Wednesday, June 8, 2016

Counterfactuals and the randomness objection to libertarianism

The randomness objection to libertarian free will says that undetermined choices could only be random rather than reason-governed.

I want to consider a bit of a reply to this. Suppose that you are choosing between A and B. You have a reason R for A and a reason S for B, and you freely end up choosing A. I think the following will be true, and I think the libertarian can say that they are true as well:

  1. If the reason R for A were stronger, you'd still have chosen A.
  2. If the reason S for B were weaker, you'd still have chosen A.
  3. If the reason R for A were noticeably weaker, you might not have chosen A.
  4. If the reason S for B were noticeably stronger, you might not have chosen B.
If this is right, then there is a real counterfactual dependence of your action on the reasons. The dependence isn't as precise as the compatibilist's dependence. The compatibilist's story may allow for precise values of strengths of reasons that produce counterfactuals like (3) and (4) with quantitative antecedents and "would" rather than "might". Still, I don't think anything so precise is needed for reasons-governance of our actions.

So, I think that if I am right that the libertarian can reasonably affirm (1)-(4), then the randomness objection fails. Of the four, I don't think there is any difficulty with (3) and (4): even if there were pure randomness, we would expect (3) and (4) to be true. So the question is: Can the libertarian affirm (1) and (2)? And I think (1) and (2) are in the same boat, so really the question is whether the libertarian can affirm (1)?

And I say: Why not? At the same time, I know that when I've talked with fellow libertarians about this, they've been pretty sceptical about counterfactuals like (1). Their main reason for scepticism was van Inwagen's re-run argument: In indeterministic free choice situations, if you repeated the same circumstances, you'd get different results. And you'd expect to get different results in repeat runs even if you somewhat raised the strength of the reasons for A.

I agree with the re-run intuition here, but I don't see it as incompatible with (1). The re-run intuition is about what we would get at a later time, albeit in a situation that is qualitatively the same. But (1) is about what would have happened at the time you made the original choice, albeit in a situation that was tweaked to favor A more.


Michael Gonzalez said...

Pruss: I'm confused. Are you saying Libertarians doubt that, in a given situation, if I had had even STRONGER reason for choosing A, I would still have chosen A?? Or are you saying they doubt that such a choice is still free? Help me out, please, because I think I have a couple of relevant thoughts on this, but first I need to understand the question.

Peter said...

In (1), do you mean the very same reason, R, is stronger than in the actual cases, or do you mean the agent has some other reason, R1, which is stronger than R. If you mean the latter, then (1) seems to be false. Suppose I decide to go for a run. My reason, R, for doing so is that I enjoy running. Suppose that I actually had a different, but stronger reason, R1, for example, that my doctor has told me that I need to do so to improve my health. It seems that if we replace R with R1, then (assuming libertarianism), we cannot say what I would have done. This would seem to be the case even if the reason is both objectively more weighty and if I judge it to be more weighty.

If we take (1) to mean that R remains the same, but is stronger, then I am not sure I understand. What would it mean for the same reason to be stronger than it is? Does it just mean that the agent assigns more weight to the same reason?

Peter said...

I wonder whether event-causal libertarians may have a problem with (1), (especially if they are naturalists).
Consider a physical object acting indeterministically. Let’s stick with a sci-fi case: a bomb is created, and it is indeterminstic whether it will go off. Supply whatever causal mechanism one would like to explain this indeterminism, as long as it is inanimate (maybe there is some deterministic sensor, which will be set off by some indeterministic event at the micro level, or maybe someone discovered or developed a macro-level object that acts indeterminsitically).
Now let’s say the bomb actually goes off.
Now consider proposition (4): “If x, the bomb would still have gone off.”
Whether this sentence is true depends, I think, on what we substitute for x. Suppose we replace x with “some completely unrelated event, light years from the bomb, is slightly altered, at the very moment the bomb goes off.” It seems plausible that (4) would then be true. Although we could run van Inwagen-style replays, I think we can dismiss them, as you wish to do, since they concern other times.
But what if somehow we find a way to modify the probabilities governing whether the indeterministic element of the detonator acts this way or that. Suppose in the original case, this element has a .5 probability of acting such that the bomb will be detonated. What if it had been possible for us to keep the very same detonator, but increase the probability that the bomb would go off, not by adding an additional way for the bomb to explode, but simply by changing the most basic probabilities governing whether the indeterministic element would act in this way or that? Suppose we could have done so, but didn’t in the original scenario. Is the following true:
(5) If we had increased the probability that the bomb would detonate, the bomb would have detonated.
I am not sure it is. By altering the cause, I think the outcome of the event is put into question, even though the probability that the bomb will go off is higher than in the original case.
Do you think (5) is true?

Alexander R Pruss said...


Anecdotally, when I talk to libertarians about this, they are sceptical about whether I would have still chosen A had the reason for A been stronger. They think: Well, you're free, so you might have chosen B.

Alexander R Pruss said...


I'd like to be able to cite your response if I publish on this. If you are willing to share your name for purposes of such citation, please email it to me (

I did mean the same reason. The individuation of reasons is tricky here. Let's say you offer me a pound of silver to do some task, and I do it. Had you offered a pound of gold, would that have provided me with the same reason to a stronger degree? That depends on whether my interest was in the monetary value of the silver or in something else like its aesthetic value. If my interest were in the monetary value alone, then my reason was: "Get such-and-such an amount of money." And the pound of gold gives a lot more money, so it's the same reason, but stronger. On the other hand, if my interest in the metal were aesthetic, then the pound of gold might provide a *different* aesthetic reason, since gold and silver instantiate different aesthetic goods (gold is colorful while silver is elegant, say). In that case, the reasons might be incommensurable.

Your naturalism comment is really interesting and makes me more sympathetic to those libertarians who deny (1). It really does seem to depend on the mechanics of the case.

Intuition: You have a rope. It breaks indeterministically when a 100 lb person sits on it. Then if that person were to weigh 200 lb and were to have loaded the rope in the same way, it would still have broken.

But it's crucial that the rope was loaded the same way. There was presumably some first fiber of the rope to break. You could put a 200 lb load on the rope in such a way that if it were to break, a different fiber would have to give way first. Then no guarantee that the rope would still have broken. Moreover, it is nigh impossible in practice to load the rope the same way but with a larger load. In other words, the fine structure of the situation matters a lot.

In the mental case, if dualism is true, it could be that there is no microstructure. A choice looks simply like this: "There is a tuple of reasons, with varying strengths, some favoring A and some favoring B." And then it makes sense to simply talk of varying the strength of a reason.

But if naturalism is true, there is microstructure. The reasons are grounded in arrangements of particles. Increasing the strength of a reason changes the microstructure. There might be no way of changing the microstructure in such a way that it's plausibly true that you would still have chosen the same way.


Alexander R Pruss said...


This is really interesting. Suppose that (1) is not only sufficient but necessary to answer the randomness objection. Then we have this cool argument for dualism:
i. Freedom is incompatible with determinism. (Insert your favorite argument for it.)
ii. If in general not (1), then we have no freedom.
iii. If naturalism and indeterminism, then in general not (1).
iv. But we have freedom.
v. So, indeterminism. (i and iv)
vi. So, sometimes (1). (ii and iv)
v. So, not naturalism. (iii, v, and vi)

In fact, maybe we can even give a direct argument against naturalism and indeterminism along these lines. If naturalism, then no doubt our mental activity can be multiply instantiated by our microstructure. (Compare how when you run the same computer program multiple times on a modern computer, it gets loaded in different parts of physical memory.) It's false that had the microstructure been different but the mental states been the same, I would have chosen the same way. In fact, in cases where the choice made wasn't more likely than not, it's not even true that had the microstructure been different, I would likely have chosen the same way. This means that if indeterminism + naturalism, then our choices depend in an objectionable way on microstructure that's significantly underdetermined by our mental states, and we're not free. So if freedom + indeterminism, then not naturalism.

But I don't think it matters too much for this argument whether agent- or event-causation is the right story.

I am also worried that some such argument can be run against the dualist, too.

Michael Gonzalez said...

How can the Libertarian accept (1), or anything like it, if they don't even actually think the following is true?

1*) Given the exact same circumstances, with R being no weaker or stronger, you'd have chosen A.

If it is true that you would always have chosen A, given that R had the strength it had, under those very same circumstances, then those who follow van Inwagen's reasoning would say it's starting to look rather deterministic. And, surely (1) is a step beyond (1*), simply by being different in any relevant way from the situation in which we know you picked A.

As for your (iii), which was based on the reasoning in the previous post, I'm completely confused as to why we should believe that a naturalistic case, in which changes to the microstructure are, in this scenario, such that the weight in favor of R is INCREASED, should somehow negate (1)! (1) says that, had the reasons been stronger, I would still have chosen A. Naturalism says that the meaning of "reasons had been stronger" has to do with microstructures being geared MORE FAVORABLY toward those reasons. This FAVORS (1), does it not?

Alexander R Pruss said...

The libertarian can accept (1*) simply by accepting the Centering rule for subjunctive conditionals (if p and q, then p→q).

As for (iii), think about this scenario. You're playing a simple dice game where you roll one die and you win if you roll a one. You roll a 20-sided die and get a one. Hurrah! Now consider the counterfactual:
(1**) If you had rolled a 6-sided die, you would still have rolled a one.
No, that's not true. You might well have rolled a five or a two or whatever. It's a different die, so the actual world's outcome with a 20-sided die isn't very relevant.

On the other hand, the following could perhaps be true:
(1***) If the die was loaded a bit more in favor of one, you would still have rolled a one.
Whether it is true depends, I think, on the details of the physics of die rolling, the degree of chaos involved, etc.

Here's another case. You roll 20-sided die and you win if you have one. In the actual world, you win. Then plausibly:
(1****) If the rules were that you win if you roll either one or two, you'd still have won.

On the other hand:
Not: (1*****) If the rules were that you win if you roll two or more, you'd still have won.
Even though those rules favor winning!

Heath White said...

What about

(*) In any case of freely choosing between A and B, you might not choose A.

Given (*), it is never the case that "If X, you would freely choose A", for any X. Because "would choose A" is incompatible with "might not choose A."

Note that (*) is entailed by

(**) In any case of freely choosing between A and B, there is some non-zero chance you will choose B.

Heath White said...

Addendum: the libertarian who wants to accept (1) is committed to denying:

"If it is possible that X happens, X might happen"

which sounds like a pretty strong principle to me.

Michael Gonzalez said...

Pruss: You folks keep comparing Libertarian choices to dice rolls (which are actually physically determined, and you admitted that it was those physically determining factors which came into play in loading the die or rolling a lesser-sided die, etc). Even if we left that behind and went to straight quantum inderminacy, how is that anything like choices we freely (in a Libertarian sense) make?

Heath: There are Molinists who CLAIM Libertarianism about free will, and yet they insist there are conditional statements about what we WOULD freely do given a set of circumstances. However, I'm with you on this, and I think that is clearly mistaken.

One question, though: Do we treat the "might" statements probabilistically? If we can assign such probabilities, does that take anything away from the freedom of the choice or mechanize it in some unacceptable way?

Cameron said...

I don't really see the difference between the modified re-run and (1). You say (1) is not about what would happen at a later time, but then neither is the re-run. The re-run is about re-running history. And (1) is about re-running history (albeit slightly tweaked).

Alexander R Pruss said...


In the so-called "re-run", we are inducing at a *later* time a situation that occurred *earlier*. In (1) we are considering what it would have been like had the *original* situation at the *original* time been different.

Alexander R Pruss said...


I think there is an equivocation on "might" somewhere in our discussion. The "might" and "possible" you're thinking of is "might" or "possible" given a particular past. The "might" I'm thinking of is the David Lewis "might" which is dual to "would", and which takes into account all of time.

Centering is trivial on Lewis semantics. If p and q are true, then the closest world where p is true is the actual world, where q is true, so p→q is guaranteed to be true.

Cameron said...

Thanks for the response. Seems to me that van Inwagen's re-run could be construed either way. And the intuitions don't really seem to differ all that much between them.

Alexander R Pruss said...

Think about similarity of worlds. The world that the same choice is made in is more similar, other things being equal, than the one where a different choice is made.

Alexander R Pruss said...

Think about similarity of worlds. The world that the same choice is made in is more similar, other things being equal, than the one where a different choice is made.

Heath White said...


I was also worried about an equivocation, but for (I think) different reasons. Consider

If X, you would do A, though it is possible you don’t do A.

What “would do A” means is that in all the CLOSEST X-worlds, you do A. What “possible you don’t do A” means is that in SOME (not necessarily close) X-worlds, you don’t do A. So the statement is consistent: it says that in the close X-worlds you always do A but in some more distant X-worlds you don’t.

This means the libertarian who wants to say this kind of thing should either reject outright “(*) In any case of freely choosing between A and B, you might not choose A”, as well as “If it is possible that X happens, X might happen,” or alternatively should say that in these sentences, the context of ‘might’ is set in ways that are wider than they are in e.g. Molinist CFs (contextualism about ‘might’).

I don’t think any of this has to do with time. (And IIRC, Van Inwagen’s rollback argument is framed in terms of rolling back time itself to replay the very same event, so he is not talking about later times either.)

Maybe the problem, though, is understanding what the distance metaphor is supposed to mean under libertarianism. For example, revisit

(1) If the reason R for A were stronger, you'd still have chosen A.

In the actual world, R and A. The claim is that in all the closest R-is-stronger worlds, you choose A and not B. But it can’t be the claim that in ALL the R-is-stronger worlds you choose A, because then R makes A necessary and you lose libertarian freedom. You can’t say R exerts a stronger but still non-determining force or influence towards A, because then (intuitively) the R-is-stronger&B worlds are rarer but just as close. You can’t say it’s simply a matter of similarity to the actual world, because that makes (1) trivial and I don’t think it’s supposed to be trivial; you’re trying to say you’d still have chosen A BECAUSE R was even stronger.

So the problem is subtler than I thought but I don’t think it’s entirely solved.

Michael Gonzalez said...
This comment has been removed by the author.
Michael Gonzalez said...

This is just a nascent, barely developed thought I've been having for some time, but what if Libertarian free will has more to do with steering the overall flow or course of our lives, than with making individual choices? Of course, the course of our lives is, in some sense, made up of individual choices, but then... is an "individual choice" really coherent? Isn't it embedded in various other choices we've made and are making, and more poignantly are actively upholding throughout large sections of our lives?

Peter Hacker gives a good illustration: Think about the words I'm typing right now. Each of them is voluntarily and freely written, and I am morally praiseworthy/blameworthy for what I write. And yet each word is clearly not preceded by an individual volitional act. This is just as clear (perhaps clearer) in spoken speech. A stream of words pours forth, and I am not individually performing an act of volition to speak each word, and yet the whole discourse is free and I am morally accountable for it.

Again, Raymond Tallis talks about the Libet-style experiments and insists that lifting a finger in response to an urge is hardly the paradigm of free will or moral accountability. Rather, the real choice that is being made (or "being upheld" in Merleau-Ponty's terms) is "participating in Libet's experiment". And perhaps that's embedded in an even larger "choice" that we are upholding (being active in our society maybe?). We are free agents in that we steer the course of these large choices, but within that there are specific individual choices that are just manifestations of that overall, larger set of choices that I'm upholding.

This is very convincing to me, and becomes more so as I consider the various other things I do. For example, I walk, run, or bicycle, and yet in each case the individual leg movements (while surely being "free" and voluntary) are not individually preceded by a volition or act of will. I'm just walking and the individual parts fall into place to uphold that larger choice.

Is this helpful? I fear I may be rambling, but it seems to me that this sort of approach could shift the paradigm and render this sort of question moot while opening up very different questions for consideration.

Michael Gonzalez said...

(continuing...) Think about your (1) again, please, in terms of this other paradigm of thinking about free will and volition. It may well be true that I would have done A, given a stronger R, but you have to add "+ having developed freely in a particular direction, X, throughout the previous years". This development is continuous and is under my control (more or less, given external factors, of course), but where it is at the moment of choosing may necessitate a particular outcome, given a particular set of reasons.

I don't know. That seems like a big step to me; but it may be nothing at all. I'm no philosopher.

Alexander R Pruss said...


Stronger-R-and-A worlds are closer than stronger-R-and-B worlds, because the actual world is an R-and-A world, which more similar to a stronger-R-and-A world than to a stronger-R-and-B world. So on similarity semantics for counterfactuals, that's some reason to think that if R were stronger, you'd choose A. There is, however, a serious difficulty: The same reasoning suggests that weaker-R-and-A is closer than weaker-R-and-B, but we don't want to say that if R were weaker, you'd choose A.

I don't quite know what to do. Perhaps similarity is a guide to counterfactuals but a fallible guide. Perhaps the similarity needs to be evaluated very carefully, but I don't know how to do that.

Alexander R Pruss said...


I distinguish between derivatively and non-derivatively free actions. It might be the case that many of our actions are only derivatively free. My concern in the post is with non-derivatively free actions.

Michael Gonzalez said...

But, Pruss... are you certain there ARE any non-derivatively free actions? It just seems to me that every attempt to nail down Libertarian free will in the case of individual choices ends up in the same sorts of messes. But, if we really reflect about the things we actually do, and the choices we actually make, do we not usually realize that they are all embedded in larger concerns which we are upholding? In other words, that they are all derivative?

I don't know if that's right, but, if it were, it might resolve this whole problem. You'd never have to worry how it can be that A and B are both metaphysically possible (which means you have the power to do either, and in some possible worlds you do one and in some you do the other, and their "nearness" to each other is determined by factors other than R, which means that technically we're yielding the choice of A or B to the extraneous circumstances which distinguish one world from the other... which then sounds like we're right back at metaphysical necessity given the exact set of circumstances.... etc etc). Choices of A and B may very well have a definite, singular possible outcome, but the situation leading up to them has to do with a continuous upholding of larger free "steerings" in the course of one's life. It is THAT over which we exercise sufficient control to be morally accountable for our actions; not the individual actions which result from it.

It just seems worth pursuing as a complete alternative to this more orthodox way in which you (and most other philosophers) usually discuss this issue. It has the potential (I think) to answer Galen Strawn's "Basic Argument", Peter van Inwagen's rewinding argument, etc.

Michael Gonzalez said...

(continuing...) I appreciate your patience with me, by the way. I may be way off-base, and I have a lot more studying to do about this, but consider the moral distinction we make between crimes of passion and premeditated ones. The act is the same. The difference is in the larger, overarching manner in which we are intentionally and voluntarily steering our course of actions.

To use a silly example, think of a person who curses in public. He says an offensive word. However, if it just "slipped out" or was a matter of accidentally giving into an urge (perhaps because he stubbed his toe or something), then we see a strong moral difference between that case and the identical case in the life of someone whose overall life course is one of flouting moral restrictions. Again, the words coming out of my mouth are, in BOTH cases, not preceded by specific acts of will, though the continuous speaking is voluntary and freely engaged in. So what distinguishes the two cases morally (and what distinguishes crimes of passion from premeditated ones) is really the larger picture of how we are steering our lives; not the individual choice which is identical in each case.

We seem to always be asking what the action flowed out of, rather than merely what the action was.

Heath White said...

Suppose I am willing to pay you D dollars to undergo a shock of V voltage. Suppose you are willing to undergo the shock for some values D* and V*. Then holding D* constant, you should be willing to undergo the shock for any VD*.

We can model your willingness to undergo shocks on a Cartesian grid, with the payoff D along the x-axis and the voltage V along the y-axis. For any coordinate where you are willing to undergo shock, we color the plane black at that point; where you are unwilling, we color it white. The result will be a region of the plane colored black below/right of a line that slopes roughly SW-NE, in such a way that its slope is always positive.

Then it seems right to say, for any D* and V* where you are willing to be shocked,
1. If the payoff D were larger, you'd still have chosen to undergo the shock. (Because for a greater payoff you’d take the same shock.)
2. If the voltage V were weaker, you'd still have chosen to undergo the shock. (Because for a lesser voltage you’d take the same payoff.)
3. If the payoff D were noticeably smaller, you might not have chosen to undergo the shock. (Because for some lesser payoff, you would not be willing to take the shock.)
4. If the voltage V were noticeably stronger, you might not have chosen to undergo the shock. (Because for some greater voltage, you would not be willing to take the same payoff.)

It is easy to envisage a deterministic system that would satisfy these constraints; just give it a function F(d,v) in the range {0,1}. But what does it mean to say that a system satisfies these constraints and is NOT deterministic?

One attempt would be to give it a function F(d,v) in the range [0,1], where the result of the function yields a probability of choosing to undergo the shock. Constrain the function so that probabilities monotonically increase as you move right on the grid (increasing payoff) or down (decreasing voltage). But that will not secure 1 and 2. Sure, you chose to take the shock this time (at D* and V*) because there was an 80% chance you would; but if the chance were 90% it doesn’t mean you necessarily would take the shock; you might not.

I think, upon reflection, that this is a long and complex way of putting the standard Luck Argument. But maybe it sharpens the argument. The challenge is to say what is missing from the simple model.

Heath White said...

Sorry for the end of that first paragraph; I was undone by HTML. It should end:

"...willing to undergo the shock for any V less than V*. And holding V* constant, you should be willing to undergo the shock for any D greater than D*."

Michael Gonzalez said...

Heath: If we establish parameters and make a deal on which Vs and Ds I will agree to, then the rest is deterministic all the way. If we go probabilistic, though, then one would have to ask what justifies those probabilities? Do we mean that, if we re-ran the same situation millions of times, the distribution of times I would or wouldn't agree would fall into those probability ranges? It's hard to see how any agent for which such a thing is true could be actually free. It seems as free as a flipped coin or a rolled die, and therefore exactly as morally accountable as such things.

Heath White said...


Yes, that is more or less the argument.

Michael Gonzalez said...

Heath: I can see why this has perplexed people (along with Strawson's Basic Argument). What are your thoughts on giving up individual decisions, and seeing those as just steps in upholding longer-term, overarching trajectories throughout life, and that our freedom (and moral accountability) has to do with the latter; not the former? After all, as Hacker mentions: Every word I'm typing is voluntary, but I am not performing an individual volitional act in each case. I am saying something, in an overarching way, and the individual words are still considered free and voluntary, but not with reference to individual acts of will OR with any regard for what words I might have typed given a rewind of the exact same circumstances. Whether or not I would converse with you and Pruss about this, and share my thoughts, is a much broader, continuous, and repeatedly upheld course of action. The instances of upholding are not really the paradigm of freedom, but the continuous voluntary engagement in the conversation is.

When philosophers talk about free acts in a traditional way, they tend to talk about things which seem to me to be more a matter of giving into urges (or refraining from urges) at particular instants. That doesn't seem to me to be the paradigm of Libertarian freedom, for which we are morally accountable.

Heath White said...


My first thought is that this proposal moves but does not solve the problem. After all, big life trajectories are either deterministic, or not. If deterministic, no freedom. If not deterministic, then probabilistic, thus no freedom. The argument is the same.

Michael Gonzalez said...

It seemed to me that the rewinding of the tape, and getting different results from which to calculate probabilities, would only apply in the case of discrete, individual choices. So, the odds that I would choose A or B in a particular instance might have some definite probability. This would be analogous to the odds that I would type a particular word in this sentence rather than some synonym of it, or perhaps a re-phrasing of the clause, etc. But, having the conversation at all seems different. It isn't located any particular instant, and indeed involves my pondering future responses during timeframes in which I am categorically not "engaging in the conversation" in any sense. It doesn't seem like a discrete event at all, but rather an overarching trajectory which the discrete events uphold.

The picture I'm coming up with here is indeed vague and spotty at the moment. I'd read Tallis and some Hacker, but I emailed Hacker and got the suggestion of reading several other people on the topic (including Kenny, White, and Hampshire, whose books are now on my Amazon wish list). So, maybe I'll need to do the rest of my research before fully fleshing out (or discarding) this idea.

In the meantime, what do you think of the idea of a non-discrete, continuous path which is upheld in the discrete instances, but over which we have free, agent-causal control? It would fit better with the undergirding agential powers that we have (such as freely walking without an act of volition for each step, or freely speaking without an act of volition for each word freely uttered).

Alexander R Pruss said...


Try this. Suppose that D ranges from 0 to 100 and V ranges from 0 to 100 as well. Choose a random number R between -100 and 100. If D-V is bigger than R, accept the deal. Otherwise, reject.

Suppose all I know is that you accepted the deal, but I don't know what the value of R was.

Then your (1) and (2) are true, because if D were higher or V were lower, then D-V would still have been bigger than R. On the other hand, if D were lower or V were higher, then D-V might have been smaller than R (since I don't know what the value of R was), and so you might not have gone for the deal.

But that's not really a solution to your challenge because I only get (3) and (4) with an epistemic "might".

I do have the intuition that one could get some counterfactuals like this in quantum cases. Suppose that an electron has a wavefunction spread along some axis from 0 to 1 inches, and you use an electron-detector calibrated to check if the electron lies in the 0.2 to 0.5 inch segment of the axis. Suppose the detector detects the electron, so that on a collapse interpretation the wavefunction collapses into one that is non-zero in the 0.2 to 0.5 segment and zero elsewhere. Now, consider counterfactuals about calibrating different bounds in the electron detector. Intuitively:
(1) If the lower-bound of the interval (actually 0.2) had been lower, an electron would still have been detected.
(2) If the upper-bound of the interval (actually 0.5) had been higher, an electron would still have been detected.
(3) If the lower-bound of the interval had been higher, an electron might and might not have been detected.
(4) If the upper-bound of the interval had been higher, an electron might and might not have been detected.

In this case, I am confident about (3) and (4), but less so about (1) and (2). (I worry that I am importing classical intuitions into (1) and (2).) Still, all four claims do seem intuitive.

Alexander R Pruss said...


I don't understand how there could be derivative cases without non-derivative ones, except given circularity or infinite regress. (But there is no time-travel in our lives, so no circularity, and we have a finite past, so no infinite regress.)

IanS said...


I suspect that (1) and (2) import the classical intuition that the electron has a definite position. As I understand it, on standard Copenhagen-style interpretations, if the detector had been set differently, it would have been a different experiment.

Michael Gonzalez said...

Pruss: My point was to stop thinking about "cases" altogether; at least, in the sense of specific instances or decisions. It's like with Libet experiments, which were all about the instant of choosing to lift a finger (which is urge-based and not at all the paradigm of free choices anyway). The elephant in the room, which Libet was missing, was that the paradigm of free will is the larger choice to "participate in Libet's experiment". That involved all sorts of particular decisions, each of which upholds that larger one, and many of which don't seem to have anything to do with the experiment or with the end-result of lifting a finger. Perhaps someone wants to call "participating in the experiment" a non-derivative choice, but when did it happen? It's completely vague, time-wise. Indeed, I was upholding that choice even when scheduling other tasks while I was at work, chatting with friends at the bar the previous night, setting my alarm clock, yelling obscenities at the bus driver for leaving me behind and making me late, etc etc.

It seems to me that animals are embedded in a continuous flow of living, not a discrete set of individual choices. We walk around while also looking, hearing, daydreaming, carrying on conversations, etc. It's a lot of what you would call "derivative" actions, but where is the non-derivative one? Living? Coping? (Heidegger's term).

I don't see circularity or infinite regress. You get down to the basic things which are upholding our lives (breathing, heartbeat, etc), then go up to mostly-instinctive but still voluntary things we do constantly (walking, dodging obstacles, conversing), then go up to more thought-out voluntary actions... but it's just the continuous engagement of an animal with its world.

Anyway, I know this isn't quite the topic of the thread, so I'll leave it at that. I just think the entire paradigm in which the thread question is asked is mistaken. It's part of a Bild (as Wittgenstein would say) which may be fundamentally mistaken.

Alexander R Pruss said...


Suppose for simplicity that time is discrete. Then there is a first moment at which it is true that one is responsible for something.

It's just logically necessary that if you existed for a finite number of moments, and at some of them you were responsible for something, then there was a first one at which you were responsible for anything.

At the first moment at which you were responsible, your responsibility surely is not derivative.

If time is non-discrete, then things are more complicated. But since nothing of human significance appears to happen in a Planck time, we can divide your past life into Planck times, and there will be a first Planck time during which you have responsibility for something or other. And a Planck time isn't long enough to get non-derivative responsibility. So there will be a case of non-derivative responsibility.

What about vagueness? Well, supervaluate, and say what I said for any precisification of "responsible". For each precisification, what I said will work: there will be a first moment or first Planck time at which you are non-derivatively responsible.

Alexander R Pruss said...


I think the interpretations of QM are silent on questions of counterfactuals. But you're right that I may be importing classical intuitions.

Michael Gonzalez said...

Pruss: Please bear with me.... If time is discrete, then perhaps there is a moment when I became morally praiseworthy for "exercising at the gym". When did that happen? Was it when I made the plans? When I rejected other plans that would have interfered? When I dragged myself out of bed? When I first arrived at the gym? etc etc... Or, was it at the last rep of the last exercise, after which it could be said that I completed my workout? If we choose this last one (which, I guess, seems most natural), how do we then interpret Peter van Inwagen's rewinding problem? Given the exact same circumstances, could I have at that instant freely NOT done the last rep? Possibly. That rep does follow rather strongly from all the previous reps, from what I'm trying to do overall, etc. There are probably physical factors that play a very large role here. Maybe there's even a probability distribution like Heath was talking about? In any case, I don't think the Libertarian should mind, since the morally praiseworthy deed was not the instantaneous completion of a single rep! It was "working out at the gym"!

I'm not denying that there is an instant at which we become morally responsible. For example, there is an instant at which I spoke the offending word, or placed the offending footstep. And I did those things voluntarily, but not by an individual act of will. They were part of doing something larger. I was continuously engaged in a course, over which I have continuous control (satisfying what the Libertarian seems to want). Each event within that course is, of course, discrete (especially if time is discrete). But the discrete events are just part of doing something else. Or, if they are just the result of a sudden urge, then they are treated differently in courts of law ("crimes of passion"?) and we understand them to be morally distinct.

I'm rambling, so let me just show quickly why I think questions like your original one here might dissolve in this paradigm:

The reasons R for decision A, were strong at the instant of choice, and I chose A. If this choice is utterly unconnected to anything leading up to it, then it is a simple case of giving in to an urge. It's quite possible that we are not free in doing that. However, if, like all such decisions in real life, it is embedded in much larger considerations which we have been upholding for some time, then rewinding to that exact instant reveals nothing. Whether the reasons were greater or lesser won't reveal much either. Why did I choose this word instead of another to express myself at this instant? The answer to that will have much to do with my education, my social background, etc, but what matters is that it was available and I was engaged in something bigger than that single word, so I used what was available. Is there a probability distribution or a set of determining factors over the voluntary use of that particular word at that moment? Maybe. But my freedom (and my moral responsibility) is in engaging in this conversation at all. This includes not just selecting words, but refraining from pursuing various side-points, trimming up the post to get rid of as much extraneous stuff as I can because I want Pruss to actually read it, etc etc. What is the probability distribution on doing THAT? I doubt any such thing could make sense. I doubt any sort of "rewinding" could make any sense in such a case. I don't even see how you could pinpoint when I "start" and "stop" having this conversation, since I'm thinking about it as I go about various other tasks.... Do you see why I think the whole paradigm in which your question (and van Inwagen's and Galen's) is posed may be the problem?

Michael Gonzalez said...

IanS: On the Copenhagen interpretation, we have a real problem as to how the detectors and laboratories and experimenters all seem to exist continuously despite being made of particles which are not at that moment being subjected to "an observation", whatever that is. I think Pruss is right to retain his non-Copenhagen intuitions! LOL. In any case, the probabilities on each experimental set-up are set facts which can be determined, no matter what interpretation of QM we're working with. That's why Many Worlds seems to be a dead-end, since it has no explanation of why we would observe these inviolable probability distributions.