Problem Of Induction
Note: I’ve moved the topic here to keep things neat. {[pink Thanks! ]}
—
Is the evolution of the universe entirely predictable?
Even if you could justify the assumption that you are observing reality directly and completely it would be impossible to answer yes to that question with any degree of certainty. The best one can do is to observe the evolution of the universe over a period of time and deduce from observation that, given some laws and the initial state, that all the events for the duration of the period are predictable. However, even if it is possible to do that for a period (however long) it is impossible to determine that the universe will not all of a sudden start to evolve in an unpredictable manner because, after all, it would be unpredictable.
Don
—
It is generally believed that the universe obeys certain exceptionless laws - i.e mass distorts space time and the like - and from these laws we can make predictions about a future state - thus, most consider the universe to be deterministic - if we can know the state of the universe at one instant we can know it’s future states. (I’m totally ignoring quantum mechanics here, forgive me, but I believe this generally works on the macro scales). Anyway, what I’m getting at, is that something would have to cause the universe to start evolving in a weird manner and thus these weird changes would not be unpredictable since they have a cause(again, I’m ignoring quantum mechanics). But then again, the universe has to be the ultimate chaotic system. So, perhaps while one can imagine that it is possible to know the (macro) state of the universe in principle, perhaps it is impossible in practice, and thus the universe is then not entirely predictable.
Kaitlin.
—
My point was, it is impossible to distinguish between two (or any number for that matter) sets of laws of physics if until now they would behave exactly identically given the initial state of the universe.
Suppose we knew the laws of physics such that given any state we can predict the future and also the past to agree with observation. Now if we modified those laws so that every 15 billion years some physical constants changed or started changing with time then we could still determine the past given the current state but predictions for the future would obviously be different. (The universe is roughly 14 billion years old)
So now we have two sets of laws of physics that agree perfectly with observation but predict a different future for the universe. Presumably we will only experience one history of the universe so one set of laws is incorrect. But it’s impossible to determine which set.
“something would have to cause the universe to start evolving in a weird manner”
I see no reason to assume this. After all, it could just be a function of the laws of physics, albeit a completely unpredictable one.
Don
—
I’m sorry if I misunderstood you before. You raise an interesting point. I still hold that any (macro) effect requires a cause, but I may be mistaken.
It’s interesting - as we look back in time, we note that the laws of physics have not changed (hence why we call the laws exceptionless) because they apply throughout time and space. Actually, I recall that there was an experiment done to determine whether the value for the gravitational constant (G) has changed over the age of the planet (4.5 billion years?). By measuring the size of crystals of a certain mineral along a very deep mine. As far as I recall they found that the value has not changed enough to be measured.
And that raises another issue; if the so called ‘laws’ of the universe are subject to change, is it even reasonable to study physics?
Kaitlin
—
The point that (I believe) Don is making hints at David Hume’s observations on the problem of induction.
Induction is used to infer general laws or principles from finite sets of past observations. E.g. All past observations of X resulted in Y, therefore future observations of X will result in Y.
However this assumes that X is the cause of Y, and as Hume notes this can only be assumed. All we actually observe is constant conjunction between X and Y (or correlation), not causation in any direct sense. The causal connection is assumed.
But if we can observe only correlation not direct causation then it is possible we are simply at the end of a very long run of luck - there may be no causal connection between X and Y (or law of nature governing their interaction). The correlation observed in the past could just be good luck and is no indication that it will continue in the future.
Appealing to experiments (such as Kaitlin’s change in gravity example) doesn’t help because it is the validity of such experiments that Hume’s argument undermines.
In response to Kaitlin’s last point (i.e. given this is it sensible to study physics?), Hume himself noted that although this is an interesting problem it certainly shouldn’t stop us using induction in both science and everyday life. Without it we’d have no reason to do anything at all - without it there is no reason to believe that eating and drinking won’t kill us even though it kept us healthy the past 100,000 times we did it. So although from a strictly philosphical point of view indunction might be unjustified, belief in indunction is still reasonable because the alternative is even sillier.
I apologise if I’ve misrepresented anyone here or made any other mistakes.
For a much better explanation you can check out the Stanford Encyclopaedia of Philosophy entry here: http://plato.stanford.edu/entries/induction-problem/#IndJus
Cheers, Toby M
—
It is possible that the laws of physics could change without precedent at any time. However (disregarding the problematic nature of induction), we have no reason to believe they should. In terms of dramatic law-changing events in the universe’s history, we have an experimental sample size of 0 based on human observation. There may be events that we have not detected or been present for, or ones that haven’t occured yet, but we have no justification for or against these assumptions. They are just that, assumptions.
Just to open another can of worms, some would argue that every possible event should take place at some point in the distant future, as a result of parallel, divergent universes that contain all possible eventualities. Would these all be governed by the same laws, or are there worlds where even they change?
I would be inclined to say yes. The most important discoveries of the past century in physics have shown us that the laws were made to be broken, or at least bent. Special relativity shows us that fundamental concepts such as mass and time are mutable, and in 2007 John Wheeler’s delayed-choice experiment was carried out, apparently proving that obervation can define the position of particle IN THE PAST as well as the present (see http://sciencenow.sciencemag.org/cgi/content/full/2007/216/4)..) It may be that there are underlying, immutable laws that underpin ours, but allow for these unusual, counter-intuitive happenings.
Andrew W.
—
The discussion above is interesting but i don’t have anything to add to it at this point.
I just have some thoughts off the back of the first two lectures. We noted that we now know that the earth is no longer the centre of much. It isn’t the centre of the solar system, our galaxy isn’t the centre of all of them. I was just thinking that it isn’t particularly important that earth isn’t the centre of much. I just do not see how being in the centre should be of any importance. Obviously we have to look at it since it was a common idea through history.
Robin
—
“It is possible that the laws of physics could change without precedent at any time. However (disregarding the problematic nature of induction), we have no reason to believe they should.”
That’s certainly true and raises an interesting question on onus of proof. Hume didn’t believe they should, only that we have no reason to believe they won’t.
“Just to open another can of worms, some would argue that every possible event should take place at some point in the distant future, as a result of parallel, divergent universes that contain all possible eventualities. Would these all be governed by the same laws, or are there worlds where even they change?”
Interesting question. I only know a little about Lewis’ possible worlds theory, and I’ve heard of others in physics. I’m not sure if his theory is at all comparable - if not I’d like to hear about the others.
On Lewis’ view (as I understand it) anything logically possible must be instantiated in at least one possible world/universe. Likewise anything logically necessary must exist in all, and anything logically impossible exists in none. So I suppose on Lewis’ theory, if there is no possible world in which laws of physics change then it must be logically impossible. But we don’t have access to Lewis’ possible worlds so it doesn’t get us far - Lewis’ theory is more an explanation of why statements about possibility are meaningful.
Interesting read on the link you posted, thanks - though I don’t see how it supports the claim that laws of nature are liable to change. It seems more like a possible example of backwards causation. But lets not go there.
Cheers, Toby
—
This is returning to the earlier discussion of predictability of the universe and of the possible physical laws in general.
If your unpredictability is constructed specifically that it defies observation and testing excepting at the one point in time when nobody’s looking, then it’s empty. Sure, it’s a plausible universe, but at the same time, there are infinitely many plausible universes that can be defined in such a way as to have both unpredictability and be observable in a general sense. The mechanism by which an unpredictably unpredictable universe remains unpredictable and beyond statistical observation is somewhat interesting and novel, but the specific situations that unpredictability leads to aren’t, in a large-scale sense - the unique thing about a chaotic universe is the mechanism behind the chaos, and where it gets the chaotic patterns of action from.
This ties into the idea of a discontinuity in the laws of physics. It seems interesting in the sense that it requires a mechanism by which your laws are suddenly but unpredictably changed, without having exhibited any flux in the past. This can be explained by randomness, but so can anything, which is why I’m distrusting of any arguments revolving around it.
Hume’s argument is something to keep in mind, but if statistical correlations are the best you can get outside of pure mathematics (which this would seem to prove), and if said correlations are sufficiently close to 1 and there’s some observable flux between the objects being measured then yada yada math is boring. It’s empty to assume any kind of general randomness, and equally empty to assume that any localised unpredictability will refuse observation and calculation if its variations. I’d love it if anyone could put out a theoretical example of some subset of the Universe that had large-scale unpredictability, while still being a universe that held information, because it’s defying me at the moment.
Moving away from that, I’ve been interested about the nature of plausible universes - is there any other set of fundamental constants that could lead to the large-scale complexity in our universe? I’ve read a bit about how if they were too far off the current values, then it would be impossible for electrons and protons to form and so the universe would be a tasty energy soup, but it’s something I know next to nothing about what to be searching for.
Apologies if this is isn’t very readable. I can’t think of how to distill it or word it better than I have here - it was put together in notepad in about an hour. Is editing our posts kosher, so long as the general point remains the same? I’m mostly a math/physics student (as if it wasn’t obvious) so if there’s anything outside those areas that I’m missing that should answer any of my points, I’d love to read/watch/etc. it.
Andrew K.
—
in following on from the idea that the universe may in fact be completely unpredictable, similar may be said about our perception of reality: there is no way in which we can prove that what we perceive to be real is in fact so, however obviously for the sake of practicality we must assume that our perceptions of reality are correct, as there is no real validity in assuming otherwise. While our perceptions are possibly flawed, this assumption restricts our ability to investigate and predict.
This is true for science as well, and in this light Andrew K’s point rings true. While of course it is possible that there are unpredictable aspects of the universe (and therefore we wouldn’t be able to predict how or in which way they might effect the universe in the future so we would have no way of knowing whether these aspects of the universe exist), assuming that this is true is pointless in the same sense that it is pointless to assume that reality differs from our perceptions. This is because the very nature of science is to predict based apon observation. Scientific theory may not be correct, but until there is a counter-example against current theories, there is no reason for them to be revised (so while it is possible that we have an incorrect assumption of cause and effect, it is meaningless to assume this without an observed counter example).
In this sense it is not scientifically valid to assume that the universe is unpredictable despite the fact that we can in no way say that it isn’t. Philosophically it is a perfectly valid question, in the sense that it is indeed true that our prediction that it is predictable may indeed be false (and there is no way to prove otherwise) however for the sake of practicality the sentiment that the universe is unpredictalbe cannot have any merit in a scientific context (as that defeats the whole purpose of science)
David Mc Manus
—
I’d like to take a slight tangent on David’s point about the need to assume that our perceptions of reality are correct - or, to be precise, accurately descriptive of the “true” underlying reality.
Should this requirement hold even as we develop greater understanding of the nature of our perception itself?
Information about everything we observe is conveyed to our consciousness by electromagnetic interactions, be they in the form of sight, hearing, touch or smell. As a result, our perception of this information is constrained by the speed of these interactions, which is that of light itself. All our accumulated scientific knowledge also reinforces the theory that this particular speed governs the very behaviour of energy and matter in the universe (small u) - it is a cosmological constant, and quite a sacred one at that.
But what if the speed of light is merely the upper limit on our own perception, rather than a universal constant? If superluminal interactions were commonplace throughout the Universe (big U), any attempt we might make at understanding them would prove futile as we would be quite incapable of perceiving them accurately, if at all. This would have different implications on our current interpretation of reality - the Universe that we aim to describe via science is merely the universe we are able to perceive.
It could be argued that any speculation on the nature of reality beyond the observable universe is pointless - mainly due to practical considerations, as science has proven valuable and more than competent in explaining interactions at speeds comfortably within human perception.
But as we explore near the fringes of our senses (interactions close to, at, or even beyond light-speed), wouldn’t it be worthwhile to consider the possibility - likelihood, even - that we are hampered by our imperfect perception of reality?
Alvin Chua
—
I don’t want to labour the point but I feel Hume is being misrepresented here.
In response to the criticisms of Hume’s assumption that the universe is random I’d reiterate my point that Hume isn’t saying this. In fact he assumes the opposite.
His argument is intended only to demonstrate that it is an assumption. The conclusion is that induction is based on an unverifiable assumed premise and therefore not a certain basis for knowledge.
Obviously we can’t operate on the assumption that induction doesn’t work or that there is no order to the universe. Hume explicitly states this as justification for assuming the validity of induction.
The assumption that all we have access to is perceptions is the underpinning doctrine of Hume’s empiricism. Given we only have access to perception, and we cannot perceive causal connections only conjunctions/correlations, causal laws can only be assumed since it is possible the correlations are merely fortuitous.
All this implies that if scientific knowledge is built on induction, it is uncertain. Given the rate at which scientific theories are produced, refined and outdated (one of the main themes of the first lectures), I don’t see why this claim should be controversial.
Some more recent philosophers such as Popper have argued that science does not proceed by induction but by other means such as falsification, but sadly I don’t know much about those yet.
Cheers, Toby
