Transcendental Arguments for Freedom I
He who says that all things happen of necessity can hardly find fault with one who denies that all happens by necessity; for on his own theory this very argument is voiced by necessity. (Epicurus 1964: XL)
8.1 Three transcendental arguments
Why cannot the determinist find fault with one who opposes determinism? Epicurus does not expand upon what he offers us in the fragment quoted, but something is supposedly denied us by determinism. Such peritrope or ‘transcendental’ arguments seek to turn whatever this is against determinism itself.
Arguments like those of Epicurus are the concern of this and the final chapter. These are often represented as being against determinism, but they plausibly oppose many indeterminist views as well. These arguments would be better interpreted as being against any conception of determinism or indeterminism where such views are seen as opposing a certain, strong conception of freedom; or, more positively, as transcendental arguments for freedom. Determinism is believed by incompatibilists to deny us the freedom to choose our actions, and is believed even by compatibilists to deny us certain conceptions of choice. Structurally related transcendental arguments then emerge within several different fields: the conative, the ethical and the epistemic.
The conative transcendental argument
1. Necessarily, given a complete state of the world at some time in the remote past and all the laws of nature, then p, for any arbitrary p (Df: determinism).
2. We are powerless to avoid or alter the remote past or the laws of nature. (‘Powerless’ being a modal notion – meaning ‘it is impossible’ in some sense: we cannot do so, we are unable in some radical sense.)
3. We are powerless to avoid or alter p, for any arbitrary p.
4. It is futile to attempt to avoid or alter something that we are powerless to avoid or alter (and which we know we are powerless to avoid or alter).
5. Every executive, regulative, human activity, where this includes mental activity, can be taken as an attempt, by an agent, to avoid or alter something.
6. So, if determinism were true, the following would be futile: urging it; striving to understand it; attempting to derive or follow through any consequences of it; attempting make it function, cognitively, such that it was fit to count as knowledge; affecting to possess such knowledge so as then to profit from it; entertaining the bare possibility that it was nevertheless true (though pragmatically/rhetorically indefensible); pessimistically acknowledging its likely truth (whilst admitting it rendered that acknowledgement and everything else futile); and so on.
Of course, if it is not futile to urge futility, then something is not futile, and so any view that entails every human activity is futile is false.
Determinists (even fairly hard determinists/incompatibilists) typically do not see themselves as vulnerable to the conative transcendental argument because they do not see their position as leading to any thesis of futility or impotence. Many of their reasons for supposing this we have already evaluated in the previous chapter – chief among them the ‘co-fated’ and the ‘epistemically non-futile’ responses. Determinism maintains that externally characterizable natural law wholly fixes whether and what efforts may be made, but allows that these (wholly determined) efforts may sometimes bring about their intended outcome, and certainly always contribute to ‘the outcome’.
Which premise such compatibilist approaches should be seen as attacking is often rather arbitrary; they may be seen as differing in emphasis, but as threatening the thrust of the argument in any of several ways. It is the same, broad compatibilist tendency that motivates the response, whichever form it comes in – that is, whichever specific premise it is taken as undermining.
So, such compatibilists will oppose the consequence argument of 1–3, typically by opposing the modal transfer from 2 to 3: granted, we are powerless in one sense – the sense of premise 2 – to change the laws of nature or the Big Bang. However, we are not powerless in another sense, for we have the power to act on our determined desires – thus to bring about their intended outcome. Forms of these compatibilist approaches are sometimes phrased as rejecting or qualifying 2 itself (the ‘local miracle’ and ‘backtracking counterfactuals’ approaches).
Or, the same broad thrust can be framed more as being directed against premise 5: we aren’t trying to avoid or alter something (an ‘alternative possibility’) – we’re trying positively to bring something about, albeit something that is inevitably going to be brought about. This rather odd response was also encountered in several places in the preceding chapter – being considered in remarks concerning the point to our actions in addressing both the ‘co-destined’ and the ‘epistemically non-futile’ responses.
Or, as has been seen already in appraising the Lazy Argument, there are other approaches threatening premise 4 (and especially its parenthetic clause): it may be acknowledged that it is metaphysically futile to avoid something one is powerless to avoid or alter without it being epistemically futile – since we don’t know what it is that is inevitable.
Historically, of course, the Kantian development of this family of Epicurean arguments came after the development of soft determinism by Hobbes, Locke and Hume. Nevertheless, soft determinism – or at least some kind of ‘softening’ of determinism – is here needed by their successors as a response to such transcendental arguments. It is, then, important to note that this softening of determinism is not any concession on the part of the determinist, but much of what stands between determinism and the transcendental argument given.
The ethical transcendental argument
1. ‘Ought’ implies ‘can’: so, if one ought to perform action a, it must be within one’s power to perform action a. And if one ought not to perform action a′, it must be within one’s power to refrain from performing action a′.
2. Necessarily, given a complete state of the world at some time in the remote past and all the laws of nature, then p, for any arbitrary p (Df: determinism).
3. We are powerless to avoid or alter the remote past or the laws of nature.
4. We are powerless to avoid or alter p, for any arbitrary p.
5. We are powerless to perform any action a, that we do not perform, or to refrain from performing any action a′ that we do perform.
6. So, if determinism is true, it is, will be, and has always been, on all occasions false1 that one ought to perform any action one does not actually perform, or ought not to perform any action that one does perform. Whether by omission or commission, no-one can ever be, or has ever been, culpable, blameworthy, irresponsible.
7. Because determinism globally denies us the negative, ‘culpability’ aspect of any deontic moral distinctions, it removes from us the ability to distinguish and use the positive ‘commendability’ aspect of such distinctions. (If one affects to make no sense of anything being not red, one cannot distinguish and use the predicate red.)
8. If determinism were true, there could be no deontic, ‘oughts-based’ moral theory. (N.b. This as a totalizing thesis.)
In light of the common counter to these kinds of argument – that they beg the question against the determinist – it is significant that the reasoning to this stage is of a type that determinists (very many determinists, including soft determinists) have been concerned to urge themselves. Here is an example of the argument to 8, in Grünbaum’s opposition, on deterministic grounds, to the deontic (pure retributive) conception of punishment:
The humane determinist rejects as barbarous the primitive vengeful idea of retaliatory, retributive, or vindictive punishment. He condemns hurting a man simply because that man has hurt others. … [T]he decision whether pain is to be inflicted on the culprit, and, if so, to what extent, is governed solely by the conduciveness of such punishment to … reform and re-education. … The requirement that punishment can be expected to be reformatory does not itself specify what choice is to be made among two equally effective punishments of differing severity. But the use of the moral requirement that gratuitous suffering be avoided as a principle of justice does indeed make this choice unique. (Grünbaum 1971: 307, emphasis in original)2
So we are to take it that because no-one is free to decide, the criminal was not free to decide, and that therefore we should take our decision with the aim of reform and not retribution in mind. That because a criminal is not free to live up to any deontic requirements not to hurt others, therefore it is a requirement on us (as a ‘principle of justice’) that we minimize the unnecessary hurt he suffers. That because no-one has free choice, the criminal could not have chosen to commit these crimes, and therefore our choice should be made with care, lest we choose the ‘primitive’, ‘vengeful’ and ‘barbarous’ option of retributive punishment. In the event that we do choose this option, the humane determinist condemns us.
Now surely one can concede the following point (which may capture some part of the confused motivation behind such passages): in a world in which some acts are free and others determined, the free ones may be responsible or culpable as the case may be; the determined ones may not – hence at best perhaps subject to a consequentialist ethics. Thus, we ought not to take a smugly judgemental or vengeful attitude to crimes that may turn out, say, for reasons of psychological or criminological detail, not to warrant such a stance. But in making this concession, the last ‘ought’ has to be a moral ought, a deontic obligation upon us, which renders false the totalizing claim that nothing warrants deontic judgement.
The conclusion to the ethical transcendental argument then is:
9. If there were no deontic basis to morality at all, it could not be said of determinists that they ought to oppose deontology – say, for a consequentialist ethics – or that their opponents ought not to maintain their own, retributive, ethics. Determinists would have no basis for opposing those who rejected their morality – even on a whim – whether rejecting it for deontology or anything else.
Determinists must, however, be able to justify their position and oppose their opponents’ positions. The framework for such reflexive justification must be left in place – no metaphysics can be so powerful, so totalizing, as to undermine this.
The response to the ethical transcendental argument is in step with the soft determinist response to the conative transcendental argument: to soften determinism. Rather more than with the conative argument, a number of different avenues then open up – for instance, in terms of which of the numbered premises this approach rejects. At this stage, note only that like the response to the conative transcendental argument, this softening of determinism is not a mere concession, but a response to a counterargument that otherwise undermines determinism as such.
I think that each of the arguments just given represents a serious challenge to determinism; but we have been addressing epistemic issues in this work for a reason, and it is the following ‘indirect (standard) epistemic transcendental argument’ that I wish to proceed with. The indirect argument is so called (appropriating terminology from van Inwagen 1983) because it doesn’t argue that determinism removes epistemic justification directly; rather, determinism removes some conception of leeway – with this kind of leeway then held to be essential for justification. (A ‘direct’ epistemic transcendental argument is considered in Chapter 10.) That is, the indirect argument proceeds via two things: firstly, the aforementioned deontic leeway requirement, which here is OIC – though a quicker argument would be available were the stronger principle of alternative possibilities (PAP) to be employed. Secondly, it proceeds via the Consequence Argument, which maintains that the leeway necessary for that deontological epistemic justification is denied us by determinism.
The indirect (standard) epistemic transcendental argument
1. ‘Ought’ implies ‘can’: so, if one ought to argue, reflect, ratiocinate, generally: reason r, it must be within one’s power to reason r. And if one ought not to reason r′, it must be within one’s power not to reason r′.
2. Necessarily, given a complete state of the world at some time in the remote past and all the laws of nature, then p, for any arbitrary p (Df: determinism).
3. We are powerless to avoid or alter the remote past or the laws of nature.
4. We are powerless to avoid or alter p, for any arbitrary p.
5. We are powerless to argue, reflect, ratiocinate, generally: reason r, for any argument, course of reasoning or conclusion that we do not actually undertake or accept; and we are powerless to refrain from reasoning r′ for any argument, course of reasoning or conclusion that we do actually undertake or accept.
6. So, if determinism is true, it is, will be, and has always been, on all occasions false that anyone ought to argue, reason or conclude otherwise than they do. Whether by omission or commission, no-one can ever be or has ever been (deontically) irrational, unreasonable, epistemically unjustified, intellectually blameworthy.
7. Because determinism globally denies us the negative, ‘irrational’, ‘unjustified’ aspect of any internalist value terms, it removes from us the ability to distinguish and use the positive ‘rational’, ‘justified’ aspect of such terms. (If one affects to make no sense of anything being not red, one cannot distinguish and use the predicate red.)
8. If determinism were true, there could be no internalist, oughts-based, epistemic theory. (This as a totalizing thesis.)
9. But then, of no argument for determinism or the entailment from it to externalism, or for that externalism direct, could it be said that we ought to believe it – that we are required to believe it. And no course of reasoning against determinism, or against the entailment from determinism to externalism, or against externalism direct, can be seen as such that we ought not to believe it – that in accepting it we would be (deontically) irrational or unjustified. There would be no libertarianism, internalism or whatever else, such that we ought not to believe it. (The determinist’s opponents could, however, consistently maintain that one ought to accept their libertarian internalism, and ought not to accept determinism.)
In summary, we get the following: if determinism is true, then no-one can do otherwise (2–4) and therefore no-one may reason otherwise (5). Assuming that the ability to reason otherwise is necessary for someone to be held epistemically irresponsible (1), no-one may then be held responsible for their intellectually wrong actions or unjustified, irrational cognition (6). But if no-one is responsible for their unjustified cognition, then no-one is epistemically justified either (7) – in the intended, internalist sense (8). If no-one is ever, under any circumstances, epistemically justified (in the intended, internalist sense), then one who contends that determinism is true is without the kind of epistemic justification they require to make or defend that contention (9). So, one cannot be epistemically justified in claiming that determinism is true. So, determinism is an intrinsically unjustified theory.
I take it that this would be a wholly unsustainable position for the determinist to be in – that the determinist simply must resist the conclusion of this argument. Determinists must be able to justify their position and oppose their opponents’ positions. The framework for such justification must be in place – no metaphysics can be so powerful, so totalizing, as to undermine it.
Determinists will respond here, and the responses available to them will be similar to those offered to the other transcendental arguments – particularly the ethical. At the heart of many of these responses will be some softening of determinism. Various proprietary versions of compatibilism (old and new) will compete to furnish us with a sense in which the demands of this argument can be met – as, say, for the classic compatibilists, where the ‘can’ component of the internalist implication is satisfied: at its crudest perhaps just a freedom from constraint, though other requirements – deliberation, for example – could be added if wished.
Whether or not the argument survives such counters, let us scotch the abruptly dismissive view that transcendental arguments like these are merely and immediately question-begging. Here this claim is made by Patricia Churchland in regard to Popper and Eccles’s (1977) use of this chapter’s Epicurean title quotation as an epistemic transcendental argument (Popper and Eccles had argued that the ‘weighing of reasons’ was a necessary condition on epistemic justification that would be denied us by determinism):
That the question is well and truly begged can be seen. … Determinism does not deny that our opinions are sometimes the result of the weighing of reasons. On the contrary it is a theory concerning how we are going to explain the phenomenon we call ‘the weighing of reasons’. (Churchland 1981: 100)3
Determinism, however, is not an epistemic theory, any more than it is an ethical theory – it is a metaphysical theory: that every event which occurs (including ‘opinions’, ‘the weighing of reasons’) was determined prior to the existence of any agent and any reasons by a complete state of the world at some time in the remote past together with all the laws of nature. Determinism’s epistemic consequences (or ethical consequences, or other consequences) are of the nature of commitments which follow from this: commitments which Epicurus, Kant, Popper and others have revealed it must defend or be refuted by. When the full panoply of such commitments is revealed, it will be a substantive issue as to whether determinism has the resources to reconstruct them in a way that leaves it no longer vulnerable to the transcendental argument.
8.2 Options for the determinist
Besides claiming that the above argument begs the question, the determinist possesses a number of options. Among the more obvious are these:
• The determinist may oppose internalism tout court, claiming that epistemic normativity is not to be understood on the model of ‘oughts’. This means accepting 9 and saying ‘so what?’ One embraces a totalizing, eliminative, ‘existential’ externalism. This position has been dealt with already, at length.
• The determinist may accept epistemic internalism, at least at some surface level – thereby understanding epistemic normativity on a model of intellectual ‘oughts’ – yet oppose premise 1, OIC (or, at least oppose ought not implies can refrain from). This position has been dealt with already, at length.4
• The determinist may accept OIC, but claim that some version of soft determinism satisfies the ‘can’ requirement – say, opposing 4 or 5. This was the approach of all the classic compatibilists, from Hobbes through to the many ‘conditional analysis’ accounts of the twentieth century. Although, after Frankfurt placed other resources at the compatibilists’ disposal, such accounts have substantially fallen into desuetude, I will have some brief things to say of such approaches later. However, two points deserve to be made even at this initial stage: one, these attempts have been failures. Two, this general approach – that a soft determinist ‘can’ is hard enough to satisfy the internalist conditional – is question-begging against this argument (van Inwagen 1983: 121ff). An argument has been advanced (2–4) that a soft determinist agent (any determinist agent) cannot really do otherwise. Soft determinists must engage with the specifics of that argument in order to escape it: not merely reassert their compatibilist formulations regardless of it.
• The determinist may insist this argument fails against ‘reasons-responsive’ compatibilist views: ‘we’re immune – we build reasons in from the beginning’. More will be said of this in the remaining two chapters. However, even at this initial stage, it should be pointed out that metaphysics must come first. This transcendental argument is precisely raising a problem (a metaphysical problem) with the idea that a determinist account can be reasons responsive in any sense deep enough for reflexive epistemic justification.
• The determinist may embrace global asymmetry (rejecting premise 7). The determinist could see herself as possessing epistemic justification in embracing and advancing her metaphysics, even though she could not (globally) ever have been other than thus, epistemically, justified. However, the bracketed clause gives in ellipsis the neo-Fregean counter to this that has already been advanced in Chapter 6 (Section 6.7).5
• The determinist may reject the Consequence Argument – claiming that the notion of modality employed (‘powerlessness’) is illicitly transferred across premises 2–4 in the epistemic and ethical transcendental argument (1–3 in the conative transcendental argument). This response will be considered shortly.
In fact, any intellectually serious determinist will likely be committed to more than one way of opposing this argument. So, alternating between rejecting the validity of the Consequence Argument and accepting its validity but maintaining that soft determinism can satisfy the ‘can’ aspect of this requirement is labelled by Kapitan (2002) as the ‘Main Compatibilist Response’ (to the Consequence Argument) – with the Consequence Argument still the most powerful counter to compatibilism, and the ‘softening’ of determinism that compatibilism represents being, we shall see, the chief underlying motivation for rejecting transcendental arguments of this kind.
The basic response to the Consequence Argument is that the modal notion used (here, of ‘powerlessness’) does not transfer across the premises. Though it is conceded that we are, in one sense of ‘powerless’, powerless to change the laws of nature, etc. (as in premise 3), we nevertheless are not powerless in some other sense – say, to act as we will (as in 4 or 5). But making this argument against modal transfer crucially requires the determinist to motivate, philosophically, a determinist account of ‘power’ or ‘ability’ – and that just is the enduring compatibilist project, in one form or other, from Hobbes through to Frankfurt and beyond.
The Consequence Argument is of course only one argument for incompatibilism. Among the others there is the Direct Argument, there are arguments from source and/or ultimacy, there are manipulation arguments and there are luck arguments. I think that the best versions of each of these arguments present challenges that are difficult for compatibilists to respond to in a principled fashion; and that some, where suitably qualified and developed, may rival the Consequence Argument for power and scope.
Were, then, the opponent of this transcendental argument to oppose the Consequence Argument, it would be open to the transcendental argument’s defenders to employ one of these other arguments for incompatibilism. I have already, in Chapters 5–7, drawn upon strands derived from some of these other arguments (especially the source argument), and will do so again in Chapter 10. The indirect epistemic argument above, then, may be taken to have its premises 2–4 (already only an ellipsis for the fully developed Consequence Argument) as a plug-in ‘module’ for whatever is taken to be the best argument for incompatibilism – and I shall here work with the assumption that this is (whatever is agreed to be the best version of) the Consequence Argument. It is noteworthy that other parts of this argument could be taken as separable components or ‘modules’ in turn – for example, as we have seen in Chapter 6, the conception of deontic leeway that is deployed: PAP, OIC, principle of avoidable blame (PAB), etc. The conception of epistemic justification that determinism is taken to withhold from us has likewise taken different forms in previous authors’ versions of this argument; though there I stand fast upon the thin deontic account that was articulated in the first part of this book (Figure 8.1).
In what follows, we won’t do anything like going through the numbered premises in order, giving arguments for and against each in isolation from the other premises. We will criss-cross the possible objections to premises as they arise in consideration of opposing positions. We start with some observations concerning this argument’s reliance on the Consequence Argument.
8.3 The Consequence Argument
There has been a vast body of work addressing the Consequence Argument, and in a book of this size, possessed of so many other aims, I should prefer not to add to this literature. If challenged, I would be prepared to claim to be relying on the Consequence Argument rather than defending it. Nevertheless, several points are worth registering before repairing to this position.
Look at the modal term (‘powerless’) that is used in the argument across premises 3–5, and especially its use in the (elliptical) Consequence Argument proper of 2–4. ‘Powerless’ is the same term, given the same sense, throughout. The basic compatibilist objection to the Consequence Argument is, however, that the modal notion used (here, of ‘powerlessness’) does not transfer across the premises. In other words, we are, in some one sense of ‘powerless’, powerless to change the laws of nature, etc. (as in premise 3), but we are not powerless in some quite other sense (say, for example, the classic Hobbesian sense: to ‘act as we will’ – in premises 4 or 5).
It is difficult for me to appreciate this family of counters to the Consequence Argument. The objection precisely isn’t that the Consequence Argument commits a fallacy of equivocation in the sense of its modal notion (‘powerlessness’) from its premises through to its conclusion. The objection seems to be exactly the reverse: that it doesn’t change the sense of its central modal notion (‘powerlessness’) from that used in its premises through to that used in its conclusion – and what kind of objection is that?
The opponent of the Consequence Argument appears to be arguing that modal notions should be used equivocally across 2–4 (but aren’t) – a very odd objection to any argument. One could go so far as to see this objection as requiring of us that we commit a fallacy of equivocation across 2–4; namely, that although avowedly we are powerless to change anything in the sense of ‘laws plus Big Bang’, we have some neo-Hobbesian (or other) sense of ‘power’ or ability to act on our determined desires in 4 or 5. Well, so what for that? An argument ought not to equivocate in the sense it attaches to its crucial operators or terms from across its premises and through to its conclusion. We teach undergraduates that to equivocate thus is to commit a basic fallacy. The Consequence Argument notes that if we accept determinism, we are powerless to avoid or alter p, for any arbitrary p, in exactly the sense that we are powerless to alter the laws of nature and the start-up conditions of the universe. This is the appropriate sense of ‘powerlessness’ for us to employ – rightly, it does not change across the premises to the conclusion.
8.3.2 Opacity, the main compatibilist response and belt-and-braces
There is a fundamental distinction between compatibilists and incompatibilists in their treatment of the argument surrounding 2–4. It is a commonplace that one philosopher’s modus tollens is another’s modus ponens. Consider in this regard a modified version of Moore’s notorious argument:
M2 If this is a hand, then I am not a brain in a vat.
M3 I am not a brain in a vat.
The Moorean argues by modus ponens from his being justified in maintaining M1 and M2 to conclude M3. His sceptical opponent argues by modus tollens from the fact that we cannot justify M3, to conclude that we cannot justify M1. Read purely alethically, each of these different uses of M1–M3 must be valid, but read epistemically, most maintain that these and cognate arguments do not go through (in either direction, quietist or sceptical). Many see this as being because epistemic notions like warrant, justification, etc., are not transparent, but are trans-substitutionally opaque – they are not closed under logical deduction. We have warrant, in the context of everyday perceptual judgements, in asserting M1. But M3 embeds within a context of utterance relative to the refutation of idealism or scepticism, or some other very philosophical context – and the avowed truth of M2 still may not licitly transfer epistemic warrant from M1’s epistemic context to M3’s epistemic context, of which more in Chapter 10.
Consider now, our 2–4:
2. Necessarily, given a complete state of the world at some time in the remote past and all the laws of nature, then p, for any arbitrary p.
3. We are powerless to avoid or alter the remote past or the laws of nature.
4. We are powerless to avoid or alter p, for any arbitrary p.
The compatibilist and the libertarian alike reject 4 – but then radically diverge. The libertarian rejects 4 and then concludes that 2 is false – via an acceptance of 3. Similarly, though in the other direction, the incompatibilist (‘hard’) determinist just accepts the Consequence Argument 2–4 in its entirety. Akin to Moore and his sceptical opponent, both kinds of incompatibilist (libertarians and hard incompatibilists) are such because they accept transparency: ‘power’ and ‘ability’, ‘cans’ and ‘cannots’ in the above argument are taken to have the same transparent sense, which modally transfers via 3 in the argument. In contrast, the compatibilist accepts 2 (on pain of not being a determinist), and rejects 4 (on pain of not believing in free will). But compatibilists must reject transparency, and argue for 2 and 4 having a different, equivocal sense of ‘power’, ‘can’, etc. – on pain of not having a motivation for their position. For the compatibilist, ‘ability’, ‘inability’, ‘avoid or alter’ – all our terms for intentional achievement, etc. – are trans-substitutionally opaque; for the incompatibilist of either type, they are transparent. Sometimes, the compatibilist’s position on opacity here is augmented by the ‘epistemology not metaphysics’ arguments already considered in the preceding chapter (Dennett 1984: 148; Kapitan 2002: 149). However, whether or not there is seen as being an epistemic component adding to the arguments against transparency, the compatibilist already maintains the position that there are different senses of ‘can’ (or ‘power’) operative across 4 and 3; and thus, independent of epistemic considerations, that opacity obtains and deductive closure fails. Compatibilists take as their starting point that we can do something ordinary language, common sense, pragmatics, etc., would hold we can do – a ‘p’, any arbitrary p (raise one’s hand). This is like Moore’s ‘here is one hand’. But they refuse to make the neo-Moorean, experiential-libertarian incompatibilist move – that is, employing closure to argue that thereby determinism/indeterminism is false. The truth of 2 is embraced on physical grounds (embedded within Campbell’s (1957) ‘external standpoint’), the truth of not-4 is embraced on common sense or pragmatic or experiential grounds (embedded within Campbell’s ‘internal standpoint’) and 3 is qualified, perhaps with a neo-Lewisian form of words, to prevent conflict.
Notice, though, that 3 need not be qualified or opposed by the determinist, and that it isn’t at all obvious why it so commonly is. If lemma 2–4 is opposed on grounds that 2 and 4 embed within different, opaque, non-alethic schemata (wholly different senses of modality, say), then it is not clear why some clever scholastic take on 3 is required – what it adds. Once we insist on different, subscripted senses of ‘power’ or ‘can’, the argument doesn’t go through anyway. Consider the following argument:
R1 Ratty and Mole sat down by the bank.
R2 A bank is a financial institution.
R3 So Ratty and Mole sat down by the financial institution.
This argument doesn’t go through because bankR1 and bankR2 are wholly distinct in meaning, and so the argument R1–R3 commits a fallacy of equivocation. There is no need at all to provide some additional, frankly odd, form of words in which saying R3 is ‘permitted’: bankR1 and bankR2 are distinct and that’s it – that’s all that’s needed.
Consider, then, how odd is the work of very many compatibilists in trying to establish that there is a sense in which we can change the laws (‘local miracles’) or start-up conditions (‘backtracking counterfactuals’) that would be needed to permit them as determinists to embrace premise 2, yet as believers in free will to oppose premise 4. Leave aside the already strange conception of philosophy which has it that our aim should be to search after a clever form of words that would licence us to say we can act such that the laws of nature or start-up conditions that obtained at the Big Bang would have had to have been different. Instead, ask where the point is to this, as an argumentative response, given the other things the compatibilist is anyway committed to. It’s belt-and-braces: if you’re a compatibilist, you already deny transparency – at least in some (crucial) sense – for you deny that the sense of ‘powerless’ used across the premises of this argument is the same. Why, then, try to provide a sop to an entailment that only applies if you accept transparency? The underlying motor driving compatibilism is an adherence to the claim that there is a meaningful and worthwhile sense in which it is true we can do p, and yet to insist that determinism is also true. The modal operators (the ‘cans’) applying across the premises of the argument are and must be held to be different. How, then, is there any pressure on the compatibilist to satisfy the entailment that only emerges if we carry the same cans/cannots across the premises of the argument to the conclusion? I’d suggest that working through this otherwise superfluous line of argument is a sign that the compatibilist is profoundly uneasy in dismissing the metaphysical reasoning which every undergraduate accepts until it is educated out of them; indeed, which every pre-philosophical intellect embraces – and that the rigorous work of van Inwagen, Wiggins, Ginet and others make respectable for the professional philosopher: that were determinism true, we would be powerless to avoid or alter p, for any arbitrary p.
8.3.3 Must have a motivation to oppose modal transfer
The Consequence Argument is a philosophical argument – not a ‘proof’ (van Inwagen 1983: 69; Warfield 2000: 168ff).6 In the extensive literature discussing the various versions of it, this is sometimes overlooked. There are very many counterarguments to the effect that, as a point of logic, the modal notion of ‘powerlessness’ (or other modal term used) does not transfer across the premises. In other words, we are, in some one sense of ‘powerless’, powerless to change the laws of nature, etc. (as in premises 2 or 3), but that this would not, as a point of logic, establish the validity of a claim that we are powerless in some other sense – say, to ‘act as we will’ (in 4 or 5).
It is, however, not enough to point out that as a point of logic, the modal notions used across these premises might change, and thus that we can entertain the supposition that, in some sense or other, our agent ‘could’ do otherwise – has that power, that ability. These counterarguments ultimately require their proponents to defend an analysis of ‘power’, ‘ability’, ‘can’, whereby – say, as in opposition to 4 – we ‘can’ do otherwise than the laws of nature and the Big Bang necessitate. We are all familiar with various formulations here – a small sample of which will be considered in Chapter 10. The compatibilist, however, has to motivate her chosen interpretation of these terms. Of course we ‘can’ do otherwise than we do in the sense that this is logically possible, but so what for that? It is logically possible that we change the distribution of energy at the Big Bang, or change the fundamental laws of nature – one would not normally think that this, of itself, represented a good argument for compatibilism. A famous compatibilist treatise was subtitled: ‘The varieties of free will worth wanting’. Well, one could require of the compatibilist in turn that she give us an analysis of ‘can’, ‘power’ or ‘ability’ worth wanting – not merely a form of words, that is, or an exercise in analytic casuistry; not yet another formulation of the kind that warranted Kant’s famous judgement on classic compatibilism: ‘a wretched subterfuge … a petty word-jugglery’ (Kant 2004: 76).
Thus, O’Connor put it:
When I wonder what it is now in my ability to do, I am wondering what is open to me given the way things are and have been and the laws that constrain how things might be. … An ‘ability’ to act here and now, the actual exercise of which strictly requires a prior condition that is lacking and which I cannot in any way contribute to bringing about, is, in the sense at issue, no ability at all. (O’Connor 2000: 17)
I share O’Connor’s ‘brute wondering’ in this regard. I doubt, that is, whether there is a separate front to be fought in the compatibilist/incompatibilist wars concerning the logical status of modal transfer across the many versions of the Consequence Argument. A number of compatibilists would like us to believe there is such a separate front, but I suspect that the issues all along are the classic philosophical ones, familiar in their different ways from Hobbes onwards. It is to these we now turn.
8.4 What kind of freedom do we need to resist the transcendental argument?
The transcendental argument is meant to require us to abandon determinism. What are we meant to abandon this for? The obvious answer is indeterminism. However, we are all of us surely familiar with the determinist response to this position.
The dilemma argument
This stock argument, developed by the classical compatibilists, the indeterminist disjunct of which was sardonically labelled the ‘Mind Argument’ by van Inwagen, presents us with the dilemma that our acts of will, desires, cognitions, etc., must be either determined or undetermined. If they are undetermined, they are by chance, random. They are thereby not a choice of the person, an act proper, the expression of a preference of the agent. A ‘freedom of caprice’ is not in any way fitted to ground an account of free will, of agency. So they must be determined.
The notion mankind have conceived of Liberty, is some dignity or privilege, something worth claiming. But what dignity or privilege is there, in being given up to such a wild Contingence as this, to be perfectly and constantly liable to act unreasonably, and as much without the guidance of understanding, as if we had none, or were as destitute of perception, as the smoke that is driven by the wind! (Edwards 2007: 151)
There are things to cavil at in the dilemma argument, but it at least shows that indeterminism, merely as such, even if ‘enclosed’ within an agent, would not be sufficient, and perhaps not necessary, for the personal, moral and epistemic responsibility needed to resist the transcendental argument structures given earlier.
Indeed, the transcendental argument structures seen above threaten to ‘gear-up’ the determinists’ objections to the ‘free will as simple indeterminism’ position in a way that would not have been anticipated by the dilemma argument’s historical proponents. It would be a simple matter to adapt the transcendental arguments against determinism to claim that the indeterministic agent may, as a result of his ‘wild contingence’, lack epistemic justification to reach or advance his defence of indeterminism.
However, consider the role that the dilemma argument was meant to play as a whole: not simply to undermine indeterminism, but to establish determinism as a necessary condition on personal, moral and epistemic responsibility. Does it do that? No, for two reasons. Firstly, because even if there were no alternatives, still it has not positively established that determinism is compatible with responsibility – against the pessimistic, sceptical conclusion of hard incompatibilism. This is because responsibility requires not chance but choice, with soft determinism and indeterminism both denying we can choose our acts of will, our desires. That is (secondly) because there is a positive alternative, and so the argument rests on a false dilemma: our acts of will, desires, could be neither determined by external natural law, nor undetermined and by chance, but self-determined.
By the liberty of a moral agent I understand a power over the determination of his own will. (Reid 1969: 259 – Essay 4, Ch. 1)
This power is that ‘dignity or privilege’, necessary for internalist justification, which even softer forms of determinism deny us.