Blog Coordinator

« The Summer of Sixty Nine | Main | NOWAR 3, CFA »



Feed You can follow this conversation by subscribing to the comment feed for this post.

No wonder I like you so much Joe--you always try to cut the chase to some basic point.

Any CA is as I take it:

(i) A deductive argument that thus poses a proof of its conclusion.

(ii) An argument that assumes determinism in some way as to produce a logical consequence of the singularity of the future relative to the present and past in terms of events or states of affairs.

(iii) An argument that involves some principle of unavoidability or inevitability attached to the deductive rules applied to the argument resulting in restricting any account of agency to what is logically concluded.

(iv) A tacit assumption that (iii) itself entails an inadequate account of agency. This is a contextual assumption of the relevance of presenting the argument to settling much larger disagreements about matters of agency.

That's the best I can do full of tryptophan and chardonnay.

Happy T-Day everyone at Flickers.

I'm not sure I understand (iv) but let me think about it.

What if we rephrased (ii) as:

(ii') An argument for the incompatibility between the thesis of determinism and the free will thesis.

Is this too broad? (ii) suggests that all versions of the consequence argument incorporate van Inwagen's definition of determinism but I'm not sure about that.

I also wonder about (iii). Take Galen Strawson's basic argument. There are 2 plausible readings:
1. It is an argument for the incompatibility between determinism and moral responsibility (technically speaking, these should both be characterized as theses).
2. It is an argument for the incompatibility between determinism and ultimacy, where ultimacy is a kind of freedom.

Now given the language of Strawson's argument, it is pretty clear that he wants to establish what I'll call "broad incompatibilism," the view that determinism is incompatible with moral responsibility and thus establishing (1) is his aim. But he establishes (1), I think, by establishing (2).

Even if I'm wrong about this interpretation, one could offer an argument that is similar to Strawson's where the aim is to argue for the incompatibility of determinism and ultimacy. Would that argument be a version of the consequence argument, and if not, why not?

Hi Joe--

I'd still defend (iv) on the basis of the use of phraseology like "no one has a choice about. . .., no one can falsify. . ." and the like. These are alternatives (plural) locutions that imply incompatibilist norms about agency. The conflict of the consequence with something incompatibilist has to be in there to be an instance of CA.

My (ii) was formed in trying to sweep together various ways of getting determinism into the act, being mindful of Ginet as PvI's precursor. "Therefore, no one has a choice, etc." seems to set up metaphysically closed states of affairs, though the closure of singular local future events relative to local past events would accomplish much the same to set up a conflict with incompatibilist values (explicitly or covertly folded into the argument).

I take Strawson to be best construed as arguing a conditional thesis, that is, if regulative-freedom incompatibilism (either as responsibility or control) is true then the truth of determinism entails that we have no FW. Now that very conditional statement might be a further entailment granting the soundness of any CA, but I think it is the right way to understand the direct conclusion of the BA. So I take CAs to try and establish incompatibilism as part of the conclusion, and Strawson to conclude a conditional thesis if one grants incompatibilism.

Of course, I could be wrong. But since Strawson concludes things about incompatibilism, and CAs attempt (I think unsuccessfully) to establish incompatibilism, that was part of what I was trying to get across.

Thanks for the question, Joe.
The name "CA" comes from PvI.
At least he used the name to refer to an argument (or family of arguments) for the incompatibility of causal determinism and freedom to do otherwise. In particular, the argument employs the following ingredients: the fixity of the past, the fixity of the laws, and a modal transfer principle (which PvI called "Beta" and I call "The Principle of Transfer of Powerlessness". In my view, one can construct a valid and arguably sound version of the CA without explicitly or implicitly employing the modal transfer principle. One can for examplse use Ginet's principle: Our freedom is the power to add to the given past, holding the laws fixed.

Thanks so much to both of you for your helpful comments. Still, I'm not certain. For instance, here is an often quoted version of the consequence argument:

"If determinism is true, then our acts are the consequences of the laws of nature and events in the remote past. But it is not up to us what went on before we were born, and neither is it up to us what the laws of nature are. Therefore, the consequences of these things (including our present acts) are not up to us." (van Inwagen 1983, 16)

Here the language is "up to us," which could be interpreted in a classical (alternative possibilities) way but could also be interpreted in a source way.

What do we say about this? Do we get two different arguments from this passage, one a version of the consequence argument and another something like Strawson's argument, or is there just one argument?

I'm really just playing devil's advocate since I'm convinced that usage among philosophers picks out the argument John describes. I'm just fishing around for similarities between the various arguments for incompatibilism.

Thanks Joe.
I think the logic of the CA (or its contents, I suppose) shows that PvI means by "up to us", "up to us whether or not...".

After all, PvI combines the result of the CA with his critique of FSCs to yield the ocnclusion that we are not morally responsible. That is, he takes it that the CA shows we are not free to do otherwise, if c.d. obtains. Then he thinks he's helped to establish that moral responsibility requires just such freedom to do otherwise. This structure of presentation seems to indicate that PvI inteprepets "up to us" as requiring alternative possibilities.

Further, he even later (int he book) introduces what he calls a "direct argument" for the incompatibility of casual determinism and moral responsibility, which he explicitly descries as not involving the additional step of "freedom to do otherwise".

All of this strongly suggests that PvI intends the CA to involve "up to usness" that implies freedom to choose/do otherwise.

Of course, one can think about whether a structurally similar argument (with "up to usness" interpreted in an actual-sequence way) might work. One might dub this an "Actual-Sequence CA". And, of course, as I mentioned above, there is the Direct Argument, which uses similar ingredients to the classical CA.

Good questions Joe. I'll just note something I've noted at Flickers before: This informal version of the CA is ambiguous about why determinism per se is supposed to have these consequences. If we leave the opening phrase (with determinism) in, then we have to add something to make it work, such as "then our acts are the *logical* consequences of..." which seems problematic, in part for the reasons Charles suggested in his early posts. Or "then our acts are the *sufficiently caused* consequences of..." (which I think is false given the most plausible theories of causation, because the state of the million-light-year-wide cross-section of spacetime a million years ago is not a sufficient cause of my actions, nor is the big bang, regardless of whether determinism is true). In any case, these readings have to add some modifier to the 'consequences'.

Conversely, we could leave out the opening clause and its reference to determinism, and the rest of the paragraph remains just as (im)plausible. Once we put aside implausible versions of libertarianism (e.g., self-causation), then whether causal processes (and laws) are deterministic or indeterministic, our acts are the (eventual) consequences of prior events, going back to the remote past, in accord with the laws of nature. In *some* sense of 'up to' these laws and the remote past aren't up to us. And so, in that sense of 'up to' and if some transfer principle applies to that sense, then our present acts are not up to us. In this version of the informal argument, sourcehood worries arise, but I think they dissipate when one thinks more clearly about the relevant causal relations.

So my question is: What is the relevant sense of "consequence" in the CA?

Joe that's a really good point about the relevance of a sourcehood view here. Here's my take.

The basic point of incompatibilism is that something about FW just ain't gonna work if D is true. Obviously any form of indeterminism will do that automatically. Sourcehood worries might seem independent of that, but I wonder. Since sourcehood worries usually backtrack causal strings beyond the reach of plausible agency (Strawson 1), then it seems that the only way to break into that worry is to break the strings by introducing another process--the contradictory view of indeterminism--into the boundaries of agency. Not that that alone would serve to be sufficient for sourcehood, since the indeterministic process might be chaotic or uncontrolled (Strawson 2), but I'd say that kind of insufficiency is parasitic on there being an indeterministic process in the first place. (In fact I think Strawson ingeniously restates the classic dilemma of determinism, but that's for another post.)

Any other account of sufficient sourcehood would I'd think of logical necessity be compatible with D.

Why isn't "the state of the million-light-year-wide cross-section of spacetime a million years ago ... a sufficient cause of my actions"? I guess I could see how a Humean about causation could complain that the laws of nature themselves depend on the subsequent events, which PvI's phrasing (perhaps objectionably) sweeps under the rug. But on other views ... ?

When I hear "Consequence Argument," the argument that comes to mind is one that shows that determinism is incompatible with the unconditional power to do otherwise. It can take either the shape of PVI's version or the shape of Ginet's (as John points out in the thread above). I take it that these days (most, if not all) compatibilists agree that determinism rules out the unconditional ability to do otherwise. The issue is whether the conditional ability to do otherwise that is favored by compatibilists is *robust* enough to undergird free will and moral desert. Am I right about this? Are there compatibilists who think that the unconditional ability to do otherwise is compatible with determinism as well?

p.s. I also think that it is the incompatibility of determinism with the unconditional ability to do otherwise that drives the worries about sourcehood, ultimacy, and up-to-me-ness. For if I couldn't have done something different under the *actual* circumstances of my decision/action, *I* didn't play the right kind of causal role. Pointing out that had things been slightly different I could have done something slightly different (which is often cashed out in terms of what my counterpart(s) would have done) doesn't allay these worries. At least not in my case...

Thomas, I think Vivhelin counts:

I'm pretty sure Jenann Ismael ( - 53:52 – 1:38:51 of the Day1 video ) and Carl Hoefer (“Freedom From the Inside Out,” in Time, Reality and Experience, C. Callender (ed.), Cambridge: Cambridge University Press, pp. 201–222.) count.


"Are there compatibilists who think that the unconditional ability to do otherwise is compatible with determinism as well?"

Beebee and Mele's "Humean Compatibilism" (Mind, 2002) makes that claim for a form of dual-ability, complete with libertarian-like luck concerns.

Paul and Alan,

Thanks for giving me even more to read over the upcoming break!

Paul, I'm going to do the cheesy move of turning the question back on you (or others): In what context, whether scientific, ordinary, or metaphysical, and using what theory of causation, do we, or would we or should we, *ever* talk about the state of the million-light-year-wide cross-section of spacetime a million years ago as a sufficient cause of any event that has just occurred?

Thanks for the comments!

Eddy that is an interesting observation about the CA. Let me think about it before saying anything more. It reminds me of Hoefer's paper, mentioned by Paul, which I have only looked at quickly. That paper looks very interesting, as does Jenann Ismael's paper, which for some reason I can't access. (It keeps asking me to download Silverlight -- even after I downloaded it!)

Thomas: Can you clarify the distinction between unconditional and conditional abilities? If conditional accounts of ability are committed to the claim that ability statements are analyzed solely in terms of ability (the view we think Moore held), then there are not many contemporary philosophers who embrace conditional abilities. Thus, in as much as contemporary compatibilists think that there are abilities to do otherwise they are *technically* unconditional accounts of abilities, though conditionals often play a big role.

On the other hand, things are very different if we think of unconditional abilities as being governed by something like Fischer's extension principle (the name is from Haji 2009, I think): "an agent can do X only if his doing X can be an extension of the actual past, holding the laws fixed" (Fischer 1994, 88). I don't think there are any classical compatibilists who accept this principle, though contextualists might qualify.

Essentially, the compatibilists noted above all have responses to the consequence argument, so they resist the extension principle and try to explain the ability to do otherwise in some other way, sometimes tinkering with our intuitions about laws (Humeanism) and sometimes with our intuitions about time (Hoefer). Since these folks have (or think they have) responses to the consequence argument, they can claim that nothing is left out of their account of ability, there is no stronger notion available to only incompatibilists. In as much as a compatibilist account of ability is not a conditional account, it is an unconditional account of (compatibilist) ability.

In an ordinary (but far future) context, our descendants might set up a light-years-wide cross-section of space specifically in order to cause and control events within a conical space-time region. For any such event they could point to that setup as a sufficient cause. In a scientific context, relativity theory tells us that all causes are confined to the light-cone; and ex hypothesi, we have discovered some deterministic theory to be true, so we know the complete description of the past plus the laws implies the description of the present event. So the past events are sufficient. And we infer our scientific theories by the usual interventionist methods, so it seems fair to attribute causality.

Microsoft Silverlight is an application. Your browser might not automatically install it (browsers being paranoid about internet downloads is pretty justifiable). You might need to open your downloads folder and activate the installer manually. You might need to reboot after. It's all worth it, I swear.

I really like your last paragraph. Make "intuitions about time" into "about time and causality" though.

Building on Thomas's point: can we all agree that (a)something that some people have associated with free will (call it Eli*) is incompatible with determinism, and (b) that the CA does a nice job demonstrating why it is incompatible? And even though we can't agree about how to describe Eli*, we all understand what Eli* refers to and recognize that Eli* is the target of the CA. Can we agree on that?

If so, then setting aside the questions of whether Eli* is compatible with indeterminism, or whether Eli* is worth wanting, or even coherent, can we agree that the debate in recent years has shifted more towards two questions (with special focus on the second)?

(1) Does Eli* capture what we mean by free will?
(2) Do we need Eli* in order to deserve blame and praise for our actions?

Or has my instinctive aversion to metaphysics caught up with me, resulting in this much too simplistic way of understanding the debate?

And how is it possible that Eli* won two Super Bowls?

In response to your last question, Tamler, his first SB win was the result of someone catching a football with his head. There is a rarely used SB rule stipulating that if one of your players catches a pass with his head, then you automatically win the game. They just went through the motions from there on out since they had the network time anyway.

Good set of questions! This is in fact one reason why I’m exploring the expansion of the consequence argument. Even if we agree with John and Al that the consequence argument is an argument for classical (ability to do otherwise) incompatibilism, concerns about the similarity between it and other arguments remain. Consider the transfer principle from Pruss' argument:

Beta 2: From Np, p entails q deduce Nq (2013: 430).

According to Pruss, Nr is "the statement that r holds and that no one ever had, has or will have a choice whether r holds" (430). Perhaps this is suggestive of classical free will: “No one has a choice about r,” meaning “no one can render r false” (no abilities to do otherwise).

But you could just as easily take the expression to mean that r was not up to anyone, or that no one is the ultimate source of r. Thought of in this way, it is not clear to me that, say, Strawson's basic argument doesn't share important features of the consequence argument.

As Charles warned, one must be careful since there are principles like Beta 2 (Beta, for instance) that are invalid and others (Beta 2 if Pruss is correct) that are valid. In addition, different interpretations of “Nr” affect not only the validity of salient transfer principles but also the plausibility of the key premises about the past and the laws. The Humeanist’s reason for rejecting the fixity of the past is much more compelling if we interpret “Nr” as “r and r is not up to anyone.” N(laws of nature) is true by the very definition of "Humeanism" given this new interpretation of “Nr.” On Pruss’ original interpretation, Humeanism is at least harder to defend. Discovering commonalities between the arguments is helpful, especially if you're interested in understanding these arguments from a formal point of view.

In sum, we should care about the consequence argument because, to the extent that we think it is plausible, we might think that other arguments -- for source incompatibilism, or incompatibilism about moral responsibility, or free will skepticism, or skepticism about moral responsibility -- are equally plausible.

To be honest, I don't believe this story; I think the CA is much stronger than any comparable arguments for source incompatibilism, etc. but that itself, if true, would be an interesting result and well worth establishing.

I'm going to stop posting for awhile since I've got other work to do and I'm interested to hear what Peter Tse has to say. Feel free to post and I'll respond at a later date.

Eddy and Paul,
I don't understand how we could claim that the events in a region of space in an early stage of the universe cause my present actions unless we do so by assuming that causation is transitive. Yet, there seem to be a huge host of counterexamples to the transitivity of causation in the literature. Just a few of them:

McDermott gives the following case. A right handed man plans to detonate a bomb. The day before, his dog bites off his right forefinger. So, he pushes the button with his left forefinger. The dog bite caused him to push the button with his left forefinger. Pressing the button with his left forefinger causes the explosion. Yet, the dog bite did not cause the explosion.

Harty Field gives the following case: A man places a bomb under my desk which I find and survive. Placing the bomb was a cause of my finding it. My finding it was a cause of my continued survival. Yet, placing the bomb under my desk was not a cause of my continued survival.

Igal Kvat provides the following case: A mans finger is severed in a factory injury. The injury causes him to go to an expert surgeon who attaches the finger so perfectly that a year later it is fully functioning. The accident is a cause of the surgery. The surgery is a cause of the finger being fully functional a year latter. Yet, the accident is not a cause of the finger being fully functional a year later.

There are a host of other counterexamples to the transitivity of causation in the literature. Certainly, there are those who defend the transitivity of causation against these challenges. Yet, unless causation is transitive, I can think of no reason to believe that the state of the universe at an early point in time is a cause of my current actions. If arguments like the CA depend upon the transitivity of causation, they are significantly weaker than many people in the free will literature believe.

Tamler: I think the converse of Levy's Hard Luck explains the two SB rings. Over his brother? C'mon man!

As I mentioned earlier, incompatibilism must perforce be some view that conflicts with the determinism, such as outright indeterminism (metaphysically stochastic quantum processes or vanilla libertarianism), or something that plausibly is entailed by determinism (no final traceable source of agency, no control of events or states of affairs other than causal, etc.) that is held to be objectionable on some grounds. Indeterminist-based incompatibilism is very direct: the logical incompatibility is assured. But incompatibilist views that rely on something entailed by determinism must leverage values or intuitions against such entailments, and frequently are used indirectly to bash compatibilism in the process. Posit some distant manipulator as some source of agency or just constitutive bad luck but both manipulative by some direct mechanism of causality (note I'm excluding counterfactual manipulators), and the deterministic means appear to be the culprit of plausibly exculpating a supposed victim of causal manipulation. But this is just to use incompatibilist values--inconsistent with determinism by explicit or implicit assumption--to conclude that determinism cannot conform to such assumptions. At best that is a petitio--or perhaps another appearance of John's dialectical stalemate. (This isn't to say that luck "manipulations" are necessarily incompatibilist-motivated, but that at least such ideas as constitutive luck are either compatible with incompatibilism--does your head hurt like mine?--or they are just conceptually skeptical about FW tout court.)

Ok that's enough.


Transitivity would be sufficient to conclude that the light-years-wide region is a sufficient cause (hmm, that looks redundant, but it's not). But it's not necessary. Rather, we can just extrapolate from our storehouse of interventionist-style investigations of causation. We've seen the ways manipulations of matter and energy propagate well enough to see that (assuming determinism and relativity are correct) specifying all the properties of a large region will bring about a specific evolution of the forward diminishing light-cone bounded by that region. Which bringing-about we call causation. (Of course, the backward-in-time light cone is also guaranteed by such a specification, yet we don't call *that* direction "causation", for very interesting reasons! But alas, we should probably put that aside in this thread.) When I used the word "specifying" a few sentences back I didn't mean necessarily the action of an agent, but it could be. When I imagined a future civilization setting up a light-years-wide region in order to get a certain result, perhaps as some giant art performance, I expected they would succeed. How could they not (barring misunderstanding laws of nature)? They control all the knobs.

And controlling all the knobs (or just inanimately "specifying" them) is a salient difference from the thwarted-transitivity scenarios. If Hartry Field's bomber could control all the inputs to your survival, then (unless he's incompetent at the Inspector Clouseau level) you'd be dead.

It seems like the account you are giving would be forced to accept transitivity. Certainly, in a deterministic world, specifying the properties of a large region will determine that a particular later event occurs. Yet, this is a case of distal causation. Setting those conditions only causes that later event by causing a bunch of intermediate events. We certainly do know that setting these conditions would start a causal chain that terminates in the event. So, we can ensure that the event occurs by setting these conditions. Yet, whether these conditions caused the event depends upon whether causation is transitive. If not, the mere fact that it is the first step in a causal chain that results in the event is not sufficient for it to be a cause of the event. So, it seems like your comments about the light-cone assumes transitivity.

Yet, you also seem to accept an interventionist accounts of causation. It seems to me that the chief reason to deny that causation is transitive is precisely an attraction to interventionist accounts of causation. Interventionist accounts are merely counterfactual accounts that take more than whether the event occurs into consideration(how, when, etc.). Yet, transitivity horribly fails for counterfactuals. To the extent that one accepts interventionist accounts for causation one ought to be very skeptical about transitivity of causation and very worried about the assumption that the first step in a causal chain that deterministically brings about an event is a cause of that event. (There is a significant difference between the claim that a causal chain of events that terminates in some event is a cause of that event and claiming that the first event of that chain is a cause of the event.)


I didn't make my point clearly. There may well be causal chains in the light-years-wide-light-cone case. My point however is that epistemically, we don't need to bother about them. Take the Giant Cosmic Art Performance, whose Setup, S, at time t(S), caused its grand Finale, F, at time t(F). We know that S caused F *without* having to reason our way through a gazillion sub-conclusions about events at the intermediate times. Even if transitivity of certain select causal chains also falls out of our physics, it is more clear and obvious that "S caused F" falls out of our physics (on the given assumptions). Causal chains are something we figure out later, if at all.

All our evidence on causal relations so far comes from examples of distal causation. At least that's the way it looks viewed through the lens of modern physics.

The comments to this entry are closed.