Hacker News Reader

Comment by devishard | Hacker News Reader

The free will conversations drive me insane. People can't shut up about how Heisenberg's uncertainty principle proves random phenomena (which isn't exactly true) but worse, they insist it indicates that we have free will.

But a) what it really indicates is that it would be unmeasurable whether we have free will, and b) it also doesn't really prove that, because while electron positions are unpredictable, molecular reactions, the sort that happen in the human brain are pretty predictable.

Basically, my response to this is "That would be relevant if we had black holes or large hadron colliders in our heads, but we don't, so...". But people really really want to believe in free will and will use any sliver of half-truth to believe they've proved it.

EDIT: To be clear, I do not believe free will has been proven OR disproven.

EDIT 2: More sophisticated wrong people argue from Bell's Inequalities rather than the Heisenberg Uncertainty Principle, but that also results from a misunderstanding of the theory and also neither proves nor disproves determinism.

View on HN

> [W]hile electron positions are unpredictable, molecular reactions, the sort that happen in the human brain are pretty predictable. Basically, my response to this is "That would be relevant if we had black holes or large hadron colliders in our heads, but we don't, so...".

What about those quantum effects that have recently been shown to potentially be part of brain activity:

https://www.sciencedaily.com/releases/2014/01/140116085105.h...

You could make the same argument to say that your PC adding 2+2 is indeterminate because it has the same underlay of supposed random quantum behavior. But at our material level, the world is pretty deterministic just as your PC or phone adding 2+2 is. And as far as neuroscience has gathered, quantum randomness hasn't affected it enough for those scientists to think it's non-deterministic, or else they wouldn't have a job.

Quantum effects are part of brain activity because they underly all electrochemistry--that's never been in question. The question is whether those quantum effects impart any non-determinism to the brain system, and the link you posted doesn't make any claims that they do or don't.

Given what we know about brains so far, doesn't it need to be proven, rather than disproven? Surely the burden of proof here is on people claiming a non-biological and non-predictable basis for brain function?

I don't see how subjective experience, i.e. consciousness, can come about from purely material interactions as currently understood. That means that a Turing Machine would be able to produce them, which means that a bunch of monks working with paper and pencil for thousands of years would be able to produce them. There's something fundamental that we don't understand and the scientific community is very opposed to talking about it, dismissing any and all speculation as quackery. This taboo is holding back serious inquiry into the matter.

    > I don't see how

Isn't that just The God of The Gaps? I suspect that many intelligent people 500 years ago wouldn't have been able to "see how" a 500 ton metal object could take controlled flight.

    > That means that a Turing Machine would be able to
    > produce them

Is there any specific reason to think it wouldn't?

    > the scientific community is very opposed to talking
    > about it

It sounds like you're suggesting there's a super-natural force at play, in which case, the scientific community is entirely justified in being opposed to talking about it.

I'm not suggesting anything supernatural. I'm suggesting that there's possibly physical phenomenon that do not fit our current materialist model.

Further, my big problem is with people trying to ignore the problem completely by claiming that there is no Hard Problem or even suggesting that consciousness is not real. It's a total head-in-the-sand position to take but it's the dominant stance.

Who is saying consciousness is not real? It just not non-physical and has no single set of hard properties from which to form a definitive boundary for discussion.

I'm curious at what point you think agency becomes consciousness. Is it self-awareness? Is a dog conscious? A mouse? A snail? An ant? A bacteria? A tree?

It has nothing to do with agency. I'm talking about the phenomenon of subjective experience. The difference between a computer that analyzes camera input and a being that actually sees the world. It would be the difference between you and a philosophical zombie.

If you aren't that "artificial" intelligence and can't put yourself in that subjective experience, how would you know that the AI's subjective experience differs from yours? Because you're relying on material observance? You realize your conundrum now? You have to rely on the material to gauge the material yet you try to invalidate the material experience using material observations.

> The difference between a computer that analyzes camera input and a being that actually sees the world.

How do you know there is a qualitative difference?

> That means that a Turing Machine would be able to produce them, which means that a bunch of monks working with paper and pencil for thousands of years would be able to produce them.

Yes. What's not to see? I think appealing to something other than Turing machines shows a lack of imagination; it takes a lot of thought to even scratch the surface of how profound Turing machines are.

Are you really arguing that a scribbling numbers on a piece of paper could give rise to a conscious being?

In another comment, you said:

> I'm not suggesting anything supernatural. I'm suggesting that there's possibly physical phenomenon that do not fit our current materialist model.

Absolutely. Sure, there are of course physical phenomenon that do not fit our current materialist model.

Let's say we discover those additional phenomenon. Ultimately, we'd want to build a machine to model that phenomenon. Perhaps encode a model of that phenomenon in software. Once that happens, scribbling numbers on a piece of paper indeed could give rise to a conscious being.

If it's some system that can't be covered by scientific understanding, isn't that what supernatural means? It's either some internally consistent (but possibly very complicated) set of rules, or it's supernatural.

No, it could be that the universe actually cannot be modeled computationally, that it exceeds the constraints of a Turing machine. This is an open question.

I think you're giving a good example of how poorly people visualize and understand infinity. A Turing Machine's capabilities are based on its infinite memory and infinite computation time, yet you imagine it as "a bunch of monks scribbling" and immediately dismiss its capabilities.

I would actually restrict things further, since introducing infinity can lead to unintuitive results orthogonal to computability ;)

Rather than 'possessing' an infinite tape and infinite time, I prefer to think of turing machines as always being allowed to perform one more step, and always being allowed to use one more tape cell. I find this characterisation of time quite intuitive as-is, whilst the tape can be made intuitive by imagining tape factories on each end, which produce more cells whenever the read/write head gets close.

In fact, such "tape factories" have been used to build universal turing machines in the game of life http://rendell-attic.org/gol/fullutm/index.htm

An order of non-celibate monks can also work for infinite amount of time and hold infinite data. Anyhow, it's not really relevant to my point. An infinite number of scribbles doesn't seem any more likely to give rise to consciousness than any finite number.

> Are you really arguing that a scribbling numbers on a piece of paper could give rise to a conscious being?

Yes. I find that no less plausible than a warm sack of meat giving rise to a conscious being.

I agree, which is why I suspect that consciousness is the result of some fundamental non-computable phenomenon that our minds tap into to perform some useful function.

No. The topic has simply been ceded to a subset of philosophers who fill large volumes that still don't manage to say more about this topic than you do. One big problem is that your thought experiments are designed specifically to make the null hypothesis sound silly. That's the same strategy you guys have been uding for thousands of years. Not so long ago people claimed that life couldn't be explained without some special substance.

You're aware of John Searle's Chinese Room gag?

> I don't see how subjective experience, i.e. consciousness, can come about from purely material interactions as currently understood. That means that a Turing Machine would be able to produce them, which means that a bunch of monks working with paper and pencil for thousands of years would be able to produce them.

Not without input data, they couldn't. You can produce a model, but unless you collect data to validate your model against, you're just speculating in a void.

> There's something fundamental that we don't understand and the scientific community is very opposed to talking about it, dismissing any and all speculation as quackery.

We're talking about it right now, and I haven't heard anyone opposed to talking about it.

However, I'll fully dismiss opinions as quackery if they clearly misrepresent the facts, i.e. claims that Heisenberg's Uncertainty Principle or Bell's Inequalities prove free will are quackery. This isn't because the ideas are taboo or because I'm opposed to talking about them, it's because they're misrepresentations or misunderstandings of facts which are clearly established.

In my experience, when people claim that scientists are opposed to talking about their hypothesis, it's because their hypothesis has been thoroughly and concretely disproven.

Even if we don't have free will, it's better to pretend we do. There is a difference between a driver that falls asleep at the wheel and causes an accident and a driver that flies into a rage and crashes their car into another driver. It's hard to justify the distinction if neither driver has a choice. I'd argue it's better for society to maintain that distinction.

Perhaps someone can reframe that in terms of types of error, and people prone to more serious types of error (road rage) face different consequences. But that gets real complicated real quick.

I dunno. I suspect the majority of the time i'm on autopilot. Alarm goes off, get up, brush teeth, shower. I pretend i'm capable of not doing that. i just don't choose the other way.

There are a lot of people who will not commit crime only because they fear the consequences. So it would be rational to punish criminals regardless of your thoughts on free-will. And if you don't believe in free-will then you have to accept that you have no "free" choice about punishing criminals.

Myself, I do not even know what the will is, or what it could possibly be free from. The influence of the universe?

> There are a lot of people who will not commit crime only because they fear the consequences.

Fear sounds like an implicit assertion of choice, but i think you mean something more subtle.

Let my try to put it another way. People are complex systems that can be influenced. Water tends to flow downhill, we can set up troughs to control the flow of water. The flow is complex, and some water will splash out, but generally the water does follows the system. Likewise, we can set up legal troughs to guide people to the behavior we want. (I know argument by analogy is crappy, i'm just hoping we can agree on this point, and i think we do.) There's a whole other argument about why we would set up those legal rules, but i'd like to set that aside for a moment (if it's a big deal, or fundamental in some way, i'd like to hear your side. I'm not sure i can meaningfully engage that point)

It seems like, punishment would need to be interpreted purely as societal cost. So you tally up the wreckage from the car accident example above, say $25,000 and 2 lives.

I don't really see how we can have both manslaughter and murder, without free will. What extra thing did the angry guy do? He didn't make a choice, because (for the sake of argument) he has no free will. What is the characteristic that makes what he did worse? The cost to society is the same.

> Myself, I do not even know what the will is, or what it could possibly be free from. The influence of the universe?

My personal interpretation (which is probably wrong, but something i'm willing to go back and forth on to come to consensus) Free will the mind having a choice in taking action, rather than the body just doing whatever, and the mind rationalizing it after the fact. Perhaps every action we take is just like a heartbeat. I have no control over that. Perhaps when i hold my breath, i'm not actually choosing to hold my breath, but the complex system that is my body holds it's breath and my mind is claims that it was responsible for that holding of breath. Or typing this comment, or whatever. I don't think there's any way i can know. It feels like my mind is in control, but my mind gets all of its information from (and the software runs on!) my body. my mind seems to control the hardware, but maybe i'm just rationalizing after the fact.

> I don't really see how we can have both manslaughter and murder, without free will. What extra thing did the angry guy do? He didn't make a choice, because (for the sake of argument) he has no free will. What is the characteristic that makes what he did worse? The cost to society is the same.

That doesn't require the concept of free will. Because for manslaughter and murder the reasons are different for the death. One is deliberate, one isn't.

> It seems like, punishment would need to be interpreted purely as societal cost. So you tally up the wreckage from the car accident example above, say $25,000 and 2 lives.

You would also want to calculate predicted future costs from that person, mainly from the point of view of harm to other people. If someone like them is likely to kill on average one or more people in the rest of their life then they aren't going to be allowed to interact with general society again. This doesn't need the concept of free will to implement. It only needs a sufficiently sophisticated way of determining of what harm a person like them will likely cause. This is in addition to the harm they have already caused, so there is no reason for someone to become a one time killer. And yes, getting the predictions right may be really tricky, and the way the courts currently sentence people may be a good as or better than trying to predict what people do on average. But the way courts sentence at the moment doesn't require the concept of free will anyway, because people do what they do because of reasons, human reasons, and they can differentiate there.

> It seems like, punishment would need to be interpreted purely as societal cost. So you tally up the wreckage from the car accident example above, say $25,000 and 2 lives.

I don't think so. I think determinism makes "punishment" as a concept, completely nonsensical--you can't punish someone for something they didn't choose to do. I think determinism pushes the focus to rehabilitation rather than punishment, which for many types of crimes (i.e. drug use, computer hacking, nonviolent robbery) is clearly more effective in preventing recidivism (and I'd argue that for crimes where it's not effective, that we haven't really explored rehabilititation because of the widespread belief in free will).

Agreed. But does this highlight a problem with a deterministic world view, or does it highlight a problem with the distinction between manslaughter and murder?

> But does this highlight a problem with a deterministic world view, or does it highlight a problem with the distinction between manslaughter and murder?

That is a really interesting point. I'm not convinced, but you've given me a ton to think about.

    > There are a lot of people
    > who will not commit crime
    > only because they fear the
    > consequences

Really? I'd say both crime rate and severity of consequences have taken a plunge over the last few centuries.

I used to feel this way too - I remember writing up drunken notes when I was at college saying exactly the same thing.

Punishments from crime have decreased substantially - nobody gets transported or hung for sheep theft anymore. The world has never been a safer place. States with the death penalty are no safer than those without. Fewer people genuinely believe in a Christian (or whatever) hell than in years past. At the same time, religion has faltered as a source in many people's lives.

I would argue this means that circumstances that lead people in to crime have decreased rather than a suggestion that people are making choices to commit crime but don't based on deterrence. I appreciate there exist counter arguments to this too, but also feel that if we treated crime as a function of situation rather than morality, we'd find ways to decrease it more cheaply.

I guess i'd ask the same question i asked of jeremyjh.

With no notion of free will, how do we distinguish between manslaughter and murder? When someone dies, the cost to society is the same either way. A consequence of no free will seems to be it's a distinction without difference. I disagree, but it might just be deeply embedded beliefs that are wrong.

> With no notion of free will, how do we distinguish between manslaughter and murder?

I'd argue that the distinction is not indicative of a phenomenon that exists.

I think a much better model would be predictive and preventive rather than punitive--the level assignment to rehabilitation should be based on the probability and severity of recidivism, rather than the level of punishment being based on the severity of the crime.

> Even if we don't have free will, it's better to pretend we do. There is a difference between a driver that falls asleep at the wheel and causes an accident and a driver that flies into a rage and crashes their car into another driver. It's hard to justify the distinction if neither driver has a choice. I'd argue it's better for society to maintain that distinction.

The distinction still matters without free will, because it's still a predictor of future behavior. A driver who crashes because of a preventable cause shouldn't be allowed to drive until the problem is resolved. If the problem is falling asleep, the solution may be simple (get more sleep?) or hard (narcolepsy), while anger issues are probably pretty difficult to provably solve.

Belief in free will has some seriously harmful effects on the way we deal with these situations. Most notably, it's the basis for a focus on punishment rather than rehabilitation. Deterrents don't work, largely because they don't fix the causes for bad behavior. Rehabilitation is the only sensible approach if you don't believe in free will, whereas punishment makes a bit more sense if you believe that people have free will.

> Perhaps someone can reframe that in terms of types of error, and people prone to more serious types of error (road rage) face different consequences. But that gets real complicated real quick.

This is fundamentally a free-will-based way of seeing the problem. "Consequences" are completely irrelevant in a deterministic worldview unless you can prove that they act as a deterrent (that is, that they are part of the cause). Since consequences are provably not significant deterrents for many crimes, I don't think that works.

Your argument basically that "this idea of punishment/consequences (which only makes sense in a free-will-based ideology) doesn't work in a deterministic ideology, and punishment/consequences are good in my free-will-based ideology, therefore deterministic ideologies are wrong". But you have to hold a free-will-based ideology for the preconditions of your argument to make sense.

Yes, i see that now. The distinction depends on free will, so isn't a good test.

Very insightful. Thank you.

Your talk of punishment vs rehabilitation is spot on, and very relevant to the "Doesn't the whole justice system collapse if there is no free will?" question. Have you ever heard of how some South American prisons are treating inmates with Ayahuasca, in order to rehabilitate their mind? Very interesting stuff considering the success rate.

You cannot derive is from ought.

That's the argument I use for not believing in assertions of existence ("God exists", "Pink unicorns exist"), because people basically don't think things exist that can't be proven to exist, but make a special pleading for Gods.

I guess you could argue that we generally assume a deterministic model exists and try to find it, but I hadn't considered that before.

Free will is the sensation of being in a brain that models counterfactuals. It is also a token concept that people like to use to do their "worship the mystery" so they can secretly believe that scientists are stupid and thus not have to be held responsible for their failure to understand them or their work.

Free will is akin to an emotion. Just because physicists haven't explained it in excruciating detail doesn't mean it doesn't exist, or we'd all be running around debating the existence of fear and love and curiosity.

> Free will is the sensation of being in a brain that models counterfactuals

No. Free will is the ability to make a choice not completely dependent on history.

I don't consider that free will then, that would just be "not completely" free will. If it's not dependent on a history, what would be the directive that effects the choice? To me, the notion of free will just defeats the ubiquitous cause-effect nature in everyday life and observance. What would be the nature of the effector that makes the "free" choice then?

That depends on the person defining it. Why can't it be the outcome my brain has determined for me to take which I subjectively understand as a self-chosen action? That may well depend on my history--my emotions, memories and all the things that shape me--but those are more reasons than constraints.

Isn't that like saying I want to fly, so I don't have free will if I can't?

Would flipping a coin count?

If you respond, "No, because the particles that make up the coin and the muscles that drive my thumbnail have a history that determines the outcome," then how about a slight variation? "I can't decide what to do. I have no idea what time it is, so if the minute digit on my clock is odd, I'll do X, otherwise I'll do Y."

Here I've explicitly refrained from using either a source of randomness or anything else with its own history to determine my fate. The decision to consult the clock might have been an act of free will on my part, but the outcome certainly wasn't.

I don't see why anyone should accept a useless definition as a correct one.

Nobody would argue that they had free will if they didn't feel like they had free will. Nobody is positing baseless metaphysical nonsense without some reverence in their actual experiences. If you want to usefully discuss free will, you have to account for the sensation of free will, not just some abstract model that you've merely assembled grammatical statements from.

We all understand fear and love and curiosity well enough that we can make rudimentary predictions about them. If I point a gun at someone they will experience fear and respond with fight or flight. If I spend a lot of time with someone and certain compatibility conditions are met, we will experience love and respond with caring behaviors. If I show someone something that they don't understand fully but that has properties that interest them, they'll likely experience curiosity and respond with questions and exploration.

Someone who studies these things (a psychologist or neurologist) could make even more sophisticated predictions.

We may not have a complete low-level electrochemical model of these phenomenon, but we have a good enough model to make some predictions. But if anything, one of the most salient criticisms of free will is that it doesn't allow us to model and make predictions--in fact, it eschews predictions altogether.

Right, I would clarify though, "single electron positions are unpredictable exactly[0]" but, I know that there are more electrons on this side of the cap than the other side, and it's physics at that scale (S >> \hbar) that matters.

[0] To be more pedantic than enlightening, add before wavefunction collapse.

Exactly.

On the flip side of this, superdeterminism (that we have absolutely no free will at all) is a third way of explaining Bell inequality violations (apart from Copenhagen and hidden variables), but essentially everyone in the physics community refuses to consider superdeterminism seriously.

It's not that they don't take it seriously -- it's that it's boring and says nothing useful. Yes, everything could be a giant conspiracy that makes it appear as if the standard ways of approaching quantum mechanics are simple, nice, and right, but ...

That's not what superdeterminism would indicate at all.

"Conspiracy" indicates free will on a large number of actors.

And a superdeterministic system could exist which differs from standard ways of approaching quantum mechanics. One obvious way in which it could exist is if our means of gathering data remain as limited as they are currently--i.e. we don't ever find a way around Heisenberg's Uncertainty Principle and therefore can't ever model quanta deterministically. Superdeterminism doesn't mean that we'd always be able to measure the deterministic phenomena.

Many people believe that free will and determinism are not incompatible (compatibilists).

Personally, I'm not even sure introducing non-determinism matters much. What would it even mean? That beyond determinsitic actions defined by our precise history, we get a bit of randomness permuting the whole thing? Does that really make it any freer?

> Many people believe that free will and determinism are not incompatible (compatibilists).

Could you explain how that works? I'm unfamiliar with that, but I suspect that they may be using a different definition of "free will" than I am familiar with.

> Personally, I'm not even sure introducing non-determinism matters much. What would it even mean? That beyond determinsitic actions defined by our precise history, we get a bit of randomness permuting the whole thing? Does that really make it any freer?

I think free will and randomness are both forms of non-determinism. Non-determinism doesn't prove free will, but it allows for the possibility that it might exist (and post people intuit that randomness could be indistinguishable from non-determinism).

To me, free will is the lack of external forces controlling my decisions and actions. That is, I can decide things for myself, even if those decisions are a deterministic product of my current state.

How are you defining it such that it must be non-deterministic and distinct from randomness? I don't really understand what that means. Assuming there's a single shared reality, what's the source of the non-determinism? Even if you say that's the free will itself, you'd be asserting that our "wills" are somehow capable of randomness.

> To me, free will is the lack of external forces controlling my decisions and actions. That is, I can decide things for myself, even if those decisions are a deterministic product of my current state.

Without some place in the sequence of events that isn't deterministic, I don't see how you can separate your current state from external forces. That is, if your decisions are a deterministic product of your current state and your current state is a deterministic product of other states before it (including states that existed before you did, such as your parents) then I don't see that as "the lack of external forces controlling my decisions and actions".

> How are you defining it such that it must be non-deterministic and distinct from randomness? I don't really understand what that means. Assuming there's a single shared reality, what's the source of the non-determinism? Even if you say that's the free will itself, you'd be asserting that our "wills" are somehow capable of randomness.

Part of the difficulty with talking about free will is that there isn't a clear definition of what it is. However, I do think I can say what it isn't--I think most people don't consider randomness or determinism to be free will. But I also think that it would be nigh impossible to distinguish randomness from free will without a clearer definition of free will.

The fallout of this belief is that I think we could disprove free will if we could prove determinism (but, I don't think we can prove determinism currently). But I don't think we can prove free will, even if we could disprove determinism (I don't think we can do that either). In order to prove free will we would have to disprove both determinism and randomness.

I think in the end we will find out that both interpretations of whether free will exists might be correct, but with determinism being more fundamental. Think about a deterministic machine learning algorithm with a fixed input of training data. The agent/algorithm learns by reacting to it's data input/environment, has goals in the form of a cost function it is actively trying to minimize, has options to adjust its neural weights depending on what it experiences, yet we know the system including the environment is predetermined from the moment we kick off the program. -Khatchig, from "Digital Physics" (the movie).

Can tou explain why qm and bell's inequality do not categorically undermine determinism and prove deep randomness ?

Is it because on a macro scale amd normal energy levels effects are negligible? Because there are well followed probability distributions?

Bell's inequality refutes theories with local hidden variables (e.g. some unknown property of particles or spacetime). That leaves two possibilities: no hidden variables (the results truly are non-deterministic) or non-local hidden variables (e.g. some property which exists outside the spacetime which we're constrained to)

There's a third possibility, superdeterminism, in which literally everything is determined by previous conditions, so that no communication is needed, even via a property outside of our spacetime, because everything was determined by preconditions.

Think of it this way: you set two clocks to the same time. The next day, they both have the same time on them. This isn't because the clocks are communicating via nonlocal hidden variables, it's because their precondition was the same and the sequence of events which followed was deterministic.

How would you test free will?

Reply on news.ycombinator.com