Lack of Freewill doesn't mean lack of choice

That depends completely on the given situation. For the cue ball to knock the nine ball into the pocket, there is a multitude of factors that need to be fulfilled. In that particular situation, we can determine all the relevant factors. That’s my understanding at least. Could you point to something in this example that would be a good illustration that causation still can’t be said to have happened? I think it would be really helpful for my understanding.

I’ll just state some things to give some more detail of how I see it, just for clarity.

We have solid objects – the pool cue, two balls, the table. We know enough about how these objects can interact, to make reliable predictions about their interactions given different scenarios. We have very good reason to assume that none of the objects will can occupy the same space at the same time. We know enough that we are also aware, that in order for that to happen, a lot of the details of why it is would have to change for it to be possible. And given all we are aware of, we have a very high level of confidence that those things are not going to change. You could maybe say that the probability is vanishingly low, but I would say that’s just being careful – normally in a situation like this, there’s no reason to think the conditions would suddenly change, if no indication of that is observed. Of course it doesn’t mean we are saying that the laws of nature could never ever change. All of this is just actually sort of predicated on what we see can justifiably be said to be possible.

This was just a difference in the understanding of ‘possible’. It may be that I should replace it with ‘plausible’. I think what I should have said is: something being possible doesn’t mean it’s plausible.

Thanks for this! Saved me from posting my own screed that probably would have said the same things in a much more longwinded manner… :sweat_smile:

Well, first of all, it’s clear (I think?) that things could’ve happened differently: if there’s some element of randomness involved, then you might’ve easily acted differently (I pointedly am not saying ‘reached a different decision’, because of course, random factors aren’t decisions). So it’s not impossible, or even particularly noteworthy, for the same situation to yield a different outcome.

Now it’s just that this different outcome needs to be coupled to your will, your intention, your volition. And here, you’re saying that can’t be the case: because there’s some set of determining factors—your personality, your knowledge, your preferences—and these are what determine your will.

But that’s basically assuming your conclusion: that the will is the sort of thing where the inputs determine the output (ignoring the possibility of random variation, which we can ignore for the time being). But that already means you’re expecting the will to work in a mechanistic way, like a computer, something chugging away on the input to produce an output, according to some procedure, some algorithm. Something whose evolution is determined by the initial conditions—a deterministic system. But that’s exactly what’s at issue.

It’s here that things like Thomson’s lamp come in. For any finite stretch of switches, the initial conditions determine the final state. Given the initial state, and the number of switches, the final state is perfectly determined. But if you allow it to complete its two-second run of operations, then the state after those two seconds is no longer determined by the initial conditions; it decouples from them. But that doesn’t make it random: systems of this sort can solve well-defined problems that finite systems can’t solve, like the halting problem for arbitrary Turing machines.

There’s an element of self-determination, of self-reference to such systems: the state of Thomson’s lamp is always given by the state of Thomson’s lamp, plus another state switch. So then, your will is determined by your preferences, your personality, your knowledge, and your will—which is determined by your preferences, your personality, your knowledge, and your will. And so on.

Of course, I have no idea if such a thing is possible. Given a more reasonable alternative, I’d chuck it out in a heartbeat. But it seems that we always run into such problems of self-determination. It’s clearest with the notion of randomness: to generate a random bit requires exactly the same (hyper-)computational capacity that the infinite machines I’ve just described possess. So if I believe in randomness, what grounds do I have to deny free will?

Eh, I don’t know. I don’t think this place is quite that strict. Certainly, don’t worry about offending or insulting me. Either you’re trying to offend me—in which case, why would I want to give you the satisfaction of succeeding? Or you’re not—in which case, what would be the point of being offended? :wink:

Well, for instance, one factor is conservation of momentum. But why should momentum be conserved? What is it that makes it so? What language does the cue ball use to tell the object ball to move in that particular direction, at that particular speed?

In some way, the properties of the cue ball must dictate the properties of the object ball. All we know is that they do—but we don’t know anything about how. Again, I think the Rule 110 toy example is useful here. One of its nice properties is that it’s computationally universal—so if you subscribe to the idea that our universe might be a simulation, then it might be the case that underneath it all, it runs on Rule 110. So the universe might, for all intents and purposes, just be a particular evolution of Rule 110.

So let’s look at the evolution rules for Rule 110. As noted above, the pattern \blacksquare\square\blacksquare leads to the state \blacksquare for the middle cell in the next round. That’s the analogy to the cue ball knocking the object ball around. So the question of causality, then, is: how? What is it about the arrangement \blacksquare\square\blacksquare that makes it necessary that the middle square transitions to \blacksquare? All we know, if we were studying the evolution of a particular instance of Rule 110 over some finite amount of time, is that this seems to be the rule; but no amount of poring over the picture—no empirical investigation—will tell us why. It always remains an option that, in the next row, we’ll suddenly find \blacksquare and \square trading places—if we’re looking at a kind of ‘block universe’.

So all that we know is that, on the part of the ‘universe’ we’ve observed, there’s a persistent regularity that has \blacksquare\square\blacksquare transitioning to \blacksquare. But to claim, from that, an understanding of causality needs something more: it needs knowledge of how \blacksquare\square\blacksquare makes the state \blacksquare necessary. But that’s clearly not achievable, as at any point, things could work out differently.

Likewise with the billiards example. We may know that two billiard balls interact such as to conserve momentum; but we have no idea what it is that determines that (Noether’s theorem doesn’t help). We have no certainty that things couldn’t stop happening that way, but if we had knowledge of causality, then we would have that certainty, as we would know why the reaction of the object ball is necessitated by the cue ball’s impact. Hence, we have no knowledge of causality.

Every event is predicated by relevant conditions being fulfilled.

It lets us reliably predict any event that we know the required conditions of. Causes aren’t just any events correlating with the predicted effect, they are known conditions for it.

Yes. But what do we and can we know about will? As a model, what predictions has it made possible? Compare to: “If causality exists, knowing the cause will let us predict what follows from it.” And this principle seems to be in wide use. Only, your definitions have thrown me for a loop, and I haven’t yet been able to satisfactorily parse what you would accept as or understand with “cause”.

That’s because the statement is left a bit incomplete. Let’s try to build this further:

Q: How did X happen? A: It was caused by {conditions listed here}
(Each condition can be examined further, following the chain of events.)

Q: How did X happen? A: It was willed by {person? entity?}
(Does a human willing something cause a god to will all the things needed to happen for that human’s will to be fullfilled? No, since causality does not exist, therefore one thing cannot cause another.)

This is my view, but I have to again add the caveat that I have not yet been able to study the matter enough to be able to confidently say I really understand what you ultimately mean with causality or cause. So I think I will just have to concede that I am unable to grasp enough of the concepts to really be able to say much more about them at this point. It looks like I need to much better understand what is meant by the concepts of constant conjunction, correlation and causation in this context.

I’m out of bullets, at least for now. And even if my proverbial gun did have any, it could probably never cause the bullets to travel through the air, and even if it did, the bullets would never cause any damage to anything they hit. :wink:


That’s not a mechanism, it’s just a restatement of the notion of causation. The mechanism is how those ‘relevant conditions’ predicate the event. What is it that they do? How does A make it so that B necessarily occurs?

Why does causality make predictions more reliable? All we ever have is the observed regularities. On the basis that they continue to hold, we then make predictions. Whether there’s a causal mediation at work that ensures that the regularity holds, whether the regularity just holds under certain conditions, or whether the regularity is ensured in another way—statistical effects, god, the structure of the block universe—simply doesn’t matter.

The notion that the regularity does hold is a methodological principle, not unlike Occam’s razor. The simplest possible explanation isn’t guaranteed to be the right one, it’s just the only sensible place to start; similarly, the expectation that the regularity continues to hold isn’t guaranteed—can never be guaranteed—but is the only sensible assumption to make, on the basis that you have no grounds to expect it to fail. So observing the sun to rise in the east on successive days, you’re warranted to bet it’ll rise tomorrow; you have no warrant to bet that it’ll suddenly rise in the west, or not rise at all.

Well, there’s a trivial way to know things about the will: ask. Also, there’s of course empathy: you can explain other’s action by putting yourself in their shoes, and guess their intention. That’s typically going to be more effective than analyzing their neural activity to see what muscular activity is likely to be consequent from that.

And again, what do and can we know about causes? You seem to be reluctant to engage with the example I’m offering, but it really makes things more clear, in my opinion. In the string \blacksquare\square\blacksquare, what do we know about the central cell? We know it’s white, and that it’s surrounded by black cells. That’s it. We also observe that, on some set of observations, it changes its color in the next step. That’s the regularity.

Suppose now we were to believe that this gives us knowledge about causality. So say we attribute to it another property: ‘turns black if surrounded by black cells’. If we could know that it has that property, then we could make perfectly reliable predictions about its behavior. But, unlike the property of being white, we can never know that it has that property: even if we’ve observed it changing color in this circumstances a million times, that doesn’t mean it must do so the million-and-first time. I could easily set up an example where it doesn’t.

So no amount of observation will ever tell us that the cell has this property. Consequently, all that we have is the observed regularity, and the reasonable expectation—which may turn out to be wrong—that it continues. And our ability to predict is, consequently, just predicated on that regularity—but not on it being ensured by some causal factor.

Well, but every such chain either goes on forever, or terminates at some you-can’t-ask-anymore-questions point. It’s the same with the will: if you choose to buy a Milky Way, I can ask you why you did that, and you might say, because I wanted to, and I could ask, why did you want to, and you could say, because I like chocolate, and I could ask, but do you always just buy what you like, and so on. It either goes on forever, or terminates at some ‘because I wanted to’.

What you’re asking about further is the question I’m claiming we don’t have any answer to: how the will works—what its mechanism is. But this question, we also can’t answer for causality—or random chance, for that matter, or god’s intervention, or the reason for the constitution of the block universe—all those are black boxes. But neither is any worse off than the others.

You’re happy with leaving the black box closed in the case of causation, and paper over it by saying there’s some sort of mysterious ‘predication’ that happens when conditions are fulfilled—but that’s like saying that opium’s sleep-inducing powers stem from its virtus dormitiva. In the case of free will, however, you insist we ought to be able to pry open the box. Of course, this is fine: we all have to take some things for granted for a starting point. But you can’t then claim that your starting point is any better than another; it’s just a matter of metaphysical taste.

When you arrive at Alpha Centauri won’t my clock still say T+1 hour from your perspective? So from both of our perspectives the marriage takes place after you arrive in Alpha Centauri. My marriage would take place at T+2 hours according to my clock as observed by both of us. Me and my bride can follow you to Alpha Centauri the instant we see you arrive safely, then when we arrive both clocks would match at T+4.5 years + 1 hour.

Therefore it seems to me you can use the clock as an absolute measure of time. That is, my marriage took place when my local clock reads T+2h, which is the same moment that your local clock reads T+2h.

I’m reading about kinematic time dilation online and find it equally incomprehensible, but as that is generally accepted physics, perhaps I’d better excuse myself from holding any further opinions on metaphysics until I understand it.


No. It will be less than T+1 hour, as in my reference frame your clock ticked more slowly than mine (even though in your reference frame, my clock is the slow one).

I wouldn’t claim to understand it well either.
I just think you’re on a hiding to nothing trying to consider spacetime as spatial slices through time. e.g. Relativity of simultaneity

For anyone wondering why we’re talking about this, the claim was that the “block universe” concept would essentially mean that the universe was nothing more than static frames through time, like a picture book. However, there is no universal analogy to “frames” possible here, and causality still has a privileged position even here (observers can disagree only about the order of non-causally connected events)

That was a mistake on my part. I think the part where you travel 4.5 lightyears in one hour at 99.999% the speed of light confused me. I’ll try again assuming it takes you 4.5 years by your own clock to reach Alpha Centauri.

Your clock reading Your perspective
T mission start, clocks synchronized, you leave for Alpha Centauri
T + 2h The Earth and all the stars completely disappeared because the light is blueshifted out of the visible spectrum, but I don’t think space would look black, rather you’d be looking at radio waves or something like CMB. But with the right sci-fi equipment I guess you could still observe the Earth, it just would be slowed down to a practical standstill: you see that my clock still reads T
T + 4.5y You reach Alpha Centauri, you observe Earth as it was when you left, maybe a second later; you can observe that my clock still reads T
T + 4.5y + 2h You observe my wedding, and you observe that my clock reads T + 2h
T + 13.5y You observe me leaving for Alpha Centauri, and you observe that my clock reads T + 4.5y + 2h
T + 18y We meet, my clock matches yours
My clock reading My perspective
T mission start, clocks synchronized, you leave for Alpha Centauri
T + 2h I get married. I can’t observe you directly at all because light from your ship is redshifted way out of the visible spectrum. With high powered sci-fi equipment I can observe you as being ~7 AU away from the Earth, and I can see that your clock reads T+1h
T + 9y I observe you reaching Alpha Centauri, I observe your clock reads T + 2h, and I set off on my own spaceship with wife & kids
T + 18y We meet and our clocks match
Absolute time event
T mission start
T + 2h I marry
T + 4.5y You reach Alpha Centauri
T + 4.5y + 2h You observe my wedding
T + 9y I observe your arrival at Alpha Centauri and leave with my family to meet you
T + 13.5y You observe me leaving Earth
T + 18y I reach Alpha Centauri


Let me get something straight here, because this discussion always makes me dizzy.

I tend to look at this debate in relation to practical outcomes, e.g. the concept that it’s immoral (whatever that means, that’s another debate entirely in my mind) to punish people for their actions, because they were inevitable and they didn’t have a chance to choose differently. Am I correct that in this case, the response seems to be, “what does it mean to ‘choose’?”

Again, it’s perfectly well possible to formulate relativity in terms of an explicit decomposition into slices through time. In the ADM formalism, the dynamical variables of the theory are the three-dimensional metric on the slices and their conjugate variables (usually called ‘momenta’). The way the slices are ‘glued together’ is described by four constraint functions. Special relativity can also be formulated in terms of a Lorentzian ether, yielding an empirically equivalent theory with an explicitly preferred frame.

Besides, the question is rather moot. If we live in a block universe such that our laws of physics are merely due to regularities—accidental or due to whatever else—in some part of it, then relativity would simply not apply to the universe in total—so arguing from the idea that relativity constrains the type of block universe we may live in is just begging the question.

Sorry I forgot about this thread

It is not fallacious and I just explained why.
We used Newton’s theory of gravity to predict the existence of a planet, which we named Uranus. We used the big bang theory to predict the microwave background frequency. That is what science is, it is not just about “observing regularities”.

And thank you for acknowledging that no part of your argument has been to support free will, merely to try (and fail) to show that alternative explanations are also useless.

Once again, this kind of argument is trying to pull at the foundations of how we know anything at all, the very basis of how we make empirical deductions.

And it is very telling.
There is nothing about “free will” that means it should have universal scope: if it exists, it is a phenomenon limited to some subset of organic life. We shouldn’t need to question the very foundations of induction and inference to think of some way to support it.

So you’re agreeing with my claim, right?
If you’re saying other things are incoherent as well that’s implicitly agreeing that free will is incoherent.

I cannot claim free will doesn’t have a concrete definition without myself giving it a concrete definition?

Well I guess you got me.

I thought they discovered Uranus with a simple telescope? Not based on a prediction.

I thought it went the other way around? nvm


I haven’t said it’s just about ‘observing regularities’; I have said that all we know about is these regularities, which we can then use to make predictions. What we don’t know is that these regularities are due to causal mediation—and no amount of observation can ever tell us that.

What we’ve used to predict the existence of Uranus is a certain mathematical relation. That relation is a description of a certain regularity. A prediction is then made by proposing that this regularity continues to hold. If that prediction happens to come true, it thus demonstrates that the regularity has, in fact, held. But that doesn’t tell us anything about why that regularity held, or indeed, if it is going to continue to hold. But that’s the question of causality.

Indeed, any universe that’s not totally random will have certain regularities that can be used to make predictions. That’s how data compression works. If you compress a .txt-file into a .zip, the reverse process takes the data in the .zip—the observations, so to speak—and uses them to predict the rest of the .txt. Because there are regularities. That doesn’t make those regularities due to causal mediation, though. One letter I type doesn’t cause the next, I can always just ihaotiha noina otiuah.

Take the example I gave above:

We can clearly infer the regularity from observation. That’s trivial. But we can never infer the causal mediation—we simply have no idea why \blacksquare\square\blacksquare facilitates a transition to \blacksquare. We can’t have any such idea, because at any point, it’s logically possible for \blacksquare\square\blacksquare to not lead to \blacksquare. It’s easy to construct a cellular automaton universe where it does for the first ten million lines, and then doesn’t anymore. So observing those ten million lines never allows us certainty about the transition. And consequently, it never allows us to know about causality—because if we could know how \blacksquare\square\blacksquare makes \blacksquare, we’d also know that it must continue to do so. But we don’t. So we can’t.

Again, this isn’t some point I’ve made up out of thin air. It’s the reason Russell says that physics has ceased looking for causes, because there simply are none. It’s the reason Hume calls the inference of causation a mere habit of thought.

And once more, this does nothing at all to threaten our ability to make predictions. In the cellular automaton, I can observe the evolution, deduce the rules for state transitions, and make predictions based on them—that’s entirely rational. But those predictions might turn out to be wrong.

The real world is also replete with successful predictions made of systems where there is no causal mediation. One such example are statistical effects—extraordinary performance is generally followed by poorer results, due to regression to the mean. That doesn’t mean there’s a causal relation there. Or take the example of thermodynamics. Entropy increases because there are more high-entropy states than low-entropy states, so any change in a system is (astronomically) more likely to lead towards a higher-entropy state—but it doesn’t have to. It’s not caused to do so. But perfectly good predictions are possible on this basis.

So the notion that our ability to make predictions is, in any way, shape, or form, linked to our ability to know causal relations is just nonsense.

Acknowledge? I’ve told you that, time and again, starting from my second post in this thread; yet you’ve tried over and over to misrepresent my argument as one in favor of free will, despite my attempts to clarify. Again—the whole history of this conversation is right here. There’s really no use in trying to rewrite it.

Once again, it doesn’t. Inductive reasoning works perfectly well in the examples I gave. You’re trying to give it a power it simply doesn’t have—to infer from regularities to the principles responsible for these regularities. That sort of overreach weakens the position of those arguing for a naturalistic explanation of the universe—because if such an explanation could only come at the price of insisting that we can know things we manifestly can’t know, such that the observation of the ‘constant conjunction’ of events allows us to infer their causal connection, the whole edifice would just be built on sand. So it’s your reasoning that threatens ‘the foundations of how we know anything at all’, not mine.

Well, the quotes around ‘incoherence’ do some work here, but broadly, yes, I think—as I’ve made explicit at multiple points throughout this thread—that if one thinks the problems of free will make that notion incoherent, then so must every other notion of ‘how stuff happens’ be. But the other way around works, too: if one thinks that the supposed ‘incoherence’ is merely apparent, then free will isn’t any less reasonable then, say, random chance.

The only thing you can’t do is go and say that free will is incoherent, but causality is all fine and dandy. That’s just being inconsistent.

That’s not all you are claiming, though. You’re claiming to be able to show that there is no free will—but you can’t both say ‘there is no x’ and ‘I don’t know what x is’; you can say ‘nobody really knows what x is, so nobody knows whether there is x’, but that’s not what you’ve been doing.

There are lots of perfectly good working definitions of free will, all of which are just a second’s google query away. There are also issues with pretty much all of these definitions. That doesn’t mean that we can’t intelligibly discuss free will—the same thing is true for ‘knowledge’, for ‘local realism’, for ‘life’, hell, even for ‘cat’. I can’t precisely tell you where ‘orange’ begins and ends in the rainbow, but that doesn’t mean I don’t know what ‘orange’ is, or don’t know whether something is orange or not. It’s just that simple, box-like concepts aren’t always a great match to a world of vagueness and continuity that doesn’t cleanly delineate into neat little well-separated taxa.

My resistance in giving a definition of ‘free will’ is just due to that fact—for one, we each have a pretty good understanding of the term; and two, once we start taking out the dictionary, the discussion is over, anyway. A discussion about semantics and definitions can be had, but isn’t one I’m interested in, and let’s be honest, would in all likelihood just be a game of lame gotchas. Free will means just that you can do as you will, that you can choose between open options, that you’re not riding through the world on a one-track rollercoaster. That’s all there is. The game of definitions is always futile.

Neptune was the planet detected following a mathematical prediction, not Uranus.

Thanks, I didn’t know that - I had assumed all the planets were discovered by telescope. Learn something new every day.


In our experience this may be true, but there is no reason to expect that robots and AI cannot have free will, assuming that free will exists at all in any meaningful way (which seems doubtful).

Oops, thanks.

I was just being concise there. The point is, free will is not a part of most physical phenomena and therefore we shouldn’t need to throw out our very view of the universe to try to defend it.

Or at least: we shouldn’t if the concept is not complete garbage.

This is the kind of statement you were making when invoking “regularities”:

This is false. It makes a difference. Consider again the big bang.
Using the big bang model we have been able to make many accurate inferences about cosmic history, the CMB being just the most famous. These models implicitly assume causality: X causing Y causing Z.
And it works.

Of course we could “God of the gaps” it, and say God is responsible for X (or somehow “mediates” causal interactions), but that doesn’t make the models equal. Because the god hypothesis did not enable us to make any inferences in itself.

We’re getting somewhere, but let’s be clear that the “problems of free will” is in its lack of clear definition, something not actually shared by all things. It’s not shared by Causality.

And it’s also puzzling why, again, you are categorizing “free will” as one of the theories of “how stuff happens”. The vast majority of physical events in the universe have nothing to do with humans (or AI), so whether or not free will is meaningful, or exists, we would still have most of our work ahead of us in explaining “how stuff happens”.

What?! In this post you’ve accused me of misstating your position, yet you throw in this whopper.

I have clearly stated, from my first post and many times since, that I think the free will concept itself is garbage and therefore it’s meaningless to say it exists or not. Heck, I used the expression “not even wrong” three times alone.

You know, I realized something. I keep giving you explicit examples of how the above just isn’t right, and arguments, and links, and cites, and puzzling over how none of that seems to be able to make any difference. You just stick to your belief, no matter the weight of counterpoints.

But then again, I’m doing the same thing, aren’t I. I keep expecting you to act reasonably, to reconsider your position in the face of explicit counterexamples, and make the same arguments over and over again, to no effect, which you then ignore and continue to repeat the same erroneous statements again and again.

So I’ll try one last time, and then learn my lesson. No, of course there’s nothing that hinges on causality in the big bang model. The Rule 110-example (or any of the others) should make it clear why: a regularity, observed a finite number of times, never allows us to conclude that it holds invariably. But causality is a relation of necessity: for A to cause B means that, absent defeaters, A’s occurrence necessitates B’s. But no matter how often we’ve observed that, it always remains possible that A occurs, and B fails to. If we can’t exclude this, we don’t know that there’s a causal relationship between the two at all.

And now for a more concrete example. Nelsonian stochastic mechanics (link goes to pdf) is an alternative interpretation of quantum mechanics that aims to derive the Schrödinger equation from a fundamental, irreducibly indeterministic diffusion process. Similarly, as I’ve already pointed out to no great resonance, Einstein’s equations describing general relativity can be derived from an underlying stochastic, thermodynamical system. Taken together, then, we have a proposal that the fundamental dynamics of the universe is irreducibly stochastic—that there’s no causality at all, only emergent regularities of the sort present in thermodynamics.

Nevertheless, this will produce the same universe as we see around us. We’re going to observe, to the limits of experimental accuracy, the same behaviors, deduce the same laws, and make the same predictions. But then—and that’s the crucial point—causality never was relevant to the predictions in the first place. If I claim Y (our ability to make certain predictions) is a consequence of X (causality), and thus, Y gives us abductive reason to believe in X, and it turns out that you can get Y without X, then I can’t appeal to Y as giving us license to believe in X anymore.

This lesson stands even if it turns out that Nelson’s program or the thermodynamic interpretation of general relativity can’t be carried through. We know random systems can give rise to stable, predictable regularities. Laws can emerge from fundamental chaos, so the mere existence of laws, and the prediction we make using them, can never certify any causal connection.

And again, it’s not just randomness that can yield the same predictive ability. We could be living in a simulation—and any number of observations of B following A, and rules deduced and predictions made on that basis, won’t help when the code says that after 10^{61} simulation steps, A won’t be followed by B anymore. So there is no sense in which A necessitates B in this case, and hence, no causal relation between the two. But we have just the same ability to make predictions.

By the way, there’s a certain irony in you choosing the CMB as an example of a prediction based on ‘causality’. The radiation follows a black-body distribution, which is described by Planck’s law, which in turn arises from the Bose-Einstein statistics of the photons. So that is has the form it does is actually due to statistical effects—in principle, the system could fluctuate away from equilibrium, leading to a different radiation distribution. There is no causal principle here that necessitates the spectrum to have the form it does; it’s just the expected behavior for (bosonic) particles randomly distributed over available energy levels. So it’s an example of a successful prediction based on stochastic behavior—thus refuting the idea that there need to be causal relations for us to be able to make successful predictions.

Of course, that doesn’t mean that there’s an actual stochastic process at work in the generation of a black-body spectrum. The caution against overinterpreting what our models allow us to conclude works both ways: just because something can be modeled statistically, doesn’t entail there is fundamental randomness. Large populations of people can be modeled statistically, but that doesn’t mean that they make random decisions.

Modeling works by means of an equivalence of behavior between two different systems. But that system A, regarding some relevant part of its properties, behaves in the same way system B does, does not entail that the behavior of system B must be produced in the same way as it is in system A. Random systems can be modeled by deterministic ones, and vice versa. That’s a feature, not a bug—it would be disastrous to our ability to make any sort of prediction at all if it were only possible given the existence of causal relations—because these relations are always beyond our knowledge. Causality is a relation of necessity: A necessitates B. But such a relation is always beyond knowability—just because A has always led to B in the past, doesn’t entail that it always will (just because you’ve only seen white swans doesn’t entail all swans are white). So we can never know whether A causes B. If now, as you say, prediction is only possible if there is a causal relation, then there’d simply be no way to predict anything, as there’s no way to know whether there’s a causal relation.

Luckily, however, it doesn’t matter one whit for out ability to make predictions whether the behavior of the system is due to causal mediation, chance, the structure of the block universe, or the whim of one deity of one sort or another. All that matters is that this behavior is stable over time. We can never be able to know whether it is—but we don’t need to.

The only real problem of free will is the infinite regress it gives rise to. If not for this, the model I gave above—the will as determined by your personality, habits, preferences, and circumstances, and the will—is perfectly adequate: the will determines itself, at least partially. It’s not an outcome of the unfolding of the universe, but an input into it, similar to a random event—an irreducible instance of the creation of new information.

Of course, it’s entirely valid to hold that this circularity dooms the notion. But the same circularity exists in randomness. And causality just seems to be clearly defined, to you, because you essentially accept it as a primitive notion—that is, you don’t ask further questions, and leave the black box untouched. If you didn’t, then the notion of A causing B would immediately open up the question of how it is that A causes B—what it is about A that makes it necessitate B. But this just leads down the same rabbit hole. Suppose that it is property C that makes A cause B. Well then, how is it that property C makes A B-causing? This yields something like Carroll’s paradox: if A’s occurrence is the reason ^0 for B’s, what is the reason ^1 why it is that reason ^0? And what is the reason ^2 why the reason ^1 makes A be the reason ^0 for B? And so on.

Sure—I didn’t say that free will is supposed to explain how everything happens. But it is a story that purports to explain, for an event, how it came about; just as causality and randomness does.

You’ve made that claim, yes, but you’ve (to your credit) also attempted to bolster it with an argument:

This proceeds from making a claim regarding what free will would entail—namely, a ‘non-determined determination’. My point was just that you must have some idea of what free will is supposed to be in order to make such an argument. Otherwise, how do you know it’s a ‘non-determined determination’?

And in fact, that’s all the idea needed for the discussion we’re having here. This is the argument I came into this thread to attack:

So, or at least that’s how it seems to me, we have a reasonably close idea of what free will would entail: namely, something being determinate, yet undetermined (by anything else). Hence, all this clamoring for a nice and tidy definition is just superfluous—we’re already essentially on the same page, there. The difference is just that you believe that this conflict makes the notion of free will impossible, while I hold that, if one honestly investigates matters, the other notions of how stuff happens aren’t better off—we always bump up against a foundational antinomy. How does the will determine itself? How does a random event occur, excluding equally possible options? How does A cause B (and what got the whole business started in the first place)? Whence the regularities in the block universe? And so on—we must accept some black box, something that, to the best of our analysis, seems impossible and unknowable, whether it is a prime mover, or things ‘just happening’ for no reason, or things being their own effective cause.

The reason I claim these are, ultimately, the same mysteries is because they all have the same answer: if things can cause themselves, then the will can cause itself; then random events can cause themselves; then the initial conditions of the universe can cause themselves. But every analysis of such self-causation ends in circularity. But this may just tell us something about the limits of human enquiry; that doesn’t mean it tells us anything about the limits of what’s possible in the world.

You mean the ultimate uncaused cause?