Quantum Computing and Free will: The impossibility of Science without Free Choice

[LegionOnomaMoi]

I’ve often encountered the view here that “free will” is somehow impossible because “either everything is determined or it is random.” I’ve never understood this, but it was interesting to find that, whilst reading a popular book on quantum computing I’d started a while ago but never finished, this view was found to be fundamentally logically flawed:
“Now on to the misconception of the anti-free-will camp. I’ve often heard the argument that says that, not only is there no free will, but the very concept of free will is incoherent. Why? Because either our actions are determined by something, or else they’re not determined by anything, in which case they’re random. In neither case can we ascribe them to “free will.”
For me, the glaring fallacy in the argument lies in the implication Not Determined ⇒ Random.”
p. 291 of Quantum Computing Since Democritus

The author goes on to specify just how ridiculous it is to equate “randomness” with “not determined” via examples of non-random indeterminism and deterministic randomness, but these examples are not important here. I put down the book originally because I fundamentally disagree with much of what the author was stating. Better still, though, is the realization that free will is required not simply for quantum computing or even quantum physics, but (as we realized from theoretical and empirical findings within quantum physics) for the whole of the scientific endeavor. In particular, analyses of Bell’s theorem/inequality have shown that whatever the particular formulation of this inequality, one assumption required is that of freedom of choice, but this assumption underlies the entirety of science:

“The condition that the choice of the experiments is taken to be a free one means that the experimentalist must be thought to be able to choose them at will, without being unconsciously forced to one or the other choice by some hidden determinism. This condition has an important role in the proof of the theorem. It is often left implicit because of its apparent obviousness. Here it is explicitly stated. But let it be observed that, when all is said and done, it appears as constituting the very condition of the possibility of any empirical science.”
p. 64 of On Physics and Philosophy by physicist Bernard d’Espagnat

Is, then, free will not only compatible with science, but required by it?

 

[ratiocinator]

The author goes on to specify just how ridiculous it is to equate “randomness” with “not determined” via examples of non-random indeterminism and deterministic randomness, but these examples are not important here.

Actually, I think they are very important. Establishing this point is central to any claim of "free will" - as is a proper definition of what the proposed "free will" entails (how does it work?) and the connection between it and any examples of things that aren't determined or random (or a combination).

 

[LegionOnomaMoi]

Actually, I think they are very important.

How? You don't know what the examples are, so how can you consider them to be important?
I think you meant that examples of the type referred to are important. In particular, you seem to think that definitions of randomness and determinism are important in a definitional sense. Ok, let us say that "determined" here, in the context of "free will", means that it is impossible for one to act or choose in such a way as to allow for the possibility that one could have chosen differently. Said otherwise, free will entails the possibility that effects can (but need not always) be determined by free choice such that the agent in question could have chosen differently.

Establishing this point is central to any claim of "free will" - as is a proper definition of what the proposed "free will" entails (how does it work?)

How "free will" works is utterly irrelevant, insofar as it is nonsensical to demand a definition of "free will" and what is entailed whilst simultaneously claiming
1) the relevancy of any concepts, let alone randomness and determinism, to "free will" (if either of these are relevant, then there is some conception of free will that makes them so and therefore an implicit definition)
2) how it "works" somehow matters in a way other than the descriptional paradigm used in everything from so-called "special" sciences to fundamental physics. We do not ask "how" the action of a Lagrangian (or lagrangian density) works in order to force functional derivatives to be stationary, nor how the particular groups (e.g., SU(1), etc.) of the standard model are those required, nor even how gravitation exists as a force rather than as the dynamical evolution of local spacetime metrics in the presence of matter/mass/physical systems (or doesn't). We note that such things exist in some sense via empirical methods.
Freedom of choice is required for the entirety of empirical inquiry as there is no logically justified inferences from observations given that the experimental/observational designs/choices were not "free". Thus, we need only that there must exist the ability to choose that which one could have not chosen (some freedom of choice) as a minimal basis for a definition of free will. This alone renders the innane "determined or random" nonsense as such, and makes any claims that such a dichotomy exists contrary to the whole of the scientific endeavor.

 

[Satans_Serrated_Edge]

I’ve often encountered the view here that “free will” is somehow impossible because “either everything is determined or it is random.” I’ve never understood this, but it was interesting to find that, whilst reading a popular book on quantum computing I’d started a while ago but never finished, this view was found to be fundamentally logically flawed:
“Now on to the misconception of the anti-free-will camp. I’ve often heard the argument that says that, not only is there no free will, but the very concept of free will is incoherent. Why? Because either our actions are determined by something, or else they’re not determined by anything, in which case they’re random. In neither case can we ascribe them to “free will.”
For me, the glaring fallacy in the argument lies in the implication Not Determined ⇒ Random.”
p. 291 of Quantum Computing Since Democritus

The author goes on to specify just how ridiculous it is to equate “randomness” with “not determined” via examples of non-random indeterminism and deterministic randomness, but these examples are not important here. I put down the book originally because I fundamentally disagree with much of what the author was stating. Better still, though, is the realization that free will is required not simply for quantum computing or even quantum physics, but (as we realized from theoretical and empirical findings within quantum physics) for the whole of the scientific endeavor. In particular, analyses of Bell’s theorem/inequality have shown that whatever the particular formulation of this inequality, one assumption required is that of freedom of choice, but this assumption underlies the entirety of science:

“The condition that the choice of the experiments is taken to be a free one means that the experimentalist must be thought to be able to choose them at will, without being unconsciously forced to one or the other choice by some hidden determinism. This condition has an important role in the proof of the theorem. It is often left implicit because of its apparent obviousness. Here it is explicitly stated. But let it be observed that, when all is said and done, it appears as constituting the very condition of the possibility of any empirical science.”
p. 64 of On Physics and Philosophy by physicist Bernard d’Espagnat

Is, then, free will not only compatible with science, but required by it?

That isn't a problem of freewill I've heard. Pretty weak actually.

The real problem is that it's actually a pretty extraordinary claim. Freewill, broken down, is simply the claim that either;
A) everything has a cause(or multiple causes) right up until 'us' then BOOM <insert unquantified assertion here>

Or b) nothing has a cause at all, including our actions.

The extraordinary evidence it would take to support either assertion is still pending.

 

[ratiocinator]

How? You don't know what the examples are, so how can you consider them to be important?

They are important in that we need to establish that anything at all can be not determined, not random and not a combination.

I would like evidence that "the implication Not Determined ⇒ Random." is a "glaring fallacy". So, please provide one of these supposed counterexamples.

 

[ratiocinator]

How "free will" works is utterly irrelevant, insofar as it is nonsensical to demand a definition of "free will" and what is entailed whilst simultaneously claiming
1) the relevancy of any concepts, let alone randomness and determinism, to "free will" (if either of these are relevant, then there is some conception of free will that makes them so and therefore an implicit definition)
2) how it "works" somehow matters in a way other than the descriptional paradigm used in everything from so-called "special" sciences to fundamental physics.

You are arguing that something exists and yet you can't define what that something is?

1) If it isn't the result of determinism or randomness, then what is it the result of?
2) A description along the lines of your examples would be fine.

 

[ratiocinator]

“The condition that the choice of the experiments is taken to be a free one means that the experimentalist must be thought to be able to choose them at will, without being unconsciously forced to one or the other choice by some hidden determinism. This condition has an important role in the proof of the theorem. It is often left implicit because of its apparent obviousness. Here it is explicitly stated. But let it be observed that, when all is said and done, it appears as constituting the very condition of the possibility of any empirical science.”

I've tracked this quote down and looked at the immediate context (I haven't read the whole book).

There seems little justification for the last (highlighted) remark. In Bell's inequalities the requirement for a "free choice" is simply that it isn't correlated with the system being measured - random or even pseudo-random would do.

It is also interesting (as is noted before the quoted section) that violations of the inequalities could be explained by abandoning any of the premises, which include the free choice (the others being locality and 'realism' [hidden variables]).

 

[Skwim]

Boy! Talk about having a severe case of deja vu all over again. ......Oops! .....................That's right, it's about time for the biannual hunt for deterministic prey to begin. And the field appears ripe for picking.


Welcome back LOM.

 

[Polymath257]

I've never found a completely satisfactory definition of the concept of 'free will'. And different people seem to use different definitions. So, for example, in Daniel Dennett's book 'Freedom Evolves', there is a definition relying on, essentially, chaotic dynamics, but allowing for a deterministic system.

At base, it seems that 'free will' requires two things from physics:

1) That in some situations there is more than one 'possible' future.

2) By some activity of 'us', we can determine which of the possible futures is realized.

While the first seems, at least with quantum mechanics, to be quite likely, it is far from being clear how the second is to be shown.

In Dennett's book, chaotic dynamics is used to say there are several possible 'nearby' futures and that the 'choices' that determine these nearby futures happen inside our brains (and are thereby 'us'). So, even in a deterministic system, it is possible to make sense of the term 'free will'. Sort of.

In the case of Bell's inequalities and quantum experiments, the main question seems to be to what extent we can set up different experiments at all. So, if some sort of 'superdeterminism' is in play, then there is no issue. Alternatively, we can say that we *do* in fact, set up different experiments and can correlate the results. And that is quite sufficient for the development of science. Even in the absence of 'free will', we still have a variety of experiments being done and we can collect and interpret the data from these experiments.

 

[LegionOnomaMoi]

On the necessity of free choice for all empirical sciences (via assumptions concerning generalizing from "random samples" which wouldn't be random if everything was determined) and quantum physics more specifically:

The entirety of empirical science depends crucially on the assumption that experiments involve the exercise of free will. I cannot determine anything about anything if I cannot even e.g., argue that my choice to pick a particular sample of gold, or use a particular kind of atom, or particular cells, etc., can be generalized to the populations my samples were obtained from. The whole of empirical science rests upon assumptions regarding the nature of sampling from a population such that the sample was sufficiently random to justify the inference that the sample is indeed that: a representative of a particular population (of some element, chemical, virus, animal, genetic sequence, etc.). If all choices are determined or purely random, the entirety of the underlying logic of all empirical inquiry vanishes in a puff of smoke. We are no longer justified in inferring anything about some experiment beyond that we found what we must have found because we could not have done differently, and thus what might be said of the properties of e.g., photons or atoms or whatever that we investigated were anything other than specific to our particular investigation. Inference is no longer logically valid, because we can no longer infer from the specific to the general.

 

[ratiocinator]

The whole of empirical science rests upon assumptions regarding the nature of sampling from a population such that the sample was sufficiently random to justify the inference that the sample is indeed that: a representative of a particular population (of some element, chemical, virus, animal, genetic sequence, etc.). If all choices are determined or purely random, the entirety of the underlying logic of all empirical inquiry vanishes in a puff of smoke.

You contradicted yourself and the video.

There is no justification in the video for excluding purely random choice, in fact purely random not only fits perfectly with the definition provided: not correlated with anything outside of its future light cone - but is the only thing that does. If it is determined by anything in its past light cone, even probabilistically, then it doesn't satisfy the definition.

However, as was pointed out at the end of the video, the requirement doesn't need to be quite that strong and the choice doesn't need to be entirely random. He doesn't explain how totally determined choices would not work if they where, for all practical purposes, uncorrelated with the experiment.

Here is a short paper that summarizes what the video explains about "free choice": A short note on the concept of free choice (pdf)

Here is a paper with a completely different approach: Quantum Mechanics from Classical Logic (pdf)

These matters are very far from settled.

 

[LegionOnomaMoi]

You contradicted yourself and the video.

There is no justification in the video for excluding purely random choice

This is a contradiction-in-terms, or it is to negate the entire framework and logical basis for empirical science. If my choices as a scientist in some experiment are purely random, then either they aren't choices (how can I "choose" something wholly random?) or they negate any possible inference from any possible results of any possible experiment that could possibility exist other than that, given some experimental design, the outcome was determined independently of my experimental choices and thus I can't generalize any results.

in fact purely random not only fits perfectly with the definition provided: not correlated with anything outside of its future light cone - but is the only thing that does.

The problem is that this "not correlated with anything outside of its future light cone" is fundamentally at all odds with the entirety of all modern quantum physics and this correlation has been experimentally confirmed over and over and over again since at least Aspect et al 1982 (Gisin and others have confirmed correlations outside of light-cones at distances of many miles/kilometers since the 90s).

If it is determined by anything in its past light cone, even probabilistically, then it doesn't satisfy the definition.

It is determined by something outside its lightcone. Hence nonlocality.

However, as was pointed out at the end of the video, the requirement doesn't need to be quite that strong and the choice doesn't need to be entirely random.

Wrong. You've missed the point. It is true that Bell's inequality can be violated in more creative ways, but this only strengthens the idea that it is pure nonsense to suppose there is any support for the idea that we are unable to choose other than we have. When we no longer demand that particular loopholes/assumptions hold or are violated in Bell inequality tests, we do not, in general, allow that stronger assumptions fail. The opposite, in fact.

He doesn't explain how totally determined choices would not work if they where, for all practical purposes, uncorrelated with the experiment.

Totally determined choices utterly fail. That's the point. Free choice is required here at a recognizably empirical level, whereas they enter into the entirety of the scientific enterprise at a more abstract level.

Here is a short paper that summarizes what the video explains about "free choice": A short note on the concept of free choice (pdf)

Here is a paper with a completely different approach: Quantum Mechanics from Classical Logic (pdf)

These matters are very far from settled.

That the sciences depend upon free choice is not even a quantum theoretical issue: its foundational to the entirety of the scientific endeavor. It just so happens that this fact is forced upon us by quantum theory, whereas we might have ignored the crucial assumptions of free choice otherwise.

 

[LegionOnomaMoi]

I've tracked this quote down and looked at the immediate context (I haven't read the whole book).

Then see his Veiled Reality or any number of treatises, monographs, or volumes on Bell's inequality.

There seems little justification for the last (highlighted) remark.

Just physics.

In Bell's inequalities the requirement for a "free choice" is simply that it isn't correlated with the system being measured - random or even pseudo-random would do.

This is utterly, horrifically, and completely wrong. "free choice" is the basis for any and all inferences from any experiment in which any assumptions might be made in which genralizations from outcomes from experiments are made beyond the specific results. We cannot even so much as determine that there exist properties of any class of concepts/objects (electrons, nouns, elements, diseases, cells, etc.) without assuming that empirical tests of these grant "free choice". This is the most basic, fundamental component of all empirical inference.

It is also interesting (as is noted before the quoted section) that violations of the inequalities could be explained by abandoning any of the premises, which include the free choice (the others being locality and 'realism' [hidden variables]).

That's sort of the point. One can abandon the assumption of free choice here, but one would then lose the entirety of the logical justification for the whole of scientific inquiry.

 

[ratiocinator]

This is a contradiction-in-terms, or it is to negate the entire framework and logical basis for empirical science. If my choices as a scientist in some experiment are purely random, then either they aren't choices (how can I "choose" something wholly random?) or they negate any possible inference from any possible results of any possible experiment that could possibility exist other than that, given some experimental design, the outcome was determined independently of my experimental choices and thus I can't generalize any results.

Have you watched the video? He actually mentions linking the measurements to random number generators. Truly random choices about measurements meet all the requirements he set out as "free choices".

The last part of that doesn't really make sense. Randomnly choosing the measurement angles in Bell type experiments makes perfect sense and the other examples given of making sure you choose a representative sample of some population - the more random the better.

The problem is that this "not correlated with anything outside of its future light cone" is fundamentally at all odds with the entirety of all modern quantum physics and this correlation has been experimentally confirmed over and over and over again since at least Aspect et al 1982 (Gisin and others have confirmed correlations outside of light-cones at distances of many miles/kilometers since the 90s).

The point of Bells type experiments is to detect such correlations over space-like separations but the criterion for the 'free choice' of measurement was that it wasn't correlated with anything outside of its future light cone - that came from Bell and modified by the speaker - it's no my idea.

I really think you need to watch it again. I have ploughed through the video and look at papers and parts of books an none of them seem to say that the whole of science comes to an end if we don't have "free will" - something you still haven't defined.

 

[ratiocinator]

"free choice" is the basis for any and all inferences from any experiment in which any assumptions might be made in which genralizations from outcomes from experiments are made beyond the specific results. We cannot even so much as determine that there exist properties of any class of concepts/objects (electrons, nouns, elements, diseases, cells, etc.) without assuming that empirical tests of these grant "free choice". This is the most basic, fundamental component of all empirical inference.

There is a specific definition supplied by Roger Colbeck and Renato Renner about what "free choice" means in this context, and it's the one about being uncorrelated with anything outside of its future light cone - the one you rejected.

See: A short note on the concept of free choice (pdf) - it's only one page (without the references).

A purely random choice would fit the bill but I can see nothing in the arguments to say that it needs to be uncorrelated with items (outside of its future light cone) that have no connection with the experiment. Why would it matter if it were correlated with what the experimenter had for breakfast or whether she's having a good day..?

 

[LegionOnomaMoi]

Have you watched the video?

I was there. I spoke to him and to others (esp. Marissa Giustina, as I knew already from debates with Khrennikov that Zeilinger's post-docs were working on so-called "foundational" question in QM). I'm quite familiar with the video and, moreover, the issues related to it.

He actually mentions linking the measurements to random number generators. Truly random choices about measurements meet all the requirements he set out as "free choices".

You mistake fundamentals here. The "free choice" isn't related to the randomness that can be generated by quantum processes or dynamics related to physical systems governed by randomness. This is trivial. It is a fact, as the lecture covers in a manner I would have thought obvious even to the uninformed, that without freedom of choice there is no basis to any inference that generalizes from experimental results. Thus, the ability to choice that which we could have chosen differently underlies the entirety of empirical science.

Randomnly choosing the measurement angles in Bell type experiments makes perfect sense and the other examples given of making sure you choose a representative sample of some population - the more random the better.

In order to choose randomly, one must be capable of choice. In particular, one must assume that one's choice could have been differently than that choice. In other words, in order to assert that one has made any random choices in the design and implementation of some experiment, one has to assert that one could have chosen differently.

The point of Bells type experiments is to detect such correlations over space-like separations

Wrong.

but the criterion for the 'free choice' of measurement was that it wasn't correlated with anything outside of its future light cone

Spectacularly wrong.
Bell didn't propose any experiments. He used Bohm's reformulation of EPR to design a condition in which local realism would necessarily conflict with quantum mechanical predictions. This inequality was empirically realized in Aspect et al.'s 1982 experiment and has been repeatedly realized over distances of many miles/kilometers since. Free choice was a way out of the violation of causality as it was assumed (as in empirical science in general) in any and all possible experiments with any and all possible systems. The criterion of free choice was simply the criterion of the ability of scientists to do science: that is, the ability to assume that the particulars of an experiment could be generalized.

I really think you need to watch it again.

What exactly is your background?

 

[viole]

I’ve often encountered the view here that “free will” is somehow impossible because “either everything is determined or it is random.” I’ve never understood this, but it was interesting to find that, whilst reading a popular book on quantum computing I’d started a while ago but never finished, this view was found to be fundamentally logically flawed:
“Now on to the misconception of the anti-free-will camp. I’ve often heard the argument that says that, not only is there no free will, but the very concept of free will is incoherent. Why? Because either our actions are determined by something, or else they’re not determined by anything, in which case they’re random. In neither case can we ascribe them to “free will.”
For me, the glaring fallacy in the argument lies in the implication Not Determined ⇒ Random.”
p. 291 of Quantum Computing Since Democritus

The author goes on to specify just how ridiculous it is to equate “randomness” with “not determined” via examples of non-random indeterminism and deterministic randomness, but these examples are not important here. I put down the book originally because I fundamentally disagree with much of what the author was stating. Better still, though, is the realization that free will is required not simply for quantum computing or even quantum physics, but (as we realized from theoretical and empirical findings within quantum physics) for the whole of the scientific endeavor. In particular, analyses of Bell’s theorem/inequality have shown that whatever the particular formulation of this inequality, one assumption required is that of freedom of choice, but this assumption underlies the entirety of science:

“The condition that the choice of the experiments is taken to be a free one means that the experimentalist must be thought to be able to choose them at will, without being unconsciously forced to one or the other choice by some hidden determinism. This condition has an important role in the proof of the theorem. It is often left implicit because of its apparent obviousness. Here it is explicitly stated. But let it be observed that, when all is said and done, it appears as constituting the very condition of the possibility of any empirical science.”
p. 64 of On Physics and Philosophy by physicist Bernard d’Espagnat

Is, then, free will not only compatible with science, but required by it?

I have a question for you.

Suppose that out of my free will I decide to kick a ball.

Is the new state of Universe (including ball there instead of here) inferrable/not inferrable from the state of the Universe prior to my decision? Say, one billion years ago?

In other words: did I introduce new information into the system? Or, by reversing the direction of time, did I make it possible to have two states (ball kicked/ball not kicked) merging into one?

I am not a physicist, so I am not sure. I am mostly interested because I am a quite fundamentalist atemporalist.

Ciao

- viole

 

[Polymath257]

This is a contradiction-in-terms, or it is to negate the entire framework and logical basis for empirical science. If my choices as a scientist in some experiment are purely random, then either they aren't choices (how can I "choose" something wholly random?) or they negate any possible inference from any possible results of any possible experiment that could possibility exist other than that, given some experimental design, the outcome was determined independently of my experimental choices and thus I can't generalize any results.

How does it follow that you cannot generalize? Even if the choice of experiment was pre-determined, it still seems that you can make valid deductions from the data and correlations to derive hypotheses that are tested via future (even pre-determined) experiments.

The problem is that this "not correlated with anything outside of its future light cone" is fundamentally at all odds with the entirety of all modern quantum physics and this correlation has been experimentally confirmed over and over and over again since at least Aspect et al 1982 (Gisin and others have confirmed correlations outside of light-cones at distances of many miles/kilometers since the 90s).

It is determined by something outside its lightcone. Hence nonlocality.

The correlation was formed when the particles became entangled. Both sides of the experiment are within the forward light cone of that event. And neither side was 'determined' by the other side. The correlation was fixed, but in a valid causal way.

The point is that quantum mechanics *is* a local theory. It just isn't a 'realist' theory. The probabilities and correlations are determined, not the specific outcomes.

 

[Polymath257]

The entirety of empirical science depends crucially on the assumption that experiments involve the exercise of free will. I cannot determine anything about anything if I cannot even e.g., argue that my choice to pick a particular sample of gold, or use a particular kind of atom, or particular cells, etc., can be generalized to the populations my samples were obtained from. The whole of empirical science rests upon assumptions regarding the nature of sampling from a population such that the sample was sufficiently random to justify the inference that the sample is indeed that: a representative of a particular population (of some element, chemical, virus, animal, genetic sequence, etc.). If all choices are determined or purely random, the entirety of the underlying logic of all empirical inquiry vanishes in a puff of smoke. We are no longer justified in inferring anything about some experiment beyond that we found what we must have found because we could not have done differently, and thus what might be said of the properties of e.g., photons or atoms or whatever that we investigated were anything other than specific to our particular investigation. Inference is no longer logically valid, because we can no longer infer from the specific to the general.

It seems that superdeterminism is only an issue if the experiments performed are superdetermined to be non-representative of the reality. So, in the situation where you have an element with isotopes, the issue only appears if only if we are pre-determined to select isotopes that are not representative of the sample. But, if this is *always* the case, throughout every possible measurement, then no possible observation will be of anything other than the 'chosen' isotope, and our derived theory will be predictive of observations. Since that is the way scientific theories are challenged, we will have a scientifically valid result.

The basic question boils down to whether it is a reasonable concern that something will exist that we can never detect because of superdeterminism. But I would say that even if such a thing exists, it is irrelevant to us. Hence the basic logic of the science doesn't break down.

 

[Polymath257]

I have a question for you.

Suppose that out of my free will I decide to kick a ball.

Is the new state of Universe (including ball there instead of here) inferrable/not inferrable from the state of the Universe prior to my decision? Say, one billion years ago?

In other words: did I introduce new information into the system? Or, by reversing the direction of time, did I make it possible to have two states (ball kicked/ball not kicked) merging into one?

I am not a physicist, so I am not sure. I am mostly interested because I am a quite fundamentalist atemporalist.

Ciao

- viole

At this level, nobody knows. If some version of superdeterminism is true, then yes, everything would be inferable from the previous state of the universe and no, you would not have introduced new information (although some care in concepts is required here).

Much more likely is that the state of the universe 1 billion years ago does NOT determine the current state to such precision because of quantum indeterminacy. But it is still quite possible that the new state would have been derivable from the state a day before. One complication is that a brain is an exceedingly complex system where very small changes in conditions at one time can lead to large differences later. So, a small change yesterday may well lead to a different result today.