Metaphysics » Lecture 11
There are various words and phrases we use in ascribing free action to people: … ‘acted freely’ and ‘did it of his own free will’… We learn these phrases by watching people apply them in concrete situations in everyday life, just as we learn, for example, colour words. These concrete situations serve as paradigms for the application of these words: the words mean things of that sort. Therefore they must apply … at least to the paradigmatic objects or situations. Careful investigation, philosophical or scientific, of these situations may indeed yield information about what freedom of choice really consists in, but it cannot show us that there is no such thing as freedom of choice. This is strictly parallel to the following proposition: careful investigation, philosophical or scientific, may show us what colour really consists in, but it cannot show us that there is no such thing as colour. (van Inwagen 1983: 107; cf. Flew 1955)
This is an argument that free will exists; how is it an argument for compatibilism?
So something more is needed to get from the premise that we correctly call some actions free to the compatibility of freedom and determinism.
Taking up this irrelevance idea: perhaps the needed premise is that esoteric scientific hypotheses, like determinism, could not possibly undermine our ordinary usage, based as it is on regular observation:
only … superficial [i.e., ordinarily observable] features of people and their acts … can be relevant to determining whether to apply the term ‘free’: one must be able effectively to compare some present act in which one is interested with the paradigms of free action … in order to see whether the present act is sufficiently similar … to be correctly called ‘free’. (van Inwagen 1983: 110)
This supplementary premise is motivated by the slogan that meaning is use: that the meanings of our expressions are fixed by the global pattern of usage; that differences in meaning can only arise when usage is different; and that usage can only be different when there is some difference language users can detect and respond to.
That green means what it does is fixed by the fact that most English speakers, most of the time, use it to label surfaces within the same fairly natural range of visible surface reflectances, and don’t use it to label other surfaces.
Likewise for free. Common usage applies free to some actions, in contrast to those which are coerced, involuntary, etc. Those uses depend on speakers deploying those terms on the basis of overt properties of what they are talking about.
We might thus attempt to run a paradigm case argument with this key premise:
This premise, coupled with the possibility of a deterministic world in which our actual usage is determined to occur, will entail that in that world determinism is true and there are actions that are correctly in world called free. Hence compatibilism is correct.
The possibility premise establishes that in a deterministic world, speakers could establish a convention in which free applies to some actions.
But we need (same usage \(⇒\) same meaning) to establish that their word free has the same meaning as our word. And that premise seems to be subject to counterexample.
Consider Twin Earth. On Twin Earth, there is no H2O, but there is an indistinguishable clear drinkable liquid called XYZ (Putnam 1973: 701). Speakers on Twin Earth use water to refer to XYZ, with the same pattern of usage as our uses of water to refer to H2O.
Once the facts about XYZ are revealed, it is correct to say
Since XYZ ≠ H2O, these two uses have different meanings, despite the same use (remember that use is based on observable features, and Twin Earth looks just like Earth from the point of view of an external observer of both).
It is usage plus environment which makes for meaning. The internal concepts that implement our usage aren’t the meanings, it is the product of that usage in that environment which fixes meaning.
Cut the pie any way you like, ‘meanings’ just ain’t in the head! (Putnam 1973: 704)
Van Inwagen gives an example where the usage of free is the same, but – as in the Twin Earth case – there is some hidden environmental variability:
The world in which (M) is true is only unobservably different from the actual world, and the patterns of usage are the same. So in the (M)-world the term free has an extension.
But in that world, the word free is only superficially like our word free. In fact, none of the actions called free in the (M)-world are in fact free by our standards.
Since compatibilism about free will and Martian control doesn’t follow from the existence of a non-empty extension for free in the (M)-world, likewise, compatibilism about about free will and determinism doesn’t follow from the existence of a non-empty extension for free in some deterministic world.
The internal principles and motives [of human nature] may operate in a uniform manner, notwithstanding these seeming irregularities; in the same manner as the winds, rain, clouds, and other variations of the weather are supposed to be governed by steady principles; though not easily discoverable by human sagacity and enquiry. (Hume 1777: §8, ¶15)
[L]iberty, when applied to voluntary actions … cannot surely mean, that actions have so little connexion with motives, inclinations, and circumstances, that one does not follow with a certain degree of uniformity from the other, and that one affords no inference by which we can conclude the existence of the other. (Hume 1777: §8, ¶23)
If indeterminism is true, our bodily actions are not fully explained by our ‘motives, inclinations, and circumstances’.
If our bodily actions are not fully explained by our motives, inclinations and circumstances, then those bodily actions are not freely chosen by us.
So, if indeterminism is true, our bodily actions are not freely chosen by us. (3, 4, logic)
If my acts are the result of undetermined events, then I have no control over those undetermined events. But how can I freely do what I have no control over?
Reply: this is to endorse the validity of rule (\(\beta\)); in which case, we must be endorsing the reasoning in the Third Formal Argument for incompatibilism.
Of course, rule (\(\beta\)) is invalid, so this can’t be a good ground to deny that probabilistically-caused acts can be free.
Anyone … can see right at the outset that a person whose ‘acts’ are the consequences of undetermined events … is not really an agent at all. Such a person is not acting, but is merely being pushed about or interfered with. (van Inwagen 1983: 134)
It should be unsurprising that the existence of quantum indeterminism, by itself, is inadequate to make the problem [of agency] disappear. It will certainly be insufficient … merely to replace a microphysically deterministic vision of the universe with a microphysically indeterministic one. For the problem … is as much about the way in which we suppose the different levels of reality relate one to another, as it is about the idea that each momentary state of the universe inexorably necessitates the next. … [Indeterminism] will not by itself supply the answer to the question how agency is possible. An answer to that question will require also an understanding of what could lead us to want to say that an organism rather than merely some part of one … has brought something about. (Steward 2012: 11)
Another sort of Humean compatibilism: analyse the ability-attributing can, or its past tense could have, by means of a conditional, something like this:
This is close to what Hume calls ‘hypothetical liberty’:
a power of acting or not acting, according to the determinations of the will; that is, if we choose to remain at rest, we may; if we choose to move, we also may. (Hume 1777: §8, ¶23)
No problem with determinism if this is correct: for even if it is determined that \(A\) won’t try to \(\phi\), still it could be that, in the nearest possibility where they did try, they would have succeeded (the laws and past would have been such as to permit them to try and to succeed).
Yet can/could have don’t act like conditionals on the standard Kratzerian approach (Kratzer 1977, 1981).
Moreover the proposal (COND) appears to be subject to direct counterexample:
Smith could have eaten one of the red candies.
This proposition is not equivalent to
If Smith had chosen to eat one of the red candies, then Smith would have eaten one of the red candies.
For suppose that Smith is pathologically afraid of the sight of blood, and the candies are the colour of blood. Then it may well be that Smith was unable to choose to eat one of the red candies. And, in that case, he could not have eaten one… Nevertheless … if he had chosen to eat one of the candies, he would have. (van Inwagen 1983: 115–16)
An instance of a general problem: Smith’s inability to eat the candy is here linked to his inability to bring himself to choose to do so. The conditional is true; but it doesn’t support the ability precisely because the ability consists in part of the agent’s liberty to make a certain choices. Since Smith can’t decide to try (not because of determinism, but because of his own pathology), he lacks the ability.
That abilities cannot be analysed in terms of conditionals only prompts the question: is there an account of abilities that is conformable to compatibilism?
Vihvelin (2013: ch. 6) offers an account that is inspired by the conditional analysis without being subject to its problems.
The conditional account of ability can be derived from two prior theses:
Put these together and we get the conditional analysis of ability (Vihvelin 2013: 196–97).
But it was long ago noted in the dispositions literature that SCA, the simple conditional analysis, fails (Johnston 1992; Martin 1994). Consider
a fragile glass that is carefully protected by packing material. It is claimed that the glass is disposed to break when struck [i.e., is fragile] but, if struck, it wouldn’t break thanks to the work of the packing material (Choi and Fara 2021: §1.2)
In this case, the disposition is present but masked; the SCA predicts the conditional will be true, but it is not.
There are also cases of the reverse sort, finks and mimics, in which the right sort of conditional is true but there is no disposition:
When a styrofoam dish is struck, it makes a distinctive sound. When the Hater of Styrofoam hears this sound, he comes and tears the dish apart by brute force. So, when the Hater is within earshot, styrofoam dishes are disposed to end up broken if struck. However, there is a certain direct and standard process whereby fragile things … break when struck, and the styrofoam dishes in the story are not at all disposed to undergo that process. (Lewis 1997: 153)
Vihvelin endorses the idea of Dispositional Abilities:
To have one of the narrow abilities in virtue of which we are agents with free will is to have some intrinsic disposition or bundle of intrinsic dispositions. (Vihvelin 2013: 175)
But Vihvelin rejects the simple conditional analysis. How then should we understand dispositions, and the abilities that are a special case of dispositions?
Vihvelin suggests that we should follow the try-conditional account but without a conditional:
For a highly interesting subset of our narrow abilities, to have the narrow ability to do \(X\) is to have an intrinsic disposition to do \(X\) in response to the stimulus of one’s trying to do \(X\). (Vihvelin 2013: 175)
Fara (2005) analyses the disposition ascription in terms of habitual constructions (like the habitual Peter sings in the shower, which seems to mean something like Normally/typically, Peter sings in the shower ):
‘\(N\) is disposed to \(M\) when \(C\)’ is true iff \(N\) has an intrinsic property in virtue of which it [habitually] \(M\)s when \(C\)’ (Fara 2005: 70)
Suppose someone – Black, let us say – wants Jones4 to perform a certain action. Black … waits until Jones4 is about to make up his mind what to do, and he does nothing unless … Jones4 is going to decide to do something other than what he [Black] wants him to do. If it does become clear that Jones4 is going to decide to do something else, Black takes effective steps to ensure that Jones4 … does do, what he wants him to do. Whatever Jones4’s initial preferences and inclinations, then, Black will have his way.
Now suppose that Black never has to show his hand because Jones4, for reasons of his own, decides to perform and does perform the very action Black wants him to perform. In that case, it seems clear, Jones4 will bear precisely the same moral responsibility for what he does as he would have borne if Black had not been ready to take steps to ensure that he do it. (Frankfurt 1969: 835–36; see also van Inwagen 1983: 162–63)
Indeed, we can link preemption cases and Frankfurt cases.
For if our action is guaranteed, then we could not have done otherwise, so we can reformulate the PAP as:
PAP2 is false whenever there are Frankfurt cases: cases in which someone’s decision caused their action, but their decision preempts an alternative cause that would have produced the same action, so the action is guaranteed.
But, as in the scenarios Frankfurt presents, one can still be morally and causally responsible for a guaranteed outcome.
Frankfurt notes than in the case of Jones4, the fact they he couldn’t have done otherwise ‘played no role at all in leading him to act as he did’ (Frankfurt 1969: 836), and this is important:
Suppose a person tells us that he did what he did because he was unable to do otherwise … We understand the person who offers the excuse to mean that he did what he did only because he was unable to do otherwise…. And we understand him to mean … that when he did what he did it was not because that was what he really wanted to do. The principle of alternate possibilities should thus be replaced … by the following principle: a person is not morally responsible for what he has done if he did it only because he could not have done otherwise. This principle does not appear to conflict with the view that moral responsibility is compatible with determinism. (Frankfurt 1969: 838, my emphasis)
The argument for the existence of free will fails if (PAP) is false, and this alternative premise doesn’t repair the damage – at least, not if we retain van Inwagen’s idea that free will is linked intimately to the ability to do otherwise than we in fact do.
The traditional kind of compatibilist attempts to argue that agents have abilities to do otherwise than they do, even under determinism (Hume 1777; Lewis 1981; Fara 2008; Vihvelin 2013).
Frankfurt cases undermine a link between responsibility and alternate possibilities.
One response to this is to focus directly on moral responsibility, rather than on free choice as a route to moral responsibility.
This is the approach of semicompatibilism, the view that
determinism is compatible with moral responsibility, apart from whether causal determinism eliminates access to alternative possibilities. (Fischer 2012: 255)
Semicompatibilism needn’t take a stand on the traditional issue of free will.
Someone who accepts the semicompatibilist idea that determinism and responsibility are compatible, and also accepts van Inwagen’s idea that freedom is necessary for responsibility, is led to endorse this conclusion: free will cannot require the ability to do otherwise.
What would such a new species of compatibilism look like? We look first to Frankfurt’s own discussion of his cases; if we can identify what Jones4’s responsibility depends on, then we might be able to identify the feature that grounds his freedom.
Our intuitive response to the Frankfurt case is that Jones4 is responsible because he did what he did because he wanted to. So, a first pass proposal:
A counterexample to FC1: the unwilling addict. Consider the addict who
hates his addiction and always struggles desperately, although to no avail, against its thrust. He tries everything that he thinks might enable him to overcome his desires for the drug. But these desires are too powerful for him to withstand, and invariably, in the end, they conquer him. He is an unwilling addict, helplessly violated by his own desires. (Frankfurt 1971: 12)
The addict acts on his desires, but these desires at at odds with what he would prefer his will to be. In that sense, he acts ‘helplessly’, but not freely.
If an agent’s preferences about their will are aligned with what they will and how they act, then we aren’t tempted to say the agent is the mere victim of their desires. In such a case, we think, the agent acts; they act from their desire; and their desire is one they fully endorse.
This gives a hierarchical picture of the will: someone is free when their action-guiding desires (‘volitions’) align with the volitions they want to have (their second-order volitions).
the statement that a person enjoys freedom of the will means (also roughly) that he [sic] is free to want what he wants to want. More precisely, it means that he is free to will what he wants to will, or to have the will he wants. Just as the question about the freedom of an agent’s action has to do with whether it is the action he wants to perform, so the question about the freedom of his will has to do with whether it is the will he wants to have.
It is in securing the conformity of his will to his second-order volitions, then, that a person exercises freedom of the will. And it is in the discrepancy between his will and his second-order volitions, or in his awareness that their coincidence is not his own doing but only a happy chance, that a person who does not have this freedom feels its lack. The unwilling addict’s will is not free. This is shown by the fact that it is not the will he wants. (Frankfurt 1971: 15)
It is worth looking more closely at PAP. It’s official formulation is ‘A person is morally responsible for what he has done only if he could have done otherwise’ (van Inwagen 1983: 162). This is a claim about the agent’s abilities.
Frankfurt’s cases are those in which there is no possibility in which the agent does otherwise: either they do it of their own volition, or the intervention causes them to do it.
These cases are a counterexample to PAP, in van Inwagen’s formulation, only if something like the following principle linking abilities and doings is correct:
Where does this leave us in the debate over free will? Two important points:
Van Inwagen takes a different response to Frankfurt cases. He accepts that PAP might be false; but (in line with the Frankfurtian compatibilist) continues to accept that moral responsibility requires free will.
Let us suppose that the Principle of Alternate Possibilities is indeed false. What follows? It does not follow that we might be morally responsible for our acts even if we lacked free will; it follows only that the usual argument for the proposition that moral responsibility entails free will has a false premiss. But might there not be other arguments for this conclusion? Might there not be other premisses from which this entailment could be derived? (van Inwagen 1983: 164)
The PPA is not susceptible to Frankfurt-style counterexample.
Suppose I witness a bank robbery in progress, and fail to call the police then. It’s also true that my phone – unbeknownst to me – won’t work, because the robbers have sabotaged all the local phone towers.
Am I responsible for that omission? Perhaps I am responsible for for not trying to call, but
Am I responsible for failing to call the police? Of course not. I couldn’t have called them. (van Inwagen 1983: 166)
Van Inwagen thinks that we also need to pay attention to the events and states of affairs involved in action.
For van Inwagen ‘no event could have had causes other than its actual causes’ (van Inwagen 1983: 170). This is a fine-grained theory of event identity.
Accordingly, there will be no Frankfurt-style counterexamples to (PPP1), because the cases where Black intervenes will feature slightly different causes, and hence slightly different events are involved: so Jones could have prevented \(e\), even though an event very similar to \(e\) would occur regardless.
Again, van Inwagen argues, there are no Frankfurt-style counterexamples here: if a certain state of affairs obtains ‘no matter what’, then the agent is not responsible for it:
because this state of affairs is a universal, it can be reached by various causal roads, some of them differing radically from the road that is in fact taken; and, in the cases we have imagined, every causal road that any choice of the agent’s might set him upon leads to this same state of affairs. This is why the agent … always turns out not to be responsible for the state of affairs he is unable to prevent. (van Inwagen 1983: 176)
If (i) no one is morally responsible for having failed to perform any act, and (ii) no one is morally responsible for any event, and (iii) no one is morally responsible for any state of affairs, then there is no such thing as moral responsibility (van Inwagen 1983: 181)
If (i) someone could have performed … some act he did not in fact perform, or (ii) someone could have prevented some event that in fact occurred, or (iii) someone could have prevented some state of affairs that in fact obtains, then the free will thesis is true. (van Inwagen 1983: 182)