfit an exostructure of "many worlds" just to avoid it, but you and I can question as to whether or not that is really necessary.

 

[Ken G]

Can it be consider deterministic because of its linearity?. That is, because it is not susceptible to slightest differences in initial conditions, so that in the evolution of one superposition of states each component evolves independently from the other?

The linearity of the Schrodinger equation doesn't mean it can't have sensitivity to initial conditions, i.e., a linear operator can treat nonlinear potentials. Chaotic systems like a 3D anharmonic oscillator can have a Hamiltonian, and nonlinearities in hamiltonian mechanics are important in the thermodynamic concept of "ergodicity".

 

[peter0302]

Right, as I keep saying, chaos does not equal random.

As I sit here I realize that the only things that are truly deterministic both in theory and in practice are things that don't exist physically, i.e., mathematical concepts. 2+2 always equals four _in theory_, but in nature you never get two things that are identical in every way (let alone 4), so one must always impart human understanding such as classificaiton to make _any_ mathematical statement whatsoever. Otherwise, there are always unpredictable (random?) differences between two physical things, which ultimately prohibit you from making absolute statements about nature, even statements so simple as 2+2=4.

Since the wavefunction is, we all agree, deterministic and governed by the Schrodinger equation, is this not an argument for the non-physicality of it?

 

[Ken G]

As I sit here I realize that the only things that are truly deterministic both in theory and in practice are things that don't exist physically, i.e., mathematical concepts.

I agree completely, we can imagine that determinism is a word that may apply to reality, but in fact it only ever applies to our models. It is important to recognize our successes, but also not to take them too seriously.

2+2 always equals four _in theory_, but in nature you never get two things that are identical in every way (let alone 4), so one must always impart human understanding such as classificaiton to make _any_ mathematical statement whatsoever.

I think that's a particularly clear way to say it. I had an argument on another forum about something very similar, I should have used that example.

Since the wavefunction is, we all agree, deterministic and governed by the Schrodinger equation, is this not an argument for the non-physicality of it?

I would agree, except to say that one does not need an "argument" for non-physicality, it should be the default stance. That's the fundamental puzzle of intelligence-- how do we make progress understanding a physical world by piecing together a bunch of non-physical mental constructs? How does a physical world build an intelligence that can understand how the intelligence got built? We may never lick that puzzle, I don't know.

 

[Hurkyl]

The way I would frame it is "everything we can say about determinism in the universe stems from application of the scientific method, therefore any inherent limitation in that method must also carry over into an inherent limitation in what we can say about determinism in the universe."

Again, I disagree. Yes, it's obvious that if you make an a priori assumption of determinism when applying the scientific method, then yes, the results of all of your analyses must agree with determinism, but that's an inherent limitation of making such an a priori assumption -- I see no logical argument that suggests that the scientific method itself implies determinism.

 

[Ken G]

The way I would frame it is "everything we can say about determinism in the universe stems from application of the scientific method, therefore any inherent limitation in that method must also carry over into an inherent limitation in what we can say about determinism in the universe."

Again, I disagree. Yes, it's obvious that if you make an a priori assumption of determinism when applying the scientific method, then yes, the results of all of your analyses must agree with determinism, but that's an inherent limitation of making such an a priori assumption -- I see no logical argument that suggests that the scientific method itself implies determinism.

Neither do I, so that is not what I said. I said two things, first that everything we know about deterministic behavior is a product of the application of the scientific method. Do you agree? Then I said that given that, if the scientific method encounters limitations in being able to label a physical process as deterministic (such as problems introduced by measurement), then that same labeling problem must extent to our very understanding of how the concept of determinism can apply to reality in that situation. Do you agree? That's all I said, so to disagree, you must disagree with either the first or the second aspect. I was talking about difficulties science has in addressing determinism, so of course I was not assuming that science must deal only with determinism.

 

[peter0302]

That's the fundamental puzzle of intelligence-- how do we make progress understanding a physical world by piecing together a bunch of non-physical mental constructs? How does a physical world build an intelligence that can understand how the intelligence got built? We may never lick that puzzle, I don't know.

It is like asking whether a computer can be self-aware? Or, if we consider a computer simulation - like the Sims - can a player, within the boundaries of the simulation, understand how the simulation works? To some extent yes, but eventually he encounter, at the smallest level, the limitations of the simulation, and reach a point where he can look no further. And moreover, since the system is not designed to let you look too deep, you might start to see some very odd and non-intuitive things the deeper you look. "Glitches" in the system resulting from its inherent limitations. So you reach a point where observation within the rules of the system is no longer possible, and you have to resort to guess-work - postulate + logic - to figure out the rest, and hope that all your observations of the "glitches" fit your hypothesis.

 

[reilly]

I don't grant this much significance yet, reilly. I can't say whether these observations made in brain science are incompatible with free will or not because I don't know what free will is. I can only think of the vaguest of definitions.

I can grant that dislike of cooked turnips may not be a free choice. No taste is, whether it is a taste for Beethoven or a taste for torturing young boys. But what about the choice to act or refrain from acting on that taste? Christians, Muslims, etc. would say therein is the free choice.

But what is a free choice? I can't isolate it to say for certain whether anything is a "free choice" or not.

I find it interesting that such a fundamental questions of physics could be at all related to a fundamental question of morality and/or religion.

I couldn't find the article I wanted to suggest, The Butler Made Me Do It; Science News...10 or more years ago. You can find lots on the free will topic: GOOGLE,[ "free will", unconscious ].

Regards,
Reilly Atkinson

 

[Ken G]

It is like asking whether a computer can be self-aware?

With one potentially important difference-- we know the computer is following an algorithm. Do we know that about reality?

So you reach a point where observation within the rules of the system is no longer possible, and you have to resort to guess-work - postulate + logic - to figure out the rest, and hope that all your observations of the "glitches" fit your hypothesis.

Yes, and face the possibility that there may be no way to "figure out the rest", not just because it's hard to do, but because there is no more that can be "figured out".

 

[reilly]

With one potentially important difference-- we know the computer is following an algorithm. Do we know that about reality?

Yes, and face the possibility that there may be no way to "figure out the rest", not just because it's hard to do, but because there is no more that can be "figured out".

Self aware? Maybe, maybe not.But I think it could be: the basic idea is, a pyramid of "watchers"(this is the core of Minsky's "Society of Mind" models.) For example; an image(s) can be stored in computer memory. At the crudest level, and somewhat oversimplified, a sequence of neural networks is trained to recognize the image, it size, color. location, duration,...Then translation networks map language descriptions into or onto the image, which can respond to queries about the image. That is, one constructs a system able to pass a Turing-like test, so that self-awareness can be demonstrated. Not easy, but certainly possible, in my opinion -- Steven Grossberg at Boston U, had completed some of this approach a some years ago, with a special type of neural network, so called Adaptive Resonance Networks. And yes, this would initially require brute force and probably more computing power than is currently available.
Regards,
Reilly Atkinson

 

[Ken G]

But I think it could be: the basic idea is, a pyramid of "watchers"(this is the core of Minsky's "Society of Mind" models.) For example; an image(s) can be stored in computer memory. At the crudest level, and somewhat oversimplified, a sequence of neural networks is trained to recognize the image, it size, color. location, duration,...Then translation networks map language descriptions into or onto the image, which can respond to queries about the image. That is, one constructs a system able to pass a Turing-like test, so that self-awareness can be demonstrated.

Yes, I would not go on record to say it can't be done, I'm just agnostic about whether or not a self-aware brain can figure out what self-awareness is. That agnosticism is related to the issue of whether or not a Turing test can really cut it-- it seems to me, a Turing test is the prescription whereby a brain can fool itself into thinking it is in contact with another awareness, but the best it can do might not be good enough. (No harm in trying of course.)

Right now, the logic seems to be, "I'm self aware, so I will posit that anything that responds to stimulus in a way that is indistinguishable from how I would must also be self aware". The best we can do, perhaps, but is it ever enough? How do we bridge the gap between an operational definition of how awareness acts, and what it is actually like to be self aware?

It reminds me of your point about Hume and causality-- if we see a close connection between two things, such as our own awareness and how we act in various situations, can we reliably reason backward from those actions by other agents and infer they have a similar self awareness? The problem, as with cause and effect, is that one cause may always be followed by a certain effect, but that effect may not always be preceded by that cause. Even if it always seems to be, we never really get to know the complete connection, the "divine providence" if you will, until we've "seen under every rock" and noted every possibility in the whole universe. How else can we rule out the possibility that something that we know is not self aware in the way we experience it could still support an architecture that could "fool" a Turing test?

It reminds me of when Kasparov was beaten in chess by Deep Blue. Kasparov knew that Deep Blue was programmed essentially expressly to beat him (it has never played a public game with anyone else, presumably because it might present "bugs" against a different style player), so it must have given him a weird feeling of looking in the mirror. Was he seeing a reflection of his own awareness in the actions of the machine? Perhaps Kasparov has a more visceral sense of artificial intelligence than anyone else, as a result, but even so, are we forever relegated to seeing only that part of ourselves when we look outside?

 

[Hurkyl]

I posit that a generic discussion of artificial intelligence or sentience belongs in either the computer or philosophy forum (depending on the content) -- if we think such notions are relevant to the thread, we ought to write down an operational definition, and work solely with that.

 

[Ken G]

The hope for sentience to be on topic in a thread on randomness is the hope that we can use the concept to understand the apparent randomness of human behavior in terms of internal degrees of freedom, internal sentience. But that just leads to the usual paradox that neither internal randomness nor internal determinism seems to explain where sentience comes into play. If we remove from the equation "sentience is what I have" on the grounds that such a subjective requirement is unscientific, I'm not sure there's anything left science can talk about, on a "randomness" thread or any other for that matter. We can look at the biological process of a brain making a decision, and look at what is inherently random and what isn't, that's the "operational definition" approach that we probably can do no better than. Like randomness, sentience then becomes a model of something else.

 

[peter0302]

I wasn't intending to talk about AI, but it is an example of why we, _inside_ the universe, cannot know everything there is to know about the universe from the inside. And I think I "proved" that theorem in an earlier post, because the "stuff" needed to observe "A" is always greater than "A".

The computer simulation is a great example because you will reach the limits of self-awareness within the confines of the system, outside of which you cannot step. You can experiment within the system to learn its rules, but you cannot discover the mechanisms enforcing those rules. That is why the computer cannot be 100% self-aware - it cannot use its own programming to learn everything about its own programming.

 

[reilly]

Determinism is plausible when and if the Law of Large Numbers/Central Limit Theorm converges for a set of experiments -- measure the initial conditions and the outcomes -- mass sliding down an inclined plane; a months movement of the earth; starting a car and getting it moving, ...Our intuition suggests these are deterministic situations; and many measurements will confirm determinism within experimental error.(This is pretty much the main idea behind Shannon's work, made rigorous by Feinstein. -- see Khinchin's Mathematical Foundations of Information Theory, Dover-- a superb book.

RE self awareness -- indeed it's not usually considered an appropriate topic for physics threads. I beg to differ, given Sir Francis Crook's take on the matter -- neural hypothesis and all that; he really talks about the physics of the brain as paramount for the study of mental phenomena. With all due respect, it would appear that few if any here have spent much time with the research literature of brain science; there are no theorems, no grand philosophical pronouncements Rather one sees articles like, Attentional Mechanisms in Visual Cortex(Maunsell and Ferrera), Neurophysiological Networks Integrating human Emotions (Halgren and Marinkovic), two of 92 papers, mostly experimental, in The Cognitive Neurosciences --edited by M.S.Gazzaniga . I have the first edition of this bible (1997); there's a revised one out. A must read if you want to get a real sense of what's going on, and how far the field has come from the days of intense AI and philosophical arguments, which are more and more becoming historical curiosities. There are tons of data on everything to the specifics of neuro-transmitters to consciousness. And that's where the action is.

Back to physics.
Regards,
Reilly

Re Turing? What's better?