Last week, we had some discussion of free will, which prompted some comments about determinism. I’m not convinced that determinism has all that much to do with free will one way or the other, but since the topic’s been raised, here are a few bullet points, jotted down late at night, which I hope will still make sense in the morning.
- It seems to me that the whole notion of determinism is quite a bit murkier than it first appears, to the point where I’m not at all sure I know what it means — and not at all sure that its defenders have any clearer sense of its meaning than I do. The remaning points will elaborate on this.
- I’ve seen people define determinism as the statement that the history of the Universe could not be anything other than what it is. That, of course, is true, because nothing can be anything other than what it is. If it is a fact that Abraham Lincoln scratched his beard in the East room of the White House on April 14, 1862, then it cannot not be a fact that Abraham Lincoln scratched his beard in the East room of the White House on April 14, 1862. If it is a fact that intelligent fish from the moons of Saturn will land a spacecraft on the Lincoln Memorial on April 14, 2062, then it cannot not be a fact that intelligent fish from the moons of Saturn will land a spacecraft on the Lincoln Memorial on April 14, 2062. So to define determinism this way is to make it trivially true, and hence uninteresting.
- Another possible definition is that the future history of the Universe is a logical consequence of its present state. This (by Godel’s Completeness Theorem) is equivalent to saying that any Univese in the same present state as ours must have the same future as ours.
But I claim this formulation makes far less sense than it appears to, because our Universe has no globally defined “present state”. The “present” is a local concept — and even that is observer-dependent. Now, the observer-dependence is perhaps not such a big problem; you could always formulate determinism to mean something like “The future, as defined by any observer, is fully predictable from the present, as defined by that same observer”. But the larger problem is that no observer has a notion of simultaneity that extends beyond his local neighborhood. (Very simplified cosmological models do incorporate global time coordinates, but I’m pretty sure nobody believes that’s an accurate depiction of reality.) So according to “determinism”, exactly which regions of spacetime are supposed to contain the information necessary to determine the entire history of the Universe? I’ve never seen a clear answer to this, which suggests that nobody has given a really coherent description of what determinism is.
Edited to Add: Roger Schlafly points out in comment #1 that one doesn’t need a global time coordinate to define determinism; one only needs a spacelike hypersurface, which might or might not be a surface of simultaneity for some observer. Point taken, and this largely dispenses with my objection here.
On the other hand — does “determinism” mean that there exists a spacelike hypersurface that completely determines the history of the Universe? Or that there exists a family of spacelike hypersurfaces that cover all of spacetime and any one of which determines the history of the Universe? Or that any spacelike hypersurface completely determines the history of the Universe? It still seems to me like there’s some fuzziness here, though I now think it’s much more likely that this fuzziness has been addressed somewhere.
- In any event, even if you take a very local notion of determinism, it’s pretty definitively ruled out by the Aspect experiments, for reasons I’ve described in Chapter 16 of The Big Questions. More precisely, the results of the Aspect experiments force you to abandon either determinism or Lorentz invariance (i.e. special relativity), and even if you think it would be crazy to abandon determinism, it can’t be as crazy as abandoning Lorentz invariance.
(A third possibility is that Nature is really out to fool us, say by infecting the brains of Alain Aspect and his colleagues so that they consistently choose to observe exactly the wrong thing, or by altering the experimental records — and the experimenters’ memories — after the fact, to create the maximally misleading false beliefs about what actually happened. Is there any determinist so determined that he’s willing to stand on this shaky foundation?)
- The above all ignores the possibility of a many-worlds or many-histories interpretation of quantum mechanics, which can indeed rescue some form of determinism from the Aspect assault, but I’m not at all sure that that sort of determinism has much to do with the determinism of the declared determinists, because it admits that the “present” state of the Universe (whatever that might mean) does not uniquely determine the future state in any given history.
- In any event, what does any of this have to do with free will? Let’s take the most extreme form of determinism imaginable: Suppose it’s the case that there is one and only one Universe, with one and only one history, in which intelligent beings emerge. Thus the entire history of the Universe is a logical consequence of your existence. What light does this shed on the question of whether it’s useful to think of your volitions as causal? And if none, what other relevant question does it address?
I disagree with steps 3 and 4. Determinism just needs a spacelike hypersurface. You can make the hypersurface simultaneous for some observer if you want, but it is not necessary. I don’t know why you say that nobody believes in cosmological time. That is what they all believe in when they say that the universe is 13.7B years old.
The Aspect experiments have almost nothing to do with determinism. They confirm certain formulas of quantum mechanics, but those formulas have deterministic and non-deterministic interpretations. It is hard to see how any experiment like that could prove or disprove determinism.
#6 – intelligent beings? like us? who compete over resources in almost the same way that single-celled organisms do? we share the biological imperative to exist with every other ‘living’ thing. im not sure that we have attained ‘intelligence’.
none, beyond the fact that i had to learn about it on my own. i was taught that free will existed by older monkeys from an early age.
the other relevant question (what im going to do tomorrow)- pretty much exactly the same thing i did last monday and most of the mondays before that. i cant be the only one who has a routine, no?
The bottom line is that our choices are either deterministic or random, meaning uncaused, both alternatives prohibit free will, and there is no third alternative.
In terms of proving determinism, you might want to check out the transcript of an episode of my show I did on it – Episode 10. Why Change as the Basic Universal Process Makes Free Will Impossible http://causalconsciousness.com/Episode%20Transcripts/10.%20%20Why%20Change%20as%20the%20Basic%20Universal%20Process%20Makes%20Free%20Will%20Impossible.htm
The basic point is that if nature wasn’t fundamentally causal, there would be no motion and all would be completely static.
Kindly note that the laws of physics, deterministic as they are at a microscopic scale, are equally deterministic going backwards in time. Which means that our “post hoc, ergo propter hoc” intuitions of causality are wrong at that level, and cannot be used to reason about issues such as we are discussing.
Roger Schlafly:
I disagree with steps 3 and 4. Determinism just needs a spacelike hypersurface. You can make the hypersurface simultaneous for some observer if you want, but it is not necessary.
Point taken. I’m editing to acknowledge this. Thanks.
George >>The bottom line is that our choices are either deterministic or random<<
This is perhaps one of the main confusions, and shows why determinism cannot easily be applied to the macroscopic world. Determinism isn't in conflict with randomness, because randomness is simply a mathematical pattern that cannot be compressed to a less disordered sequence (see Kolmogorov complexity for a fuller definition). Thus a pattern can be deterministic and random at the same time.
‘observers’. Last I checked all of those evolved, and did not exist before doing so. Unless you mean by observer ‘some other stuff that interacts’. Like a hydrogen atom.
I’m on board for #6 though.
The argument in favor of determinism and against free will becomes progressively more juvenile the more I think about it.
It seems to be that the claims go something like this:
1 – If humans make choices based on information gleaned from the universe around them, then the choice is a direct product of the information and nothing else. Therefore, there is no free will.
2 – Otherwise, human decisions are random chaos, therefore there is no free will.
Option 1 strikes me as identical to the “watchmaker” argument that proponents of Intelligent Design use to mock science. Except that the opponent’s of free will are saying – in seriousness – “Yep, that’s how it happens, exactly!” :O
But what Options 1 and 2 really have in common is that they are examples of a “Begging the Question” fallacy. The premise is included in the theory. Option 1 assumes that free will is not present in the universe as part of the information being considered, and then concludes that free will does not exist because of that very fact. Option 2 ignores the fact that free will could very well be what makes human decisions appear “chaotic,” and thus concludes that anything that cannot be explained without attending to considerations of free will must prove that free will doesn’t exist.
The argument is circular. I wrote a blog post on how easy it is to get lost in circular reasoning, and I believe that’s what’s happening to people who don’t believe in free will. See below:
http://www.stationarywaves.com/2012/04/begging-question-or-brainwashing.html
How does this arguement for free will make a distinction between living and non living things? For instance, weather systems seem to also show elements of both randomness and determinism and it is very hard/maybe impossible to specify with near perfect accuracy what specific outcome is going to occur even at a local state. We probably would not think of a weather system as having free will. Also, there are states where more or less randomness can be expected, does this mean the amount of free will varies depending upon the state of the brain at any one time? This isn’t necessarily an arguement against your definition. I am just curious how these issues would fit in. I think that some people’s arguements for free will tend to be that, there is randomness/currently unexplainable features occurring, therefore there must be a soul/something mystical accounting for that randomness or unexplainable feature. It’s that extra step in reasoning that some people make that makes it harder to discuss free will.
I think that free will is a strange self-referential loop. Suppose some advanced AI program (no doubt in attempted development by Amazon as we speak) purports to consistently predict my choice of action from a very long menu. The existence of such a program would affect my choice of action because I might then want to make a contrary choice to prove my free will. The AI program would of course have to anticipate that and predict when I would exercise that option. But I would choose when to exercise the option so as to prove my free will, etc etc.
George Ortega: “The bottom line is that our choices are either deterministic or random.”
I grant that this is true, if you accept that our choices must be predictable in terms of some sort of mathematical “process.” But that is the very thing free will theorists are inclined to deny, so I don’t think you’ve disposed of their doctrine.
I greatly appreciate your attempt to establish a coherent definition of ‘free will’ in an earlier post. Unfortunately, it seems people have already reverted to an”I know what I mean, so should you” mentality.
Since most of this post went over my head, could you explain if there is any significant difference between the model of the universe you describe here and “the universe is deterministic except for some randomness introduced on the quantum level”?
Neil – Could another AI program be created to outsmart the first using the same methods you use to prove your free will?
Martin-2
You are asking whether I think an AI machine can possess the phenomenon we call “free will”. Yes, in principle it can, and someday will, providing one does not define free will in a way that violates the laws of nature. For example, Steve’s subjective experience definition. This would mean an AI program with subjective experience. But unless you believe in magic, that must be possible too in principle. Of course we don’t have a clue right now how to accomplish that.
I am not a mathematician, but it was my understanding that Chaos theory (in the mathematical sense) had reasonably well established that there are situations in which no amount of knowledge about the present will allow complete accurate prediction of the future. Further, I thought it was reasonably well understood that significant systems in the real world (such as global weather) were chaotic, in the formal mathematical sense, and therefore unpredictable in just the same way that we agree we cannot know both the position and velocity of subatomic particles with infinite accuracy.
Given that, it would seem that if indeed the universe is deterministic, it must be unknowably so, even given the amended “spacelike hypersurface” above. It then seems to me that we have a hard time distinguishing a universe-in-which-we-live that is unknowably deterministic from one that that is not at all deterministic. This further seems remarkably analogous to the problem of determining whether G-d exists (in the Judeo/Christian/Muslim sense of a single omnipotent omniscient diety that no longer has direct verifiable interactions with the observed world).
Alan – Well said.
Martin-2, saying that “the universe is deterministic except for some randomness introduced on the quantum level” doesn’t help much. First, quantum mechanics underlies everything, so randomness at that level means randomness everywhere. 2nd, even without the quantum level, weather is chaotic, as Alan has explained. 3rd, there is no proof that there is randomness (in the sense of nondeterminism) at the quantum level.
Alan Wexelblat:
it was my understanding that Chaos theory (in the mathematical sense) had reasonably well established that there are situations in which no amount of knowledge about the present will allow complete accurate prediction of the future.
Your understanding is absolutely, unequivocally, 100% wrong.
There are (at least) two definitions of the term ‘free will’ at work here. In the earlier post, Landesburg speaks of free will as “deliberating among various options and making a choice.” We might call this the modest definition of free will; by this definition, there is no obvious conflict between free will and determinism. Landesburg takes this to be the “right” (only?) definition because “no other definition makes any sense.” I half agree with him, inasmuch as I think the other definition is probably conceptually incoherent, but if he means no one really believes in it or defends its existence, I think he is mistaken.
The other definition–call it the bold definition of ‘free will’–holds that free will is a necessary condition for a certain kind of personal responsibility, such that we can choose to refrain from doing wrong contrary to the forces of nature. I put this vaguely because it is a vague notion, but it is historically important, especially in religious discussions of the problem of evil. If people have (the bold sort of) free will, so the argument goes, that explains how it can be that an omnipotent, omniscient, perfectly good divine power can have created the world, and yet will also punish people for being evil. The modest definition of free will will not do this job; sure, we deliberate among various options and make a choice, but if our choice is entirely in accordance with the laws of physics, then any divine power that designed the physical universe and all its laws, and was able to foresee all that would happen, knew in advance that we would make evil choices, and chose to create a world in which we make evil choices and are punished, rather than a world in which we make good choices. If that’s true, it’s an uphill battle to argue that this divine power is good (or to convincingly defend the divine by appealing to quantum uncertainty.)
Lots of people, today and throughout history, want to maintain that there is a divine power that escapes these criticisms, and try to use free will to answer them. So, one current and historical definition of ‘free will’ is ‘that sort of freedom that implies the sort of personal responsibility that would excuse the divine from responsibility for our wrongdoing.’ I think this notion turns out to be incoherent, but that is not obvious at first blush, and a lot of debates about free will involve this definition.
SO, if this is a real usage of ‘free will’, and this is what Sam Harris is talking about, I think Landesburg probably wants to agree that this kind of free will doesn’t exist. He argues in BQ that we are playing word games if we “define free will by its physical impossibility–and then go on to debate whether it exists or not. . . . If you require free will . . . to be impossible by definition, then it doesn’t exist” (72). But his argument presupposes that whatever is physically impossible is altogether impossible. Yet much of the debate over the problem of evil is a debate over whether the divine makes possible things that are physically impossible–e.g., whether free will is miraculous, or the result of a sort of spiritual power that can can contravene the laws of physics. And, in point #3 above, Landesburg says, “no observer has a notion of simultaneity that extends beyond his local neighborhood”–but if defenders of free will are trying to answer the problem of evil, then they, at least, are apt to maintain that the divine will have such a notion of simultaneity, and so their opponents can hold them accountable for a notion of determinism built on that supposition.
re 18:
I think I double. Assuming Alan means no *finite* amount of knowledge. Imagine a machine that every second prints the next decimal in some measurement. You know it to 100 places. Predict what happens in the 101st cycle to within 2 …
In a chaotic system won’t there be trajectories such that for any epsilon there exists a delta where 2 initial points within delta differ more than epsilon?
So what have I got wrong?
I wish to apologize for mischaracterizing Landsburg’s views. I thought he was on board with determinism. I apparently learned nothing from Chap. 16.
That said, it remains unclear to me that Landsburg has identified anything “free” about his concept of free will that would distinguish it from other phenomena such as momentum that people generally do not characterize as free.
An example for consideration: Mary Sue suffered from Transient Global Amnesia. (Click on “ Stream”; the relevant RadioLab story starts roughly half way through the linked podcast). That is, Mary Sue lost the power to form memories – to learn, to grow. This fact may not have enabled you to predict her conduct. But it enabled the doctor to predict that her conduct would repeat. And it did. Mary Sue began repeating herself – the same questions, same words, same inflection – every 90 seconds. Given the same stimulus, she mechanistically produced the same response – pretty much exactly.
I imagine that my mind works the same way that Mary Sue’s works, except that the mechanistic nature of my responses is concealed by the fact that I learn, and so my internal state varies over time. As a consequence, my responses will change over time, even when confronted with the same stimuli.
But I don’t control the stimuli I receive. And if the stimuli are what determines my behavior, what responsibility do I bear for my behavior?
Ken B:
Chaotic systems are entirely deterministic. If you know the state at time $t$, you can calculate the state at all future times.
You’re saying, “Ah, but you need *infinitely much information* to do that calculation.” Well, sure. The same is true of *non*-chaotic systems.
Here’s a nice non-chaotic system. I have a particle that is in a certain location and will never move. Where do you predict it will be one hour from now? To answer that to infinite precision, you need to know its current state to infinite precision. But that doesn’t render false the statement that the future state is calculable from the current state.
Chaotic systems depend more sensitively on their initial conditions, but they are no less deterministic for all that.
How about this: it all boils down to what really is conscience, or mind. If it works like a very complicated computer, as we know them or can imagine them, there is no free will – every decision is some superposition of the deterministic (function of the input and internal variables) and random (there are surely random generators in the brain). If the mind can add something else on top of that, overrule somehow the “mechanical” processes, maybe through some wild positive feedbacks and infinite recurrences, maybe some complexity that breaks down the determinism (maybe we need some new math?), then I would say there might be free will. To be sure, I feel to have one and at least it is a very useful illusion..
Thank you Roger.
People seem to wonder when computer decision making will reach the level of human decision making. But there’s nothing outlandish about a computer program that maximizes a utility function, so I wonder if the seemingly magical decision making process of humans is the result of psych/neurological baggage making our minds overcomplicated and inefficient. Like peacock tails.
My uneducated guess would be that a chaotic system capable of amplifying individual quantum effects to a macro level cannot be deterministic. Somehow the full history of the Universe could not have been written into the original tiny hot speck (seems to my humble layman self)
re 22:
Steve: yep, I agree with all that. But that doesn’t conflict with Alan’s statement, which was about arbitrarily accurate prediction. One can imagine systems in which to predict at some future t within e you would only need to measure current conditions within d. If you have a chaotic system though you will be thwarted. The question, defined by the italicized snippet from AW you quoted, is one of prediction not determinacy.
In your stationary particle example I cam map e to d fairly directly …
If you want to really drive yourself crazy, contemplate the issues of free will, determinism, quantum theory, etc. in the context of Nick Bostrom’s proposition that we are all living in a computer simulation, and then try to answer these questions both for us and for the beings running the simulation (assuming they themselves are not a simulation…and the infinite regression issues that introduces).
Steve, Alan’s understanding is not wrong. First, chaotic systems are not necessarily deterministic. A mathematical model might be deterministic or not. A physical system, like the weather, may be chaotic and there is no way to know whether it is deterministic or not.
Second, if “knowledge” means finite knowledge, as it usually does, then Alan is correct that accurate prediction is impossible in a chaotic system.
Roger writes: “Second, if ‘knowledge’ means finite knowledge, as it usually does, then Alan is correct that accurate prediction is impossible in a chaotic system.”
Yes, but what does that have to do with determinism? The classic expression of causal determinism is that of Pierre-Simon Laplace:
“Given for one instant an intelligence which could comprehend all the forces by which nature is animated and the respective situation of the beings who compose it—an intelligence sufficiently vast to submit these data to analysis—it would embrace in the same formula the movements of the greatest bodies of the universe and those of the lightest atom; for it, nothing would be uncertain and the future, as the past, would be present to its eyes.” [A Philosophical Essay on Probabilities, trans. F.W. Truscott and F.L. Emory, 1902]
In other words, determinism is not about what we can know, but about what a hypothetical intelligence could know.
If the argument is about a hypothetical intelligence or God, then the universe is deterministic under a trivial argument, as in Steve’s point no. 2.
I like these posts. Interestingly, the pastor at church last weekend was preaching on the “tension” found in scriptures of free will vs God’s sovereignty. This is similar, but not identical to, the tension between ideas of free will vs determinism.
“In other words, determinism is not about what we can know, but about what a hypothetical intelligence could know.”
If this is so, it’s not clear why anyone would dogmatically insist that the universe is deterministic. After all, we haven’t been able to cook up a set of formulae that adequately describe all known experimental data – the existence of some “laws of nature” must therefore remain in part an article of faith for now.
And is it clear that deterministic micro-level laws of the kind physicists posit and test give rise to determinism at the macro level in a potentially infinite universe?
“What light does this shed on the question of whether it’s useful to think of your volitions as causal?”
Whether or not our _experience_ of volition is itself causal or not, what causes our volitions in the first place? Either we choose them, at which point we have to figure out te cause of our volitions about our volitions, or someone/something else does. Or nothing at all does and they just happen without cause (including causes like “my strength/lack of character”).
Posit strong determinism or the most conceivable state of indeterminism you can fit into a fairly predictable universe (on the macro-scale): I don’t understand where the strong form of Free Will fits anywhere into this picture, or, still, what it’s even supposed to be claiming.
We know the doctrine of Free Will only by its desired effects (our responsibility for actions in a way that is somehow above and beyond any sort of internal, predictable specific nature, and also a complete lack of responsibility for any purported Creator of actors), not by anyone ever defining the doctrine itself. That’s bizarre and pointless.
Steve, I have a few thoughts on this subject:
1. I think you take the statement “the Universe could not be anything other than what it is” a bit too literally. Consider the statement “The inventor of bifocals could have been someone other than the inventor of bifocals.” Taken literally, that statement is clearly false: obviously regardless of what happens, the inventor of bifocals is the inventor of bifocals. But it still may be possible for Benjamin Franklin, the actual inventor of bifocals, to have NOT been the inventor of bifocals. Similarly, although in the literal sense “the history of events may not have been the history of events” is nonsensical, it can still be understood in a similar fashion.
2. Regarding the Aspect experiments, Bell’s theorem is a very precise result: it states that Local Determinism + Counterfactual Definiteness + the No-Conspiracy Condition together imply various Bell inequalities (such as CHSH), which are in contradiction with the quantum mechanical results shown in experiments like Aspect’s. The “locality” part can be defined in terms of the independence of spacelike seperated events. Counterfactual definiteness (CFD), also known as “realism”, states that for any observable properties A and B of an object, if you measure property A for the object, there is still a well-defined answer as to what value you would have gotten if you had instead measured property B. For instance, CFD says that if you measure the position of a particle, then there is a definite answer to the question “what momentum would you have measured if you measured momentum instead”. The no-conspiracy condition states roughly that nature is not conspiring to force you to make exactly the sort of measurement decisions needed to make experiments like Aspect’s come out in exactly the right way. (People who disagree with the no-conspiracy condition are known as superdeterminists, and they include Nobel laureate Gerard t’Hooft in their number.)
3. Regarding both determinism and free will, I don’t feel it is particularly helpful to think in terms of alternate universes, but in any case I think it’s better to think in terms of possible universes rather than actual universes. It seems like you can have people who have clear opinions on these issues while being agnostic on how many universes there are. The definition of “A causes B” can be stated as “A happened before B and in any possible universe in which A occurs, B must also occur”, regardless of how many actual universes there are.
Keshav Srinivasan:
1) I don’t actually know what you mean by the sentence “The inventor of bifocals could have been someone else”. “Could have been” conditional on what, exactly? Conditional on all other events in the Universe being what they are? If so, the sentence is pretty clearly false. Maybe conditional on all events prior the invention of bifocals being what they were? Well, then you do have to be a bit careful about what “prior” means, but then your sentence is essentially equivalent to: “There exists a Universe that is identical to ours up to the moment when bifocals are invented, but not thereafter.” So I think your point 1) is subsumed in your point 3).
2) Re the Aspect experiment: Thanks for providing our readers with this more precise version of point 4 in the original post. I don’t believe you’re disagreeing with that point; just restating it a bit more carefully. Yes?
3) I totally do not understand your distinction between “possible” and “actual” universes, or why you would want to make such a distinction. How does one check whether a Universe is “actual” or not? This seems to me to be just the sort of metaphysical baggage that it’s always best to jettison — it’s a heavy load and it serves no purpose.
So you want to define “A causes B” as “A happened before B and in any Universe in which A occurs, B must also occur”. This seems to me to be a very strange definition of causality; it means that anything that is logically necessary is caused by everything that happens. It’s also quite hard to know how to decide whether an event in another universe counts as the same A as some event in our Universe. I’d have preferred to think of Universes as disjoint, so that if A is an event in our Universe, it is necessarily *not* an event in another.
(I am willing to allow, though, the notion of an isomorphism between some fragment of one Universe and some fragment of another, which means that with sufficiently careful phrasing your notion can be made meaningful — just not, I think, a terribly good reflection of what most people mean by causality.)
I picked up a cup of joe at my regular coffee shop this morning, for the first time in about a month. “Where have you been,” they said.
What I want to know is: was it just coincidence or was I preordained to visit my coffee shop only on the last day of July in 2012?
I understand a watch to be a deterministic system. The tension of the spring excercises a near-constant force on the gears, which in turn absorb the force so that the fast dial advance at almost exactly 1 degree per second. The presence of dust, losing of tension, temperature changes and the like will effect the speed of the dial, but it will always move within parameters which can even be calculated in advance under most circumstances.
I don’t know if organic life works in the same way, humans in particular and myself specifically. That is, if you knew enough about what I ate yesterday, how much I slept, which leg I got out first in the morning, etc etc, could you have calculated that I would show up at Jenna-Sue’s cafe this morning?
I am not vested in either answer, but I am curious what it is.
Correcting myself: a clock’s seconds dial moves at 6 degrees per second.
Just thought I’d put out this (to me) very useful write-up on the problem from 2004 by respected open source hacker Eric Raymond
http://esr.ibiblio.org/?p=161