Two thoughts:
-randomness: if the future is not determined it is completely unpredictable, in other words, it is random. Wouldn't the universe be a strange place if your decisions have a strong component of randomness? Perhaps in the next hour you would decide to take all the paper in your office and fold it into tiny boxes...
-Danger: since the future is already determined I can as well sit back and relax and don't worry. Don't fall into this trap.
Clearly, clear thinking is opaquely difficult. The future is not random, Roland, the future is just unknown. The future will have about as much structure as the now has, but we do not yet know all the details. This can possibly be construed as "randomness" in our thinking, but it is not randomness in the world. Another POV would call this "randomness in our thinking" as uncertainty. Roland "don't fall into this trap" means don't let determinism dertermine you - or worse - let Roland determine what you are determined to do. Maybe the Don't Panic button would be better and thusly your future was.
That was interesting, but I think you misunderstand time as badly as you expect us to misunderstand non-time
In regular time, the past no longer exists -- so there's no issue of whether it is changing or not -- and when we talk about the future changing, we're really referring to what is likely to happen in a future that doesn't exist yet.
A person living in a block universe could mistakenly think they have time by only perceiving the present. On the other hand, a person living in a timed universe could mistakenly think they live in a block by writing down their memories and expectations in a little diagram.
At some point in the future a SI could come a long and explain all those observations in terms of atoms etc and your "control" etc would go poof becoming epiphenomenal.
I repeat: There is a difference between explaining something, and explaining it away.
You can explain a rainbow in terms of optics, but that doesn't make it an epiphenomenon. The haunted mine has been emptied of gnomes, but the rainbow is still there.
For goodness sakes, I just covered all this territory a couple of months ago. Consider rereading the Reductionism sequence.
So you are saying that explaining something is equivalent to constructing a map that bridges an inferential distance, whereas explaining something away is refactoring thought-space to remove an unnecessary gerrymandering?
As Elezier points out, You are indeed a part of physics. Likewise, it does not make sense to consider the future determined from a human's perspective.
Let me try to explain that. You can consider any system with fixed rules and constant initial conditions to be solvable. So if you consider a rock rolling down a hill. Given any time t, you can predict the position of the rock.
If you consider a human a system in this same way, we can expect an output for any given input, however the data is much too complex to be represented graphically. For example consider writing a note to a friend. You then watch the friend write her response. As far as you're concerned, her output is determined based on the input, in this case the note, but more generally, the environment. You can then consider the function "friend(environment) = action." Indeed you can model the entire universe this way. However if you try to consider yourself this way, you will fail. See http://en.wikipedia.org/wiki/G%C3%B6del%27s_incompleteness_theorems
The debate about "free will" is mostly a matter of perspective. Looking from a third person perspective, we know that "free will" is really always a function of the environment. When we examine our own choices retrospectively, we see that our actions were a product of other factors, and we did exactly the only thing we could do.
But since any definition like this can only be considered from a third person, we say that we have free will.
In case I haven't made myself clear, consider a function f(x) = x+1. We say f is entirely determined by x. However, if you imagine yourself as the function f, you do not consider yourself predetermined. While you know that for whatever input x, you will return x+1, you have no idea what you will do until the PRESENT time, when you learn x. Our free will then is introspectively the mechanism by which we analyze our environment and make decisions.
A system with fixed deterministic rules and known initial conditions has an outcome that can be predicted. OTOH, the rules of QM are as fixed--unchanging--as you like, but do not allow definite prediction of experience.
We don't know anything about FW, including whether it is determined by the environment.
You can predict experiences quite reliably in aggregate. You can determine the percentages of you which will have experience A or B. The only uncertainty is which aspect of the amplitude flow you'll feel like you're in.
MWI doesn't just allow you to predict in ways that are of no practical use , it's "deterministic" in ways that are of no practical use.
Re: "Even without timeless physics, Special Relativity outlaws any global space of simultaneity".
Absolute time is like absolute space. The most we can say is that we haven't detected it yet. If it is ever detected, relativity will break - and what things it outlaws will no longer matter.
Yudowsky, I'm just applying parsimony/ockhams razor. If the configuration space is all that is needed to explain the next step, why do we also need to say, "a human controls the next state" as well. You are multiplying entities needlessly, going above the minimum message length, etc....
I think Will made my point. It doesn't seem to me like anything you wrote demonstrated that human choice is non-illusory. Granted that we seem to be part of physics, we can have experiences that make us think we're engaging in choice, we can experience cause and effect in our different points in configuration space, I don't see how that adds up to us actually engaging in choosing (as opposed to us experiencing thoughts and feelings of choosing, each of which are different parts of that configuration space).
I'm trying to see exactly where your assertion that humans actually have choice comes in. It's not clear to me, and the evidence I've seen is that cognitive scientists have already exposed much of the free will/human choice experience as illusory. So it seems reasonable to me that all of the human choice experience could be illusory (although the illusory experience could be a part of physics and normality too).
Being part of physics is not the same as being part of deterministic physics. We don't know whether physics is deterministic, and we don't know whether FW can or cannot be made out of the right structure under the right physics.
Human choice: why it exists, despite being inside of physics. Intelligence is the decision-making process. This is how our actions are determined. The experience of this decision-making process is called alternately "choice" and "free will". The causal relationship of our environments to our actions extends from observation, through our mental state and decision-making process, to our actions. If I use a different decision-making process, I make different decisions. This is still entirely inside of physics, but it hasn't been explained away. It can even be absolutely deterministic, when viewed from a third-person perspective. Saying we don't have "choice" is about as helpful as anything in the debate about free will
I'm trying to see exactly where your assertion that humans actually have choice comes in."choice" is a useful high-level abstraction of certain phenomena. It's a lossy abstraction, and if you had infinite amounts of memory and computing power, you would have no need for it, at least when reasoning about other entities. It exists, in exactly the same way in which books (the concept of a book is also a high-level abstraction) exist.
I also have a question of my own, regarding the rock-hill-system:
If you isolate a subsystem of reality, like a rock rolling down hill, then you can mathematically define the future-in-isolation of that subsystem; you can take the subsystem in isolation, and compute what would happen to it if you did not act on it. In this case, what would happen is that the rock would reach the bottom of the hill.How does this isolation work? Do you assume that the forces acting on the system from outside stay constant (in some undefined fashion), without explicitly modeling the outside? If I assume no further interactions with the outside, I don't expect to see the rock rolling down the hill, since there's no planet below to gravitationally attract it. Or was the planet supposed to be part of this system?
Will, Hopefully: When you say "the configuration space" you have the human. The human is part of the configuration space. It's not multiplying entities to have a human and his hand or a hand and the hand's figures. A different configuration space, one that differed only in that it contained a different human rather than this human, would cause a different future.
Tim: The wheel of science doesn't roll backwards. We know why we thought we had absolute time at a level of abstraction that encompasses those situations in which we had apparent evidence for absolute time. What would it even mean, at this point, to say that we had discovered absolute time. What would the experiments have to look like? Do you think that we might find out that there is "real heat", not just thermal motion, too?
From what I understand the configuration space adds up to humans controlling the next state. This reminds me of when Eliezer was speaking about how one can separate out all the bones, tendons and ligaments of the hand and understand each one in turn but you still have to add them all up to a hand for them to function in their purpose as a hand..A bunch of bones, tendons and ligaments laying around separately may be useful in certain situations but they can't be a hand unless you combine them.
Sebastian, Michael, Cassandra, you seem to be side-stepping the central question. It seems reasonable to me that "choice" is an illusory experience in the same way that much of what we experience visually is illusory.
http://www.sciam.com/article.cfm?id=the-neuroscience-of-illusion
I'm trying to understand where Eliezer's expressions of certainty that human are capable of choice (rather than the illusory experience of choosing) comes from?
If probability is always a part of decision-making algorithm, embedded within physics, what does it mean to attach probability to timeless physics? How can global causality be described probabilistically? What is the decision that is being considered?
Eliezer's point (a quite justified one) is that the word "choice" is a name for something that human beings do, just as the name "apple" is a name for something human beings find in the world. Whatever you think an apple is, if you say it is only an illusion, then you're not talking about apples, but something else. Likewise, whatever you might think a choice is, if you say it is only an illusion, you're not talking about choices, but something else. For choice just means one of the things that people actually do in the real world, so it is quite real, not an illusion.
I think when people say that apples and choices are illusions, they might mean that they are patterns recognizable by people but not fundamental: if some system couldn't recognize an apple (perhaps only because it never had any reason to form the concept) but did have a model of the amplitude distribution of the universe, it would get along just fine (actually it would probably just have different high-level concepts).
If your mind contains the causal model that has "Determinism" as the cause of both the "Past" and the "Future", then you will start saying things like, "But it was determined before the dawn of time that the water would spill - so not dropping the glass would have made no difference".
Nobody could be that screwed up! Not dropping the glass would have been no option. =)
About all that free-will stuff: The whole "free will" hypothesis may be so deeply rooted in our heads because the explanatory framework of identifying agents with beliefs about the world, objectives, and the "will" to change the world according to these beliefs and objectives just works so remarkably well. Much like Newtons theory of gravity: In terms of the ratio of predictive_accuracy_in_standard_situations to operational_complexity Newton's gravity kicks donkey. So does the Free Will (TM). But that don't mean it's true.
Unknown, your comment strikes me as a good way of looking at it.
The "me of now" as a region of configuration space contains residue of causal relationships to other regions of configuration space ("the past" and my memories of it). And the timeless probability field on configuration space causally connects the "me of now" to the "future" (other regions of configuration space). Just because this is true, and -- even more profoundly -- even though the "me of now" configuration space region has no special status (no shining "moment in the sun" as the privileged focus of a global clock ticking a path through configuration space), I am still what I am and I do what I do (from a local perspective which is all I have detailed information about), which includes making decisions.
Our decisions are based on what we know and believe, so an acceptance of the viewpoint Eliezer has been putting forth is likely to have some impact on decisions we make... I wonder what that impact is, and what should it be?
If I follow this correctly, choices are both deterministic and non-illusory.
The traditional line of thinking is something along the lines of "if my choices are determined by something else, they are illusion, and therefore do not matter." If the choices were illusory, then—if I have correctly understood—removal from the system would not have an effect on the system. Which is to say, 'past' leads to 'choice'—I'm unsure if 'present' is more correct in this case—leads to 'future' is indistinguishable from 'past' leads to 'future'
However, this is something akin to saying '1 + 1 = 2' is indistinguishable from '1 = 2'. Both the '+ 1' and 'choice' are integral to '2' and 'future', respectively. Furthermore, the reverse—'2 = 1 + 1' and 'future is outcome of choice that is determined by past'—also needs the middle to work correctly.
So—to relate this to terms more intuitive—the past determines the choices we are presented with and the options we take, but the future is the direct outcome of the choice, not the past. Our choices are non-illusory because they have effect. Our choices are more deterministic than we would like to think. After all, I know I like to think that I'm the only determinator in my choices, but that would ignore my thoughts on the current situation, the outcomes of similar choices made in the past, and what has happened to me recently.
Now, I wonder what happens if we have—instead of a function f(n-1) = n, where n is a node along a chain of causality—a function f(n) = n - d, where d is some distance along the chain. Put another way, what if a choice were to somehow effect what we consider the past—or, if n is negative, the future—and this discontinuous function affected the chain in such a way as to affect node n? After all, if 'past', 'present' and 'future' are only the way we interpret causality, what prevents a chain from looping back on itself as one of many inputs.
In asking the question, I may have answered it myself. I was thinking timefully, in that adding a new node would affect an old node, where instead there are no 'new' or 'old' nodes. The nodes 'are'.
Perhaps any node which appears to affect the chain is an illusion, and instead is merely another chain of causality that is similar to the chain being considered. The present doesn't change the past, it links to something that looks like the past.
The argument that determinism is inimical to FW isn't the claim that you can remove the link labelled choice without changing anything else. It is partly based on the observation that the link is not in a privileged position that should objectively make it THE cause. Not only is it not the furthest back, it is also not the nearest and most I'mmediate cause of an event, since there is always some further mechanism needed to turn a choice into a successful action.
Furthermore, deterministic "choice"doesn't have all the expected properties of free choice. Particularly , it lacks could-have-been-otherwise.
Furthermore, deterministic "choice"doesn't have all the expected properties of free choice. Particularly , it lacks could-have-been-otherwise.
There is no conflict between determinism and counterfactuals.
"Had this tumbler been made of plastic instead of glass, it would not have broken when it fell on the floor."
"Had Krakatoa not erupted, the remarkable sunsets subsequently seen around the world would not have happened."
"Had Mars been larger, it would have retained a denser atmosphere."
And a favorite on LessWrong:
"Had the millionth digit of pi been 0, this program to calculate it would have output 0."
Therefore such counterfactuals are not the right model for free choice. In particular, they are not an ability someone can exercise.
Compare and contrast:
"If I run to catch my train, I will arrive on time."
"Had I run to catch my train, I would have arrived on time."
These say the same thing, one from before the decision and one from after. If you reject the latter, how can you accept the former? If you reject the former, how can you run your life at all?
Consider "if I run to catch the train, i would arrive on time. I have the ability to choose to run or not. I chose not to run, and so missed the appointment. I could have chosen differently. I regret choosing as I did"
Free choice requires and depends on counterfactuals, but consists of more than counterfactuals.
Free choice requires and depends on counterfactuals, but consists of more than counterfactuals.
Then I don't know what you mean by "free choice". What I mean is that I go through a process of deciding; in the present example by considering the effort required, the chance of being successful, and how much it matters. When it is clear to me making that effort is the best decision, that is what I will inevitably do. That is the exercise of free will, as practically understood in real life, from where the concept comes: making decisions when one's own consideration of the issue screens off all other causes. It is still meaningful to say afterwards, "had I not caught the train, I would have been late", despite the inevitability of the actual decision.
People who have faced up to difficult moral decisions sometimes say afterwards words to the effect that "I could not have acted otherwise," meaning that the right path was so clear, despite the personal sacrifice involved, that they had to take it. In saying that, they are not claiming to be blind automata.
By freechoice I mean a putative capacity that is incompatible with determinism, and which underlies common attitudes of culpability, congratulation, regret, etc.
If your decisions are inevitable, why regret them? Regret is common attitude, so the common notion choice is one where decisions aren't inevitable.
That people do not regret some of their decisions -I could/would not have done otherwise- does indeed not mean that they regard themselves as automata whose decisions can only be inevitable. Why do you?
By freechoice I mean a putative capacity that is incompatible with determinism
Writing nondeterminism into the definition doesn't establish anything about the real world.
and which underlies common attitudes of culpability, congratulation, regret, etc.
Common attitudes do not depend on the arguments of philosophers (although they may well get influenced by such deepities as get into the popular air). Debates about determinism show up in the law courts only to the extent that a person's deliberation drastically fails to screen off other causes of their actions (such as insanity, drug influence, etc.). The law does not need to deal with philosophical spooks.
Now, there are any number of contemporary intellectuals, Sam Harris for example, ready to argue that there is no such thing as free will, choice is an illusion, we are all just sacks of competing brain modules, the self does not exist, nothing it true, all is a lie, and any number of such deepities. I think they're all wrong, but I think that free will as a spooky extra-deterministic force is also wrong.
That people do not regret some of their decisions -I could/would not have done otherwise- does indeed not mean that they regard themselves as automata whose decisions can only be inevitable. Why do you?
I don't.
I used the word "putative" in the hope of signaling that I was not attempting an armchair argument for the actual existence of FW. I was, however, launching an armchair argument for the incompatibilist concept of FW being the correct concept, as opposed to incompatible.ism. If it is correct, the actual existence of FW would depend on empirical factors, such as the actual existence of determinism (which is rather different to the situation if compatibilism is correct)
The arguments of philosophers should depend on common concepts, the notion of FW that people use and care about. The existence of regret shows that people care about a notion of FW that involves accessible contractual worlds. The compatibilist can only offer inaccessible worlds, ie if the Big Bang had been different, you would have been determined to do differently, whereas the incompatibilist maintains that you could have done differently by your own choice.
I dont bet onthe idea that FW is nonexistent , as per Harris, nor on the idea that it is triviallly compatible with determinism, as per Dennet. Incompatibilist FW only has to override determinism if determinism is actually the case, which is an empirical, not a conceptual issue.
There is no conflict between determinism and counterfactuals.
These counterfactuals are untestable, unobservable, and do not actually exist
Choose one!
We can and do test counterfactuals by by re-running experiments with different starting conditions. The claim that
These counterfactuals are untestable, unobservable, and do not actually exist
...is profoundly counter-scientific.
(There is serious issue about locality. We can't rewind the universe and vary the starting conditions , we can only do so repeat a localised experiment with variations. But science works...doesn't it.)
There is no conflict between determinism and counterfactuals.
These counterfactuals are untestable, unobservable, and do not actually exist
Choose one!
We can and do test counterfactuals ...by by re-running experiments with different starting conditions. The claim that ...
These counterfactuals are untestable, unobservable, and do not actually exist
...is profoundly counter-scientific.
(There is serious issue about locality. We can't rewind the universe and vary the starting conditions , we can only do so repeat a localised experiment with variations. But science works...doesn't it.)
All the ideas expressed in this post, as well as the "timeless physics" one, seems amazingly obvious to me, and has for all of my adult life, and compared to a lot of OB posters, I am not that bright. Since I normally find many of Elizer's posts extremely counterintuitive and/or hard to grasp, I've got to ask the question: am I missing something here? Is Elizer saying something so mind-boggling out of this world that I do not even realize he is saying it?
Eliezer, you're spot on with the "Determinator." The modern free will debate has its roots not in the clockwork universe of Newtonianism but the supposed problem of God's omnipotence and omniscience. The problem of free will was originally formulated in terms of a Determinator - God - who chose and imminently caused the future. The question was "How can we also have free will?" and free will was, of course, also an important concept in Christian theology (we're made in God's image and therefore chose and cause our futures just like God does). As is often the case in philosophy the current debate is just a secularization of the theological debate; they just switch "God" for "Universe", "soul" for "essential property," etc, and carry on having the same arguments.
And second, you can't compute the Future from the Past, except by also computing something that looks exactly like the Present; which computation just creates another copy of the Block Universe (if that statement even makes any sense), it does not affect any of the causal relations within it.
I'm not sure that statement does make sense. It sounds a bit too mystical to me. But it'd be interesting to look at it from a thermodynamic perspective. You can't predict the future from the past without doing work in the present. Perhaps the work needed would always be greater than or equal to that required for the system you're predicting to just play out regardless?
Wibby:
After all, I know I like to think that I'm the only determinator in my choices, but that would ignore my thoughts on the current situation, the outcomes of similar choices made in the past, and what has happened to me recently.
Hand is to fingers as "I" is to "my thoughts on the current situation, the outcomes of similar choices made in the past, and what has happened to me recently".
A hand has more to it than fingers, but still, the relation is whole to part, not inside to outside.
Sebastian:
How does this isolation work? Do you assume that the forces acting on the system from outside stay constant (in some undefined fashion), without explicitly modeling the outside?
That would save on computing power, yes. If you were playing a game against someone else, you would probably model yourself, the opponent, and "the hill plus the rest of the universe" but you wouldn't actually think of the whole universe, just the parts that affected the hill. You would think of gravity but not Earth, and so on. If you think as though the whole goal is to save on computing power, and that the brain is actually fairly good at this (it has to be), then you won't go far astray.
The defenses of non-illusory human capacity to make choices, as presented in this thread after my last post, seem weak to me.
Dennett's got an analogy to address how choice can be both deterministic and non-illusory. He asks his audience to consider a deterministic chess-playing algorithm. You can play the same game with this algorithm over and over -- it doesn't learn. If you look at its internal state, you can see it generating ply-reply trees and evaluating the positions thus generated. In this view, "making a choice" reduces to "running a decision-making algorithm". The computer chess player doesn't have the cognitive apparatus to have an illusory experience of doing anything, and yet it remains meaningful to speak of the reasons it has for making the choices it does.
HA, what do you think of this analogy? (tone: genuine curiosity)
Free will is about free choices, choices that could have been different. Unfree choices are trivially compatible with determinism.
I like that analogy. It helps make it clear that we may have a "free will of the gaps", phrase borrowed from "god of the gaps". What we call free will, or a claimed ability to choose arbitrarily, may be an incomplete understanding of the input factors determine our subsequent behavior and experience of that behavior. The chess-playing algorithm you describe also seems analogous to a billiard ball, bouncing around a table due to prior and current conditions. The danger in calling it non-illusory choice (in chess playing algorithms or in humans) is that that word use may play to our cognitive biases, like calling rain "god tears" or cancer "wrongdoing punishment".
What we call FW may also be a correct understanding of a real mechanism. You are priviledging one hypothesis.
"In Thou Art Physics, I pointed out that since you are within physics, anything you control is necessarily controlled by physics."
I could just as easily argue since I'm within my past self's future light cone, anything I control is/was necessarily controlled by (a younger) me. In both cases we are playing with words and muddying the waters rather than learning or teaching.
I don't see why you can't just reverse the logic and claim that since everything in my mind is controlled by physics, thought is an act of my free will. I don't believe in strong free will. But I do believe by the time a toddler can form ideals (desires ice cream) that aren't real, some free will is already at work. The theory (math is not subject to General Relativity and thus "deterministic" is this description has nothing to do with the "deterministic" used to describe human actions)of MWI may be deterministic, but playing with English language words suggests actors can't choose their world-lines by using the physics of their minds to cascade synchronized neural firing patterns that activate the parts of our brains producing minds. Maybe there is no free will, but I'd need to see a convincing theory of consciousness absent circular reasoning. The Plinko disc make fall determinalistically, but if the Plinko chip had a human CNS and accurate memories of past drops, I bet it might try to rotate in a preferred fall path, and if the Plinko chip based its decision on reflected ideals, I'd say there is some free will there (neuron firing seems at a small enough scale to harness some of the quantum-spooky-stuff that causes universes to split off, for instance. I think our brains can control the % of world-lines that decide whether to binge eat ice cream. Equating a block universe to MWI assumes there is an end state where the total ratio of all time-space co-ordinates is known. in reality, this end state does not exist (as time breaks down outside reality, like when forming the mathematical concept of a block universe). There are many random events that control which world-line an individual experiences, but I don't see why volitions can't be among the cases. I doubt few people defending free will really mean to defend their right to bring about their own birth.
Er, to try to simply my above point: in my model, energy (say, an atom) at time-sequence t1, sums up all its interactions with the rest of its local universe (such as a CNS if it is a brain atom), and this "calculation" affects the weighting of sick-of-ice-cream t2, t2a, t2b, world-lines. In claiming MWI is a block universe, you are accepting t1 ping-pongs to the subsequent split world-lines t2, t2a, t2b, without any "calculation" as described.
Ultimately it is a question of what limits are imposed on the splitting off of new world-lines in the multiverse. The speed-of-light, yes. I don't see why the physics of mind couldn't also qualify.
Eliezer:
If you think as though the whole goal is to save on computing power, and that the brain is actually fairly good at this (it has to be), then you won't go far astray.Ah, thanks! I hadn't considered why you would think about isolated subystems in practice; knowing about the motivation helps a lot in filling in the implementation details.
I'm with Phillip Huggan here.
I also noticed along with "Poke" that this is a common religious discussion, but I would remark that it is not respective to Christianity. I live in a Muslim country and this a common topic debate amongst Muslims themselves. It is not really Christian. It is an attempt to explain reality for all people.
It seems a little out of reach to try to understand the relationship of time and the block universe. Should we really expect it to be simple? Is everything understandable? Could the block universe be other than you described?
Philosophy can imagine all sorts of logical things that are not based in reality. Maybe it is explainable by determinism, but does that mean that determinism is true?
Have a few questions though.
What rules govern the block universe?
What if there were multiple block universes?
HA: How come you think I defend any "non-illusory human capacity to make choices"? I am just wondering why the illusion seems so hard to get rid of. Did I fail so miserably at making my point clear?
Eliezer:
Although the I chose bits that were part of the whole, I think they are useful to consider how parts inform the whole, and use the parts-to-whole relation to—at least crudely—model the outer-to-inner relation.
HA:
What I was attempting to say is that the human mind appears to me to be a chaotic system. While it may be entirely deterministic, the outcome can be radically changed by small inputs.
The usage of the word 'illusion' as I am interpreting it is akin to "since all things are made up of a small amount of atoms and a large amount of space, the sensation of solidity is an illusion." This, I suppose, is true, but it leaves out a rather large bit about the fundamental forces. Saying choice is an illusion appears to me to leave out the rather large bit of the workings of the brain.
Frank, Demonstrated instances of illusory free-will don't seem to me to be harder or easier to get rid of than the many other demonstrated illusory cognitive experiences. So I don't see anything exceptional about them in that regard. Doug, that's where "choice of the gaps" comes into play in my opinion. Some commenters, maybe you included, are saying "a lot of what goes on the brain is unknown. Therefore let's call 'choice' or 'free will' a subset of that unknown part." It's very analogous, in my opinion, to "god of the gaps" arguments. Non-illusory choice/free will may end up being something that many of our brains want to believe exists, like a creator god, but doesn't seem to be backed by our better empirical methods for observing and modeling reality.
The nature of time has been covered by many great minds from a religious viewpoint, as mentioned by nick. It is also an active research topic among mainstream universities. I'm not particularly interested in the question, but the best analysis I've read comes from a few N.Bostrom papers, and a book I once read called "Time Machines". The book supposes a block universe, but states very clearly that this may not be the way the universe operates. From what I understand, this means the opposite of what EY wrote. It means the Copenhagen determination (that magic causes wavefront collapses) is a block universe. From my understanding, MWI means the universe would only be deterministic if there were no tachyons (I'm not sure, but I think these are predicted in most GUTs), otherwise there would be feedbacks. Even if no tachyons, the universe would only be deterministic in a past direction. The real question is what causes universes to split off. This is deep physics. There are papers on this topic. If someone were to suggest one, I would read it. The whole point of Tipler's "The Physics of Immortality", was to use shearing forces in a collapsing universe (universe strongly appears to be open, unfortunately), as an energy source. Where would a never ending universe fit when viewed through block universe goggles? Once again I ask, don't tachyons eliminate the block universe concept for all energy except photons travelling at c?
I'm not discouraging discussion. But there are some topics where this may be a cutting edge dialectic, such as the nature of minds, the computational power limits if any to recursive AI software programs, and AGI/AI controls. But this debate is inferior to mainstream university research. Keep it up, but the real question is how much money to spend on particle accelrators and observatories, that might resolve these basic physics questions. The money people use mainstream physicists as their info sources. These mainstream physicists have written papers. If EY's "block universe" hypothesis were correct, we wouldn't experience time. Simple anthropic reasoning disproves it. Time exists. The future is more important than the past. If anyone takes the time to find papers that deal with splitting off universes, I'd attempt to read them and discuss. I hope if mildly recursive software AI systems are built in the decades ahead and the human brain/mind is modelled by IBM or whoever, that those interested here in AI/AGI will keep up with these findings and not continue to discuss "inferior" content. Maybe I'm just pissed because I realize blogs where GUT amateurs talk about time, have limits.
Off-topic, but I suggest EY's idea of an AGI using mixed chemicals to form a mobile robot (and assumedly hack the internet), is now dated. With rep-rap and ink jet polymers, rapid plastics prototyping...a far more likely scenario is that an AI would hack a printer and output some sort of shape-memory device or conducting plastic as an origami crane. Normally this is a moot point, but there may be real defenses that could be dreamed in these sorts of discussions. If it is not known whether AGI is possible with a 2000BC Egyptian wooden abacus, or needs a computer from 10000000AD, but we know people may try to use the same sort of technologies and/or hacking procedures as weapons, why not diversify one's fields-of-expertise? If I were to suggest AI/AGI prescriptions to cyber police, I'd suggest cracking down on Eastern European, Russian and Chinese virus writers and better funding the good guys.
HA, Dennett's a compatibilist, so his analogy is meant to demonstrate that making a choice is not an illusory experience. That's the part I was talking about when I said we can meaningfully discuss the choices it makes.
For example, we can analyze the game and say, "The algorithm blundered in this move -- it ignored a line of play which leads to a significant disadvantage," or perhaps, "This move was excellent -- the algorithm decided to sacrifice material for much greater activity for its developed pieces, allowing it to dominate the board; it will probably be able to force a win." The fact that we can get into the guts of the code and point to the heuristics and evaluating functions that led to these plays does not invalidate the fact that the algorithm really did make choices. For conscious beings, the content of the experience of making a choice is in the evaluating and the acting, not in the exercise of some kind of "free will" that requires the essence of choice to exist outside a deterministic physics.
Given this framework, I'm not really seeing any danger in calling choices non-illusory.
Cyan, by that logic you could say that about a fire "it made a good choice to burn here, a bad choice to burn there" given that the spread of the fire would be as determined in your model as the development, existence, performance, and persistence of the chess-playing algorithm, or the human.
If the alogorithm was determined, it seems to me that claiming directions it went were "choices" is about as accurate as claiming the directions a fire goes are "choices". I think we can discuss phenomena like an algorithm playing chess against something else, and the results (win, loss, stronger position, weaker position) without using the word "choice" just like we can describe a fire's interaction with its environment without ascribing "choice" to the fire. To selectively use that word may prey on a common bias that we're susceptible to.
I'd write more to make my thoughts on this more clear, but I'm tired and I think my previous writing in this thread holds up pretty well.
HA, your analogy fails to hold because the fire isn't performing a computation, and hence cannot be said to be evaluating the outcomes of any actions.
Cyan, I'm not sure you about assertion that the fire isn't performing a computation. For example, we could structure the fire's environment such that in burning it performs a computation. But then one could look at any "natural" environment a fire burns in as something it is similarly performing a computation of. Also, any example you'd give of computations to distinguish them from fires burning may be viewable as energy passing through structured and/or "natural" environments.
I'm sure people smarter and more expert than I am have put good thought into this, for example in information theory.
Frank, Demonstrated instances of illusory free-will don't seem to me to be harder or easier to get rid of than the many other demonstrated illusory cognitive experiences. So I don't see anything exceptional about them in that regard.
HA, I do. It is a concept I suspect we are genetically biased to hold, an outgrowth of the distinction between subject (has a will) and object (has none). Why are be biased to do so? Because, largely, it works very well as a pattern for explanations about the world. We are built to explain the world using stories, and these stories need actors. Even when you are convinced that choice does not exist, you'll still be bound to make use of that concept, if only for practical reasons. The best you can do is try to separate the "free" from the "choice" in an attempt to avoid the flawed connotation. But we have trouble conceptualising choice if it's not free; because then, how could it be a choice? All that said, I seem to remember someone saying something like: "Having established that there is no such thing as a free will, the practical thing to do is to go on and pretend there was.".
My usual Block Universe analogy is a book. What time is it right now in Middle Earth? That question does not make much sense, although we coherently ask a reader, "What point are you at?" A page is a time-slice of that universe. We stand outside that universe the way a hypothetical observer stands outside time in a Block Universe.
I have trouble putting myself back in a frame of mind where some questions sound coherent. To take Eliezer's haunted mine example, it is like asking, "But where do the gnomes go? If they have been emptied out of the mine, they must have gone somewhere. That's simple conservation of matter." There is a fundamental problem with the question.
Note that, once we start talking meta-universe, our words only have meaning as analogies. What does "outside time" really mean? Time is not the sort of thing that you get outside of, any more than you get to the left of it. You also do not get before time or outside space. There is no before without time, nor anywhere to be outside space. It would be like being faster than the number three. The syntax is right, but that is not how those words work.
My usual Block Universe analogy is a book. What time is it right now in Middle Earth? That question does not make much sense, although we coherently ask a reader, "What point are you at?" A page is a time-slice of that universe. We stand outside that universe the way a hypothetical observer stands outside time in a Block Universe.
This is a really old comment to be replying to, but the intuition pump presented here is so good (for me, anyway) that I marvel that I didn't think of it myself! I guess that's what separate merely understanding something and understanding it well enough to explain it.
HA, I think you're right that fires can be said to be performing computations (in a deterministic universe). What the chess algorithm does that makes it different from a generic computation is goal-oriented actions driven by an explicit evaluation of possible outcomes. (Computation is necessary but not sufficient for this; I took a wrong step in bringing up generic computation.)
I'll steal another analogy from Dennett. Your constituent molecules are not alive, but you are. Likewise, your constituent parts considered at a low level may not make choices, but you do. Both "life" and "choice-making" are properties of the arrangements of the bits you're made up of. Being aware of making choices is another such property.
In my view, the worthwhile things to talk about when discussing a particular choice someone or something made are (i) the information available to the choice-maker, and (ii) the evaluation function it used to rank the available actions. I strongly reject the idea that such a discussion would be invalidated or made meaningless in a deterministic universe, which is where I think the "it's dangerous to reify illusory choice" position takes us.
Frank Hirsch wrote: "Having established that there is no such thing as a free will, the practical thing to do is to go on and pretend there was.".
I would agree with this, similarly for everything else in normality ™.
Unknown wrote: "Eliezer's point (a quite justified one) is that the word "choice" is a name for something that human beings do, just as the name "apple" is a name for something human beings find in the world. Whatever you think an apple is, if you say it is only an illusion, then you're not talking about apples, but something else. Likewise, whatever you might think a choice is, if you say it is only an illusion, you're not talking about choices, but something else. For choice just means one of the things that people actually do in the real world, so it is quite real, not an illusion."
And I am saying it is a mistake to mix up this thinking with fundamental physics such as controlling the width of the worlds or other many world concepts. This is the stuff of the laplacian demon and it needs no concepts of "apple", "choice" or "human" to predict the universe. Admittedly it probably has to be outside this universe as we know it (i.e. we are in a simulator or something).
Michael Vasser: "When you say "the configuration space" you have the human. The human is part of the configuration space. It's not multiplying entities to have a human and his hand or a hand and the hand's figures. A different configuration space, one that differed only in that it contained a different human rather than this human, would cause a different future."
How do I know I have a human or not in a configuration space. Please transmit a program for determining whether I have human or not with a description of the configuration space and rules of the system. Do not increase the size of the program above one that only has the configuration space, or tell me what more having the human detection part would allow me to predict about the future state of the configuration space.
"Having established that there is no such thing as a free will, the practical thing to do is to go on and pretend there was."
The thing to do is go on and ignore the question altogether. When I deliberate, I'm not wracked with anxiety about whether I have Free Will. I just go about deliberating. "I deliberate" means I deliberate -- whatever else that means; thinking about the else won't make me a more effective deliberator.
If you're still thinking about turning this sequence into a book (or have you already done so? This business of reading and responding years after the fact is a bit disorienting sometimes.) I would strongly recommend adding an earlier discussion that makes a weaker version of the same argument from cognition.
My experience is that exploring the machinery of cognition tends to weaken the intuition of unbounded free will, and tends to be more accessible than the machinery of quantum mechanics. That is, it's easier to demonstrate to people that there are cases where we make what feel like unconstrained choices that are nevertheless constrained.
And once you've given up the naive idea of free will, once you've accepted that the experience of choosing among options is something we construct rather than something that is "just there," it's easier to accept the idea of a timeless block universe.
You have to empty the cup before you can fill it.
Of course, you do touch on this in your discussions of cognitive biases -- what is a bias, after all, if not a constraint on the thoughts that I'm "free" to think? -- but more explicitly connecting the dots might be valuable.
Similarly, I find it's helpful when exploring determinism to let go of the emotional attachment to making things go a certain way. Otherwise, you end up thinking things like "Well, if the future is determined anyway, then I might as well , because if I do then it's what I was going to do, right?" and so forth.
If you're still thinking about turning this sequence into a book (or have you already done so? This business of reading and responding years after the fact is a bit disorienting sometimes.)
I believe the first draft is nearly complete, and he's approaching the seeking-a-publisher stage, followed by the writing-a-second-and-third-draft stage
For "I did A but could have done otherwise" I see two coherent meanings:
1) My mind produced A from the local conditions, but a conceivable different mind with otherwise identical local conditions would've produced not A. My mind is therefore a crucial causal factor in the reality of A.
OR
2) From my limited knowledge, I cannot trace the causal steps to A that precede my decision well enough to determine, from those steps alone, the decision I make which leads to A.
...actually, probably both.
So, the causal steps to A include my decision (and A is inconsistent with certain decisions that differ from my real one), but I cannot trace the causal steps of my decision precisely enough to have precluded those differing decision (without already knowing the reality of my decision.)
Alternatively: if we work from full knowledge of the causal path to A, except that we treat my cognition as a black box whose outcome we don't know, we could not conclude A even with unlimited processing power.
This post really knocks it out of the park. The diagrams are very appropriate and clear. If the book of sequences is still a (re-)work in progress, I strongly encourage making this part prominent.
One must avoid mixing up timeless and timeful thinking. E.g., trying to have "Determinism" acting on things before they happen. Determinism is a timeless viewpoint, so it doesn't mix well with words like "before".
I was going to write something quite similar in my own blog - and I might still - but I thought, "I didn't read the whole free will sequence, I wonder if EY mentioned the block universe?" My thunder done been stole! Well, mostly.
Let's say there was some way to change a single moment of time.
We would then need some kind of meta-time over which time could "change".
The lamp's state would need to change from "OFF at 7:02am at 3:00meta-am" to "ON at 7:02am at 3:01meta-am".
But wait! Have you ever seen a lamp change from OFF at 7:02am at 3:00meta-am, to ON at 7:02am at 3:00meta-am? No! A single instant of meta-time never changes, so you cannot change it, and you have no control.
"Have you ever seen?" refers to a state of a (mere) time computer. Your memory is time-caused so it is not reliable evidence about meta-time mechanics. It is an important obliviousness result for chronal beings. It might not be holeless. I don't know about more general settings but with "propagation" (meta-time change being local in (normal) time), even a chronal mind in the proximity of CTL could occasionally (with some appriciable fraction of meta-time) notice things that are not compatible with a totally frozen timeline (this does need recurring meta-time switching back and forth to not be infinidesimal (in the setting that I know CTL can also provide such a meta-time switching source)).
So with two lamps with different meta-time wire lengths (essentially different proper times) you could have all four states of "A on, B on", "A on, B off", "A off, B off" and "A off, B on" cycle and have corresponding chronal observer experience each. None of them see the lamps change in (normal) time, but it still takes place.
Followup to: Timeless Physics, Timeless Causality, Thou Art Physics
People hear about many-worlds, which is deterministic, or about timeless physics, and ask:
If the future is determined by physics, how can anyone control it?
In Thou Art Physics, I pointed out that since you are within physics, anything you control is necessarily controlled by physics. Today we will talk about a different aspect of the confusion, the words "determined" and "control".
The "Block Universe" is the classical term for the universe considered from outside Time. Even without timeless physics, Special Relativity outlaws any global space of simultaneity, which is widely believed to suggest the Block Universe—spacetime as one vast 4D block.
When you take a perspective outside time, you have to be careful not to let your old, timeful intuitions run wild in the absence of their subject matter.
In the Block Universe, the future is not determined before you make your choice. "Before" is a timeful word. Once you descend so far as to start talking about time, then, of course, the future comes "after" the past, not "before" it.
If we're going to take a timeless perspective, then the past and the future have not always been there. The Block Universe is not something that hangs, motionless and static, lasting for a very long time. You might try to visualize the Block Universe hanging in front of your mind's eye, but then your mind's eye is running the clock while the universe stays still. Nor does the Block Universe exist for just a single second, and then disappear. It is not instantaneous. It is not eternal. It does not last for exactly 15 seconds. All these are timeful statements. The Block Universe is simply there.
Perhaps people imagine a Determinator—not so much an agent, perhaps, but a mysterious entity labeled "Determinism"—which, at "the dawn of time", say, 6:00am, writes down your choice at 7:00am, and separately, writes the outcome at 7:02am. In which case, indeed, the future would be determined before you made your decision...
At 7:00am you're trying to decide to turn on the light bulb. But the Determinator already decided at 6:00am whether the light bulb would be on or off at 7:02am. Back at the dawn of time when Destiny wrote out the Block Universe, which was scripted before you started experiencing it...
This, perhaps, is the kind of unspoken, intuitive mental model that might lead people to talk about "determinism" implying that the future is determined before you make your decision.
Even without the concept of the Block Universe or timeless physics, this is probably what goes on when people start talking about "deterministic physics" in which "the whole course of history" was fixed at "the dawn of time" and therefore your choices have no effect on the "future".
As described in Timeless Causality, "cause" and "effect" are things we talk about by pointing to relations within the Block Universe. E.g., we might expect to see human colonies separated by an expanding cosmological horizon; we can expect to find correlation between two regions that communicate with a mutual point in the "past", but have no light-lines to any mutual points in their "future". But we wouldn't expect to find a human colony in a distant supercluster, having arrived from the other side of the universe; we should not find correlation between regions with a shared "future" but no shared "past". This is how we can experimentally observe the orientation of the Block Universe, the direction of the river that never flows.
The Past is just there, and the Future is just there, but the relations between them have a certain kind of structure—whose ultimate nature, I do not conceive myself to understand—but which we do know a bit about mathematically; the structure is called "causality".
(I am not ruling out the possibility of causality that extends outside the Block Universe—say, some reason why the laws of physics are what they are. We can have timeless causal relations, remember? But the causal relations between, say, "dropping a glass" and "water spilling out", or between "deciding to do something" and "doing it", are causal relations embedded within the Block.)
One of the things we can do with graphical models of causality—networks of little directed arrows—is construe counterfactuals: Statements about "what would have happened if X had occurred, instead of Y".
These counterfactuals are untestable, unobservable, and do not actually exist anywhere that I've been able to find. Counterfactuals are not facts, unless you count them as mathematical properties of certain causal diagrams. We can define statistical properties we expect to see, given a causal hypothesis; but counterfactuals themselves are not observable. We cannot see what "would have happened, if I hadn't dropped the glass".
Nonetheless, if you draw the causal graph that the statistics force you to draw, within our Block Universe, and you construct the counterfactual, then you get statements like: "If I hadn't dropped the glass, the water wouldn't have spilled."
If your mind contains the causal model that has "Determinism" as the cause of both the "Past" and the "Future", then you will start saying things like, But it was determined before the dawn of time that the water would spill—so not dropping the glass would have made no difference. This would be the standard counterfactual, on the causal graph in which "Past" and "Future" are both children of some mutual ancestor, but have no connection between them.
And then there's the idea that, if you can predict the whole course of the universe by looking at the state at the beginning of time, the present must have no influence on the future...
Surely, if you can determine the Future just by looking at the Past, there's no need to look at the Present?
The problem with the right-side graph is twofold: First, it violates the beautiful locality of reality; we're supposing causal relations that go outside the immediate neighborhoods of space/time/configuration. And second, you can't compute the Future from the Past, except by also computing something that looks exactly like the Present; which computation just creates another copy of the Block Universe (if that statement even makes any sense), it does not affect any of the causal relations within it.
One must avoid mixing up timeless and timeful thinking. E.g., trying to have "Determinism" acting on things before they happen. Determinism is a timeless viewpoint, so it doesn't mix well with words like "before".
The same thing happens if you try to talk about how the Past at 6:30am determines the Future at 7:30am, and therefore, the state at 7:30am is already determined at 6:30am, so you can't control it at 7:00am, because it was determined at 6:30am earlier...
What is determined is a timeless mathematical structure whose interior includes 7:00am and 7:30am. That which you might be tempted to say "already exists" at 6:00am, does not exist before 7:00am, it is something whose existence includes the Now of 7:00am and the Now of 7:30am.
If you imagine a counterfactual surgery on the interior of the structure at 7:00am, then, according to the statistically correct way to draw the arrows of causality within the structure, the 7:30am part would be affected as well.
So it is exactly correct to say, on the one hand, "The whole future course of the universe was determined by its state at 6:30am this morning," and, on the other, "If I hadn't dropped the glass, the water wouldn't have spilled." In the former case you're talking about a mathematical object outside time; in the latter case you're talking about cause and effect inside the mathematical object. Part of what is determined is that dropping the glass in the Now of 7:00:00am, causes the water to spill in the Now of 7:00:01am.
And as pointed out in Thou Art Physics, you are inside that mathematical object too. So are your thoughts, emotions, morals, goals, beliefs, and all else that goes into the way you determine your decisions.
To say "the future is already written" is a fine example of mixed-up timeful and timeless thinking. The future is. It is not "already". What is it that writes the future? In the timeless causal relations, we do. That is what is written: that our choices control the future.
But how can you "control" something without changing it?
"Change" is a word that makes sense within time, and only within time. One observes a macroscopically persistent object, like, say, a lamp, and compares its state at 7:00am to its state at 7:02am. If the two states are different, then we say that "the lamp" changed over time.
In Timeless Physics, I observed that, while things can change from one time to another, a single moment of time is never observed to change:
But if you have to change a single moment of time, in order to be said to "control" something, you really are hosed.
Forget this whole business of deterministic physics for a moment.
Let's say there was some way to change a single moment of time.
We would then need some kind of meta-time over which time could "change".
The lamp's state would need to change from "OFF at 7:02am at 3:00meta-am" to "ON at 7:02am at 3:01meta-am".
But wait! Have you ever seen a lamp change from OFF at 7:02am at 3:00meta-am, to ON at 7:02am at 3:00meta-am? No! A single instant of meta-time never changes, so you cannot change it, and you have no control.
Now we need meta-meta time.
So if we're going to keep our concepts of "cause" and "control" and "choose"—and to discard them would leave a heck of a lot observations unexplained—then we've got to figure out some way to define them within time, within that which is written, within the Block Universe, within... well... reality.
Control lets you change things from one time to another; you can turn on a lamp that was previously off. That's one kind of control, and a fine sort of control it is to have. But trying to pull this stunt on a single moment of time, is a type error.
If you isolate a subsystem of reality, like a rock rolling down hill, then you can mathematically define the future-in-isolation of that subsystem; you can take the subsystem in isolation, and compute what would happen to it if you did not act on it. In this case, what would happen is that the rock would reach the bottom of the hill. This future-in-isolation is not something that actually happens in the Block Universe; it is a computable property of the subsystem as it exists at some particular moment. If you reach in from outside the isolation, you can stop the rock from rolling. Now if you walk away, and again leave the system isolated, the future-in-isolation will be that the rock just stays there. But perhaps someone will reach in, and tip the rock over and start it rolling again. The hill is not really isolated—the universe is a continuous whole—but we can imagine what would happen if the hill were isolated. This is a "counterfactual", so called because they are not factual.
The future-in-isolation of a subsystem can change from one time to another, as the subsystem itself changes over time as the result of actions from outside. The future of the Grand System that includes everything, cannot change as the result of outside action.
People want to place themselves outside the System, see themselves separated from it by a Cartesian boundary. But even if free will could act outside physics to change the Block Universe, we would just have a Grand System that included free-will+physics and the future would be fully determined by that. If you have "freer will" we just have an Even Grander System, and so on.
It's hard to put yourself outside Reality. Whatever is, is real.
Control lets you determine single moments of time (though they do not change from one meta-time to another). You can change what would have happened, from one time to another. But you cannot change what does happen—just determine it. Control means that you are what writes the written future, according to the laws of causality as they exist within the writing.
Or maybe look at it this way: Pretend, for a moment, that naive views of free will were correct. The future "doesn't exist yet" and can be "changed". (Note: How are these two statements compatible?) Suppose that you exercise your "free will" at 6:30am to rescue three toddlers from a burning orphanage, changing their future from horrible flamey death at 7:00am, to happy gurgling contentment at 7:00am.
But now it is 7:30am, and I say:
"Aha! The past is fixed and can never be altered! So now you cannot ever have chosen any differently than you did choose. Furthermore, the actual outcome of your actions can never change either; the outcome is now fixed, so even if your past choice did now change, the past outcome wouldn't, because they are both just determined by "The Past". While your will was once free at 6:30am to change the future at 7:00am, it is now 7:30am and this freedom no longer exists. So now your will at 6:30am is no longer free. How can your past will have been free, now that there is only one past? Therefore I do not now assign you any moral credit for saving the orphanage; you no longer could have chosen differently from how you chose."
In the Block Universe, the "past" and the "future" are just perspectives, taken from some point within the Block. So, if the fixation of the past doesn't prevent the embedded decisions from having (had?) the property of freedom, why should the determination of the future prevent those embedded decisions from having the same property?
In the Block Universe, the Future is just like the Past: it contains the Nows of people making choices that determine their outcomes, which do not change from one meta-time to another.
And given the way we draw the causal arrows, it is correct to form the (un-observable) counterfactuals, "If I hadn't saved those children from the orphanage, they would be dead," and "If I don't think carefully, my thoughts will end up in Outer Mongolia." One is a counterfactual over the past, and one is a counterfactual over the future; but they are both as correct as a counter-factual can be.
The next step in analyzing the cognitive issues surrounding free will, is to take apart the word "could"—as in "I could have decided not to save the children from the orphanage." As always, I encourage the reader to try to get it in advance—this one is easier if you know a certain simple algorithm from Artificial Intelligence.
PPS: It all adds up to normality.
Part of The Quantum Physics Sequence
Next post: "The Failures of Eld Science"
Previous post: "Thou Art Physics"