Monday, August 6, 2007

Why Did the Consequence Argument Have Consequences?

There is an issue that has confused me for some time, ever since I first encountered, on The Garden of Forking Paths blog, the claim that the Consequence Argument (CA), as formulated by Peter van Inwagen, changed the landscape of the free will debate. I have since seen this claim repeated in print in several places, and the basic take seems to be this: before the CA was well known, most philosophers were compatibilists about free will. The CA, by formally showing the difficulties present in compatibilism, led to the resurgence of incompatibilist attempts to work out the free will problem. Since I first saw this claim, however, I have yet to figure out how, exactly, the CA argument could have been such a watershed in the free will debate. I do not mean here to criticize van Inwagen, but only to express my puzzlement. The puzzlement, essentially, comes down to this: what is it about the CA that could change the mind of anyone at all who already had a view on the free will problem?

I will not present the entire formalized version of the CA (nor will I try to outline all three versions of it), but here is, in simplified (though I hope not false) form, one version:

Def: P0 is a proposition that defines the complete physical state of the universe before the existence of any human beings, and P1 defines the complete physical state of the universe at any point after P0.

Def: Determinism is the thesis that P0, together with the laws of nature, necessarily entails P1. (That is, every physical state of the universe follows necessarily from any prior state, according to the laws of nature.)
No one has or ever had any choice about P0 or about the laws of nature.
Therefore, no one has or ever had any choice about any physical state of the universe, P1, whatsoever.
It follows that we cannot have any choice about our actions (which are physical events), since these all follow strictly from a prior state of the universe according to the laws of nature.
There are some problematic premises in the argument (and almost all of them have been questioned by someone or other), but essentially it seems to me to simply express, in one formalized version, the problem of free will and determinism. But since both compatibilism and incompatibilism are responses to that problem, I can’t see how someone could have been a compatibilist or an incompatibilist before learning of CA. That is, the argument seems to me to simply formalize just the issue that the very problem is about; anyone who has a take on the issue must, then, already have some sense of what the issue is. But then, why would a formalized version of the problem change someone’s mind?

Furthermore, I don’t see why this argument could have presented any real problems for classical compatibilist arguments; or, at least, I can’t see how it could have presented any problems that classical compatibilists were not already aware of and had not already tried to deal with. Here are at least two fairly obvious resources for dealing with it:

(1) It may well be true that nobody has any choice about past events and yet has a choice about their own actions, even if determinism is true. The past of P0 happened without the involvement of any agents. The actions of agents, on the other hand, have choices as constitutive features of those actions. The occurrence of an action in some propositions P1 then depends on the occurrence of a choice in a way that nothing in P0 does.

(2) Actions are events of a particular type. Following Davidson, we could say that actions are events that admit of a description under which they are intentional. An action, then, is an event that, on some description, has a particular kind of cause: a reason. But one can hold that reasons are not reducible to physical events. Thus, any action that occurs (will occur/has occurred) may be “inevitable” given P0 and the laws of nature. But the action is still up to the agent, since otherwise it would not be an action at all. That is, if an event is describable as following necessarily from P0 and the laws of nature, but it is not describable as following from a reason, then it is not an action. But no theory of free will that I am aware of insists that we must have a choice about events that are not actions.

Now both of these arguments may well have fatal flaws. What I don’t see, however, is how CA exacerbates those flaws, since the arguments, which predate van Inwagen’s formulation of CA, are already responses to the very problem CA formalizes. So what I cannot see is why someone who thought compatibilism was a good idea before they were aware of CA might think it was a bad idea after becoming aware of CA.


  1. """....every physical state of the universe follows necessarily from any prior state, according to the laws of nature.""""

    That premise should be questioned, I would say on somewhat Humean grounds, however general . It can be posited, and may be true in most cases, yet lacking unified scientific knowledge (consider the debates pertaining to quantum physics), we certainly aren't in the position to prove how things will work out in all cases, whether in terms of human or natural events.

    Consider the action of tectonic plates: were geology more reliable the experts could predict quakes, and the experts admit they can't do that. That doesn't defeat determinism, but doesn't it seem a bit trivial to say everything's determined (whether in terms of physics or psychology), but we can't trace the causal factors?

    I'd say that's the case with human psychology as well: lacking some complete account of behavior (conditioning, genetics, cognition, etc) it's rather premature to make a claim of determinism--especially given most humans' desires to approve "moral" and/or just behavior.

    The freedom/determinism issue reminds me of the case involving the SF politicians, where the one guy--White-- who killed Moscone offered the "Twinkie defense"---really an explanation of the act (and he was convicted of manslaughter, not murder). Many objected to that: they wanted to hold White responsible for murder (and for his actions), since White was more or less competent (rational, of sound mind, etc). Is that desire to hold people responsible just an illusion, misplaced, naive? Determinists would seem to say yes.

  2. In other words, isn't the determinist required to show the causal factors say for various human actions (or "agency" in general) to some degree, instead of just saying ala Kant "every event must have a cause?" (btw Heisenberg claimed quantum physics disproved Kant's a priori causality---).

    White's attorney, in a sense a determinist, offers the "Twinkie Defense"--not sure of all the details, but I sort of doubt that much research exists showing something like a causal relationship between high sugar intake and murder (or even violence)--even if a few people get wired after some donuts, etc. that doesn't turn them into Dan Whites on murder sprees (so it was more than likely BS, but jurors were against Moscone and Milk etc).

    "Twinkie Defense determinism" should thus be seen as a type of grand generalization. It's not even really "sufficient" as y'all say, but conjecture. Deterministic, naturalist arguments may work in some cases--say with DUIs (based on many studies showing that certain amounts of alcohol impair motor functions), or poverty leading to crime (of course not all poor people are criminals)--but I don't think philosophers usually make use of that sort of fact-based analysis.

  3. Ok. Let me unpack some of this.

    First of all, neither I nor (to my knowledge) any serious hard determinist (hard determinists argue that (1) determinism is true and (2) we lack both free will and moral responsibility) are out to stand up for the Twinkie defense. In any case, it is a bit of a red herring here, depending on what you mean by and what importance you place on "the desire to hold people responsible." Why should our desire, no matter how ingrained, dictate how we answer metaphysical questions such as that of free will?

    Second, it is not at all clear that dropping determinism is going to give us a clear path to moral responsibility. Since you mentioned Hume, I'll remind you that he originated a rather powerful argument to the effect that moral responsibility actually requires determinism; an undetermined event would be merely random, and how could anyone be responsible for a random event? A number of contemporary compatibilists have argued along these lines that no libertarian account--an account that rejects determinism--is going to be sufficient to ground moral responsibility. (Thus, some philosophers--like Thomas Nagel and Galen Strawson--have argued that moral responsibility is an incoherent concept regardless of whether or not determinism is true.)

    Third, the fact that we cannot make perfect predictions or trace complete causal paths for any event does not demonstrate that the events are not determined--you recognize this, but you seem to lean pretty heavily on our lack of knowledge of specific causal paths. (1) On the empirical side, one certainly need not trace a complete causal path in order to explain what caused an event--that would make it impossible to explain any event whatsoever without a complete model of the universe. But oddly enough this lack of perfect causal explanatory powers has now shut down the sciences for good. (2) On the a priori side, there are still questions about whether quantum physics disproved Kantian causality. How can an empirical event disprove an a priori claim? That we do understand events as necessarily caused seems to be an underlying reason for why nobody can really make sense of how quantum physics works; there's an equation, and the equation works, but how it works may be a question beyond what human reason can grasp.

    I'm a bit stupefied by the suggestion that determinists would have to be able to offer complete determining causes for every action in order to make their case. If we have good evidence that determinism is true, certainly one need not provide a complete account of how every particular is determined.

    But is determinism true? First, note that this is not really relevant to the Consequence Argument. CA is supposed only to show that free will is incompatible with determinism, not that determinism is true. Of course one of the major reasons that CA matters is that determinism is a widespread underlying assumption, and thus there is a problem of how to reconcile our responsibility-attributions with it. Some libertarians argue that determinism is in fact false, and quantum indeterminacy opens the path for free will. Others, however, point out that quantum indeterminacy only applies to micro-events and does not seem to effect--as far as anyone can tell--macro-events. If so, then the assumption of determinism remains a reasonable one.

  4. I did not reject determinism, but suggest a sort of voluntary determinism (and compatibilist, as y'all say). Few would deny strict determinism applies to cannon balls, to airplanes, to trolley cars. Yet human brains while physiologicaly organs, are not trolley cars, or even CPUs; so merely claiming some necessity to human actions and decisions, seems rather premature, and reductionist (say as in chess, or tests of various sorts, etc). At least from their own subjective perspective, most humans would agree they could have done differently. However trite or "libertarian" that seems, it's still a relevant objection to strict determinism. Playing chess for instance, you have a limited range of choices, but choices regardless, and obviously one can after a losing game analyze what went wrong: ah yes, I should have done that.

    I imagine a sophisticated behaviorist might say that's all conditioned (and determined) as like a rat learning to find the right passage in a maze (and getting his food pellet) is conditioned, but the chess-maze is quite more complex that the rat's, and the solution (ie a good move) requires a certain projective quality of thought and deliberation (dare we say modality of some sort) which rats lack. At the same time, I have not suggested that some anomalous quality of rationality to humans implies any ghost in the machine.

    Hume does seem somewhat persuasive on this issue: looking at how people behave one notes certain regularities, etc. But that's hardly empirical proof. Hume himself seems a bit contradictory on the causation issue as it applies to human actions: we can posit causal chains, but cannot easily infer them or prove them. The White/Moscoe case still seems relevant to that issue. A journalist and pal of White interviewed White while he was incarcerated, and found out White planned to kill other SF liberals, such as Willie Brown. This pal said White told him it was all premeditated. So the Twinkie defense was BS, according to the pal: and that's the case with many legal situations, and causation at this present stage of human psychology. There's no easy way to tell when all the facts are in; was White's crime due to bio-chemical causation (high sucrose intake), environmental stimulus, psych. problems, random factors, some long-harbored grudge, racism? To what degree? etc. Saying it was determined is really not saying much: and for that matter, the desire of Moscone and Milk's families and SF citizens to exact justice (even a certain retribution) would also be determined. So I would say a complete account of determinism as it applies to human actions depends upon a complete account of human cognition, which is probably decades, if not centuries away.