1 The Case for the Mystery of Metaphysical Freedom Stated

According to Peter van Inwagen human beings possess, with respect to some of their actions, metaphysical freedom. Someone possesses metaphysical freedom with respect to action A at time t, provided that person is able to perform A at t and able to refrain from performing A at t. Alice has such freedom with respect to the act of her rising from her chair if she is able to rise and able to refrain from doing so. Having metaphysical freedom with respect to A at t, is having a choice about whether or not to perform A at t.Footnote 1

In a number of places Van Inwagen has argued that metaphysical freedom is a mystery.Footnote 2 It is a mystery that he cannot dispel and that he is certain that no one else has dispelled. He brings out the mystery by considering the following three theses, all of which appear to him to be true:

  1. (1)

    Either determinism is true or indeterminism is true.

  2. (2)

    Determinism is incompatible with metaphysical freedom.

  3. (3)

    Indeterminism is incompatible with metaphysical freedom.

From these it follows that.

(C) There cannot be metaphysical freedom.

There can be no metaphysical freedom because metaphysical freedom is a contradiction in terms. It is an inner impossibility much as a square circle is, or a liquid wine bottle. Let us call the argument from (1), (2), and (3) to (C) the Impossibility Argument.

The mystery for Van Inwagen is this: the Impossibility Argument seems compelling to him, yet he nonetheless believes that humans do have metaphysical freedom, and hence that the conclusion of the Impossibility Argument, (C), is false. He is thus committed to holding that at least one of the premises of the Impossibility Argument is false. But, “having thought very hard [about the Impossibility Argument, RvW] for almost thirty years”, he confesses he is unable to decide which, he is, as he says, “absolutely clueless”.Footnote 3 This is Van Inwagen’s mystery. Van Inwagen reports that it is his opinion that (2) is true and (3) is false (and he holds that (1) is true too). But he has no clue. Let us call this line of thought the Mystery Argument. Somewhat more formally it can be put as follows:

Mystery Argument

(5) The Impossibility Argument is logically valid.

(6) The conclusion of the Impossibility Argument seems false, for humans do have metaphysical freedom.

(C*) Hence, at least one of the premises of the Impossibility Argument must be false.

(7) The premises of the Impossibility Argument all seem true.

(8) If a logically valid argument has a seemingly false conclusion that follows from premises that all seem true, we face a mystery.

(C**) Hence, the Impossibility Argument presents us with a mystery.

In this paper I shall argue that the Impossibility Argument does not present us with a mystery—in other words, I shall argue that the conclusion of the Mystery Argument, (C**) is false.

2 The Case for the Mystery Analysed

To see how Van Inwagen argues for the mystery of metaphysical freedom, we must have before us his definition of Determinism:

Determinism = df. the thesis that the laws of nature and the way the world was millions of years ago together determine one unique future.

So Determinism tells us that if the particles of matter that the world consists of were, so to speak, “rolled back” to the positions they occupied in May 1945, history would, as it were, “repeat” itself. There would “again” be the cold war, the assassination of John F. Kennedy in 1963, and the fall of the Berlin Wall in 1989.

Indeterminism is the denial of determinism. It is the thesis that it isn’t the case that the laws of nature and the way the world was millions of years ago together determine one unique future. So Indeterminism tells us that if the world were to be “rolled back” to May 1945, then possibly there would be no cold war, no assassination of Kennedy in 1963 and no fall of the Berlin Wall in 1989.

Let us, given these definitions, consider the Impossibility Argument more closely. The argument is deductively valid: if the premises are true, then so must be the conclusion. If one is to escape the conclusion, one must successfully argue against the truth one of the premises. Let us therefore consider (1), (2), and (3) respectively.

Premise (1), “Either Determinism is true, or Indeterminism is true”, has much to be said for it. It is an instantiation of a logical truth: either p or not-p. It must be noted, however, as Strawson has argued, that certain what seem to be instantiations of this logical truth, are false. Take for example proposition Q: “Henry has stopped beating his wife”. If Henry has never punched his wife, for instance because he has no wife, or has one but has never beaten her, then “either Q or not-Q” is false. Strawson’s diagnosis for why it is false, would be that Q has a false presupposition. His general point was that if a proposition p is to be such that “either p or not-p” is to be true of it, p should not contain a false presupposition.Footnote 4 One way one could try to argue against (1), then, is to explore whether “Determinism is true” contains a false presupposition. But this doesn’t look promising. For Determinism’s presuppositions, viz. “There are laws of nature”, and “The world is millions of years old”, are unobjectionable. Like Van Inwagen, I shall take (1) to be true.

Premise (2), the thesis that metaphysical freedom and determinism are incompatible, seems true as well. For if the laws of nature and the way the world was millions of years ago determine one unique future, human beings can do only what they in fact do (and could do only what they in fact did, and will do only what they in fact will do). For, recall, to have metaphysical freedom is to be able to do what it is in one’s power not to do, and to be able to refrain from doing what it is in one’s power to do. Van Inwagen’s Consequence Argument, in one of is versions, is presented and widely considered to be a strong argument in support of (2).Footnote 5 I join in.

Premise (3), the thesis that metaphysical freedom is incompatible with Indeterminism looks true to Van Inwagen as well. He considers an argument that seems sound to him, although, as indicated earlier on, it is his opinion that there is something wrong with it. The following passage contains the core of the argument for (3):

Does postulating or asserting that the laws of nature are indeterministic provide any comfort to those who would like to believe in metaphysical freedom? If the laws are indeterministic, then more than one future is indeed consistent with those laws and the actual past and present - but how can anyone have any choice about which of these futures becomes actual? Isn’t it just a matter of chance which becomes actual? If God were to “return” an indeterministic world to precisely its state at some time in the past, and then let the world go forward again, things might indeed happen differently the “second” time. But then, if the world is indeterministic, isn’t it just a matter of chance how things did happen in the one, actual course of events? And if what we do is just a matter of chance - well, who would want to call that freedom?Footnote 6

The argument, then, is that if the world, or the laws of nature, are indeterministic, then what humans do is just a matter of chance, not a matter of freedom. To be more precise, the argument involves two conditionals, and both should be true if the Mystery Argument is to go through:

Conditional 1: If the laws of nature are indeterministic, then what we call ‘human actions’ are chance events.

Conditional 2: If what we call ‘human actions’ are chance events, then what we call ‘human actions’ aren’t free actions.

“Chance”, however, is a multiply ambiguous notion, and the Mystery Argument cannot be adequately discussed without identifying the notion of “chance” that is wielded.Footnote 7 This paper presents a number of different notions of chance and discusses, for each of them, whether Conditionals 1 and 2 are true. Van Inwagen assumes that there is a notion of chance such that Conditionals 1 and 2 come out true, or seem to come out true. (Seem, for, as we saw, it is his opinion that (3) is false.) My argument will be that it is unclear what that notion is, and even whether there is such a notion. And hence that Van Inwagen’s case for the mysteriousness of metaphysical freedom is not really made.

3 Why the Case for the Mysteriousness of Metaphysical Freedom is not Made

The Mystery Argument goes through only if Conditionals 1 and 2 are true. But what are the truth conditions of conditionals? Putting to one side the non-propositional view according to which conditionals never have a truth value, I will evaluate Conditionals 1 and 2 on the basis of the material conditional account, as well as on the basis Igor Douven’s inferentialist account.Footnote 8 The material conditional account says that conditional “if P then Q” is false only when P is true and Q false. In all other cases it is true. This account is very popular, even though it faces the paradoxes of material implication.

In part motivated by the wish to evade these paradoxes, Igor Douven has proposed that “a conditional is true in a given context iff the consequent follows via a number of inferential steps from the antecedent, possibly in conjunction with contextually given background knowledge, where, first, the steps are all valid in a deductive, inductive, or abductive sense; second, the consequent doesn’t follow … from the background knowledge alone; and third, the antecedent is (deductively) consistent with the background knowledge.”Footnote 9 Douven’s semantics for conditionals is motivated by the intuitively plausible idea that what is required for the truth of a conditional is that there is a connection that links antecedent and consequent. Since the conditional “If the moon is made of cheese, then Amsterdam is the capital of the Netherlands”, which comes out true on the material conditional account, is not such that the consequent follows, not even given background knowledge, from the antecedent, it doesn’t come out true on Douven’s account. On his account “If Millie misses her bus, she will be late for the movies” does come out true, as “She will be late for the movies” follows via a small number of inferential steps from “Millie misses her bus”. The inferential steps in this case aren’t deductive, but inductive. The steps could be spelled out as follows: “Millie misses her bus”; “Her bus takes between 25 and 35 minutes to get to the theatre”; “The next bus departs 10 minutes later”; “The movie starts in about 40 minutes”; “Millie will be late for the movie”.

I will now introduce four different notions of “chance” and feed Conditionals 1 and 2 with them. I will then evaluate the ensuing Conditionals first on the basis of the material conditional account of conditionals, and next on the basis of Douven’s inferentialist account.

3.1 Unpredictable

The first notion of “chance” that I introduce says that an event is a chance event provided we didn’t and perhaps couldn’t know in advance whether or not its occurrence would take place. Its occurrence is or was unpredictable. Throwing “six” with a fair die is a chance event in this sense: prior to throwing we could not know that “six” would come up. This is an epistemic notion of chance, as it concerns what we do, or rather don’t know. If we feed the Conditionals with this notions, we get:

Conditional 1a: If the laws of nature are indeterministic, then human actions are unpredictable.Footnote 10

Conditional 2a: If human actions are unpredictable, then they aren’t free.

If we apply, in the interest of their evaluation, the material conditional account to these Conditionals, 1a seems to come out true: for the antecedent (“the laws of nature are indeterministic”) is true (or at least, for the purposes of argument we suppose it to be true) and so is the consequent (“human actions are unpredictable”, or at least some of them are). Since the consequent of 1a is identical to the antecedent of 2a, this means that the antecedent of 2a is true too. But what about 2a’s consequent: is it true, is “Human actions aren’t free” true? Van Inwagen, we saw, believes that it is false, and so do libertarians, as well as many compatibilists. But then 2a is false! Since the Mystery Argument goes through only if both Conditionals are true, and on the material conditional account one of them, 2a, is false, the Mystery Argument does not go through. Conclusion: this notion of chance, on the material conditional account, doesn’t support the Mystery Argument.

If we, still in the interest of their evaluation, now apply Douven’s inferentialist semantics to the Conditionals, we must check whether the consequents, given relevant background knowledge, follow from the antecedents in a number of argumentative steps (either deductively, inductively or abductively). A propos 1a we should thus ask: does “Human actions are unpredictable events” follow from “The laws of nature are indeterministic”? I don’t see any obvious argument that connects antecedent and consequent. Any argument that does the trick must contain non obvious and perhaps even spurious bridging propositions, such as “Human actions are law like.” But I see no reason to accept this bridging proposition, nor has Van Inwagen provided one. A propos 2a we should ask whether there is an argument from “Human actions are unpredictable” to the conclusion “Human actions aren’t free”? I must say, again I can’t see any plausible argument, and Van Inwagen surely hasn’t provided one. As a matter of fact, it seems plausible to argue from the antecedent to the denial of the consequent, i.e. to argue from “Human actions are unpredictable” to “They are free actions”, certainly if we have as background knowledge the thought that free actions aren’t determined by the laws of nature and the state of the cosmos long ago. The conclusion, then, is: this notion of chance, on Douven semantics, does not support the Mystery Argument either.

So, if ‘chance’ means ‘unpredictable’, neither material conditional, nor Douven semantics clearly supports the truth of the Conditionals and hence fails to support the Mystery Argument. But perhaps it was unlikely that Van Inwagen had this epistemic notion of chance in mind in the first place when he constructed his mystery argument. After all, as he thinks of it, it is something about the world that makes human actions chance events—not something about what we know (or rather don’t know) about the world. Let us therefore look at various metaphysical notions of chance.

3.2 Lacking a Determining Cause

The first I propose to consider is this: An event is a chance-event when it has no determining cause. Certain quantum events, such as the quantum jump of an individual particle, are often held to be chance-events in this sense: they have no determining causes, i.e. there are no necessary and sufficient conditions for their occurrence. The rising of the mercury column in a thermometer, by contrast, does have a determining cause, for there are necessary and sufficient conditions for its rising. Inserting this notion of chance into the Conditionals, we get

Conditional 1b: If the laws of nature are indeterministic, then human actions have no determining cause.

Conditional 2b: If human actions have no determining cause, then their actions aren’t free.

On the material conditional account 1b comes out true. For we are again, for the sake of argument, supposing that the antecedent (“The laws of nature are indeterministic”) is true. Its consequent (“Human actions have no determining cause”) is true too—or at least so libertarians tell us, and Van Inwagen agrees, witness his opinion that (2) is true. But on the material conditional account 2b comes out false. For its antecedent is true, and its consequent (“Human actions aren’t free”—or no action is free) is false. Van Inwagen will agree that 2b’s consequent is false, witness his belief in metaphysical freedom. Conclusion: with this notion of chance inserted in the Conditionals, the material conditional account does not support the Mystery Argument.

Douven’s inferentialist semantics asks us to see whether 1b’s consequent can be inferred from its antecedent. So, can “Human actions have no determining causes” be derived, deductively, inductively or abductively, from “The laws of nature are indeterministic”? This is evidently a difficult matter, but I confess to not seeing how the former can be deduced from the latter (in conjunction with background knowledge). For what exactly is the relation between laws of nature and human actions? Is it that human actions are fully “governed” by the laws of nature, in the sense that our actions have exclusively causes of the sort that physics or brain scientists study. That we have to reckon with the laws of nature when we set out to act, is blatantly clear. But this doesn’t mean that human actions are fully “governed” by the laws of nature. As a matter of fact, it seems highly implausible that human actions are fully governed by the laws of nature. For many human actions are occasioned by such things as plans, purposes, intentions, and promises—all of which are items that don’t figure in the laws of nature.Footnote 11 For a similar reason I do not see inductive support for “Human actions have no determining causes” emerging from “The laws of nature are indeterministic”. I therefore aver that there is no strong support for 1b. With respect to 2b the situation is more or less the same. For it seems unlikely that “Human actions aren’t free” can be derived, deductively or inductively or abductively, from “Human actions have no determining cause”. That there are no necessary and sufficient conditions for a person’s action A, doesn’t mean that A wasn’t free. Certainly libertarians will agree with me on this. It would even seem more plausible that the denial of 2b’s consequent (“Human actions are free”) can be derived from 2b’s antecedent (“The laws of nature are indeterministic”). I conclude that if “chance” means “without determining cause”, and we apply Douven’s inferentialist semantics, there is little support for Conditionals 1 and 2. And hence little support for the Mystery Argument.

3.3 Lacking Agential Control

Another metaphysical notion of chance is this: events such as a person’s bodily movements and other actions are chancy provided that that person lacks control over them. Erratic tics are usually held to be chance-events in this sense: the tics take place, but the sufferer has no control over them, he could not stop the tics to take place. An event is not chancy in this sense provided an agent has control over it. Your switching on the light in ordinary situations is an example of this; for in those situations you could have refrained from switching it on, meaning that you have control over it. Inserting this notion of chance in the two Conditionals we get.

Conditional 1c: If the laws of nature are indeterministic, then humans have no control over their actions.

Conditional 2c: If humans have no control over their actions, then those aren’t free.

On the material conditional account 1c comes out false. As before, we assume 1c’s antecedent to be true. But humans do have control over their actions (certainly libertarians will say this, and certain compatibilists as wellFootnote 12), hence 1c’s consequent is false. Since 1c’s consequent is 2c’s antecedent, 2c’s antecedent is false, and hence 2c true. Conclusion: this notion of chance, on the material conditional account, lends no support to the Mystery Thesis.

Neither does Douven’s semantics. For can we derive “Humans have no control over their actions” from “The laws of nature are indeterministic”? I find it very difficult to see how we can. The reason is (again) that the relation between laws of nature and human actions is opaque. Or perhaps that is too pessimistic a view of the situation. For Ginet has argued that an uncaused act can be under an agent’s control.Footnote 13 For this reason I am reluctant to evaluate 1c as true. 2c is true, as we can we derive “Human actions aren’t free” from “Humans have no control over their actions”. If a person has no control over her action A, she cannot be said to be free in performing it. The conclusion must be: Douven’s semantics gives no clear support to the Mystery Argument if we take “chance” in the sense of absence of control.

3.4 Not Intended by an Agent

There is a final metaphysical notion of chance that I should like to present: an event is chancy when it is not intended by an agent. When Richard Dawkins says that the emergence of human being in evolutionary history is due to chance, he uses “chance” in just this sense: no God or anybody else intended humans to emerge in evolutionary history. And when we look back to the world’s history we seem to find many things that were not intended, i.e. not the working-out of anybody’s intention—they are what Aristotle called coincidences. That Shakespeare and Cervantes died on the same day, is an example of this, and so is the fact that the weather conditions during the Dunkirk operation were favourable for the Allied forces; but since this was not intended by any human, it is a chance event in this sense under consideration. If we plug this notion of chance into the Conditionals, we get.

Conditional 1d: If the laws of nature are indeterministic, then human actions are unintended.

Conditional 2d: If human actions are unintended, then they aren’t free.

On the material conditional account 1d is false. It has a true antecedent (again this is assumed for the purposes of argument) but a false consequent, for many human actions are intended, they are the working-out of intentions, they aren’t happenstances, flukes or coincidences. 2d’s antecedent is hence false too, as is its consequent, hence Conditional 2d true. Conclusion: ‘chance’ conceived of as ‘unintended’, on the material conditional account, doesn’t support the Mystery Argument.

Turning to Douven’s inferentialist semantics we should ask whether we can derive “Human actions are unintended” from “The laws of nature are indeterministic”. I don’t see how that could be done, as the relation between laws of nature and actions, as indicated before, is opaque. On the face of it there is not even the glimmering of a reason for thinking that “Human actions are unintended” could be derived, somehow, from “The laws of nature are indeterministic”. So I see no reason for holding that 1d is true. As to 2d: can we derive “Human actions aren’t free” from “Human actions are unintended”? That is problematic. It would seem that certain unintended actions, such as inadvertently knocking over a glass, can still be free actions, as the agent could have not knocked over the glass, and as nothing forced him to knock it over. This means that in its generality 2d is false. Conclusion: ‘chance’ conceived of as ‘unintended’, given Douven’s semantics, doesn’t support the Mystery Argument.

4 Conclusion

By way of conclusion, then, I have first argued that the Mystery Argument, the conclusion of which is that metaphysical freedom is a mystery, involves two conditionals that both should be true if the Mystery Argument is to go through:

Conditional 1: If the laws of nature are indeterministic, then what we call ‘human actions’ are chance events.

Conditional 2: If what we call ‘human actions’ are chance events, then what we call ‘human actions’ aren’t free actions.

Next I have presented two semantics for conditionals: the traditional material conditional account, and Douven’s inferentialist account.

Then I introduced four different notions of chance, or four different understandings of “X is a chance event”, viz. a) X is unpredictable, b) X lacks a determining cause, c) X is outside agential control, and d) X is not intended by an agent.

Squeezing these four different notions of chance into Conditionals 1 and 2, and evaluating them by the standards of the material conditional account of conditionals, as well as by Douven’s inferentialist semantics, led to the following conclusion: Conditionals 1 and 2 never come out as both true.

Hence the Mystery Argument doesn’t go through.

Hence Van Inwagen hasn’t really presented us with a case for the mysteriousness of metaphysical freedom.