1. Storytelling in the Land of the Free
‘When I act freely, I write a sentence in the story of my life’: so writes John Martin Fischer, who has done more than anyone to make vivid an account of human freedom rooted in the notion of ‘guidance control’, a sort of control expressible in the absence of genuine alternative possibilities, the presence of which are, by contrast, requisite for a more demanding form of ‘regulative control’ (Fischer Reference Fischer2009: 167). On Fischer's approach, guidance control is the only freedom-relevant condition for the ascription of moral responsibility (Fischer: Reference Fischer2009; Fischer and Ravizza Reference Fischer and Ravizza1998: 69–73). As he sees things, ‘an agent may be morally responsible but never have had genuine metaphysical access to alternative possibilities—he may never have had “freedom to do otherwise”’ (Fischer Reference Fischer2009: 167). All that matters for this form of control is that an agent be reasons-responsive and in so being come to manifest ownership of the mechanism by which her reasons eventuate in action. In this sense, ‘guidance control’ expresses a commitment to traditional compatibilism: it articulates a commitment to the compatibility of human freedom and moral responsibility consistent with the truth of universal casual determinism. Yet the form of compatibilism articulated is novel and refreshingly nonstandard.
Or perhaps it would be more apt to say not that Fischer's notion of guidance control expresses a traditional form of compatibilism but rather that it sidesteps the question as traditionally conceived. This is because guidance control finds its richest expression in terms of our responsiveness to reasons rather than in any attempt to address the question of how we as agents might or might have done otherwise relative to an array of alternative possibilities, assuming, that is, that we set aside Frankfurt-style cases where one is (it is maintained) fully free in the absence of any alternative possibilities. Its best development occurs rather in the idiom of narration and storytelling. Like David Velleman (Reference Velleman1991, especially Reference Velleman2003), Fischer identifies a locus of agentive value in the act of constructing a coherent narrative of one's life in the writing of sentences—not sentences constituting a bare chronicle of one's doings, where each catalogues an event done in its own discrete slice of time, but rather in the composing of sentences that contribute to a diachronic arc stitching together its discrete components so that they merge into a fully unified, coherent narrative. Just as Aristotle remarks that the unity of Homer resides not in a bare sequence of verses (Metaphysics 1030a7-11), so the value of our life stories emerges not in a simple litany of deeds and moments, but rather in their richly wrought telling: as we narrate our self-stories, some events prefigure other events whilst other, later events qualify and color the earlier events, complementing them, completing them, or perhaps in some instances challenging and even undermining them. A narrative is not a list but an organic, structured, integrated tapestry of deeds and themes, presented as mutually implicating and intimately interlocking.
This last observation is recognized and indeed stressed by both Fischer and Velleman. Neither, however, develops one striking implication of the narratological approach to free will: later events do not merely color and complete earlier events in a story, but alter them. In a rich and robust sense, later events in a narrative determine earlier events to become the events they became. They even manage to change them, by determining them to be other than they were. Moreover, they make them other than they were not only by rendering them precise where they were vague or indeterminate but also by determining them to be events they were. If this is so, then there turns out to be an unexpected and fruitful way of connecting the narratological conception of freedom with the traditional debate that it supersedes, or supplants, or simply sidesteps. When ‘I write a sentence in the story or my life’, I alter the complexion of my narrative; I thus change, or can change, earlier components of the narrative told.
Roman Altshuler provides a succinct description of the general way in which an action can alter earlier components of a narrative: ‘In both cases, agents change the role past motives play in their actions—and thus reconfigure the motives themselves—by changing the ongoing narrative of their lives. In speaking loosely of “changing” past motives, however, I do not mean anything spooky—like reaching back in time and rewiring one's own neural synapses. I mean only that by changing the narrative of our lives, we can change the meaning of items within that narrative’ (Reference Altshuler2015: 879).
Actions performed at one time, according to Altshuler, can alter the motives that explain those actions within an agent's narrative. Despite advocating this sort of retroactive narratological determination we find congenial, Altshuler shies away from the thought that a narrative can alter the past—a view he regards as ‘spooky’ (Reference Altshuler2015: 879). He is thus in agreement with Fischer and other narratologists in rejecting the retroactive determination of the structure of reality.
Their common reluctance to allow that an agent can make a relevant difference to the world gone by is understandable, indeed inevitable, as long as the traditional problem of free will is cast entirely in the mono-dimensional framework of a causality unfolding in a single temporal dimension in which earlier, discrete events necessitate later, discrete events, but in which later events cannot touch or alter earlier events. In this framework, all that remains for an agent, as Carl Ginet observes, is ‘to add to the given past’ (Ginet Reference Ginet1990: 103–10).
By contrast, we think that the determination invoked in narration is not subject to such constraints. We reject the view that the past is fixed, and, more to the point, we regard as question-begging at best any such claim offered in the context of an open debate about free will. The world in which we reside, the world of free agents, is not a world that respects the constraints artificially imposed by those motivating a problem they understandably report themselves unable to solve. There turns out to be an unexpected metaphysical clout to William Faulkner's oft-repeated aphorism from Requiem for a Nun: ‘The past is never dead. It's not even past’ (Reference Faulkner1950: 49).
2. A Debate at Stalemate
If we can do no more than add to the given past, then we cannot make the world such that it could have been other than it is. Though surely it could have been other than it is, had its causal trajectories been other than they were, they were in fact as they were and consequently the world is at is. Compatibilists seize on one evident truth: nothing we can do can change the causal structure of the past, so that any freedom we possess must be understood within those confines, even when those confines imply that we are straightforwardly determined, casually determined, to do what we do, even to the fine-grained point of being determined to choose what we choose. Libertarians will not relinquish another, equally evident truth: we cannot be free, genuinely free, as they say, unless it is in our power to do or forbear doing a given action, irrespective of the causal history of the world to the point of our acting (see Chisholm Reference Chisholm, Fienberg and Shafer-Landau1987: 419). They then square off and reach an intricately sophisticated stalemate—but a sophisticated stalemate, no matter how intricate, is a stalemate all the same.
Galen Strawson (Reference Strawson2010) provides an engaging presentation of the state of play. He depicts an endless merry-go-round of positions and counter-positions, jointly eventuating in an ultimately unsatisfying cycle of metaphysical speculation about free will. Compatibilism, which he offers as a prima facie reasonable view, runs into difficulties in handling intuitions about ultimate moral responsibility: we seem somehow not to be responsible for our actions if we are finally but dominoes felled from behind in accordance with exceptionalness laws that we neither have the power to alter nor, in the main, even to know. These intuitions naturally give way to an incompatibilist position. Yet, Strawson observes, incompatibilists cannot articulate a conception of freedom that makes us responsible for our actions at all; rather, they render us passive subjects, mere spectators witnessing a flow of unpredictable events passing by, randomly, outside of our control (Reference Strawson1998: 746).
So, the free will debate cycles back to compatibilist approaches, but these, again, intricate and fiendishly clever though they become, eventually run afoul of the same old intuitions about responsibility. The situation seems dire, and, as David Wiggins observes, in need of a fresh start:
Whether or not it is our world, we must continue to press the question, ‘What is the possible world which could afford the autonomy of thought or agency the libertarian craves in this one?’ I do not think an answer to this question is going to be found merely by analysing ordinary notions like can, or power, reasons, or choice. If a coherent libertarian account of freedom were already embodied in some existing human language or conceptual scheme, we would have found it long ago. What is called for is not mere analysis, but theoretical construction out of diverse materials, old and new, ordinary and not-so-ordinary—a thought experiment, if you like, that is meant to produce what Plato called a ‘likely story’, though not necessarily a familiar one. (Kane 1999: 124)
We completely endorse Wiggins's call for a new construct—to the point of offering a story that is, in fact, more than merely likely. We are somehow never both free and responsible; yet we must somehow be both, that realization notwithstanding, if we are to conceive ourselves as agents in the world.
Strawson sums up the predicament as follows: ‘Now the argument may cycle back to compatibilism. Pointing out that ‘ultimate’ moral responsibility is obviously impossible, compatibilists may claim that we should rest content with the compatibilist account of things—since it is the best we can do. But this claim reactivates the incompatibilist objection, and the cycle continues’ (Reference Strawson1998: 749). Strawson goes on to propose a way out of the cycle, of sorts. Philosophers should, he avers, simply move from the metaphysical to the psychological: ‘There is an alternative strategy at this point: quit the traditional metaphysical circle for the domain of moral psychology. The principal positions in the traditional metaphysical debate are clear. No radically new option is likely to emerge after millennia of debate. The interesting questions that remain are primarily psychological: Why do we believe we have strong free will and ultimate responsibility of the kind that can be characterized by reference to the story of heaven and hell?’ (Strawson Reference Strawson1998: 748). There is something honest in the defeatism of Strawson's way out. He proposes, like the host of a dinner party made uncomfortable by a stale, faltering discussion, that we simply change the subject.
3. Remaining on Topic
Perhaps running the risk of boorishness, we decline the host's invitation. We prefer to carry the conversation forward. While we agree that we currently lack a compelling account of either compatibilist responsibility or libertarian freedom, we suppose that a retreat to psychological diagnostics is a defeatist counsel of despair. More precisely, we remain optimistic about the possible payoff of continued metaphysical speculation at this juncture. There is, however, also something importantly correct in what Strawson says: as long as we engage in continued metaphysical speculation using only the terms structuring the stalemated debate, we will, just as he says, merely go round the merry-go-round a few turns more before ending up back where we were. We look at one of those terms, namely the conception of determination inherent in the conception of causation shared by all riders on the merry-go-round; but we do not intend to offer a general disquisition on the nature of causation. Rather, by developing an alternative conception of determination rooted in the genuine insights of narrative construction, we think we can characterize that corner of determination relevant to agency, and to do so in a manner explicative of our freedom as actors.
We do not offer a theory of action beyond saying that when acting we are in one crucial respect like authors: we are like authors in that we determine the events of our lives to become the events they are, and we at times, if not always, do so in just the way authors of fiction cause their fictions to be the fictions they are. It does not follow that we are fictions, for we are not; and it does not follow that our determination is fictional, for authorial causation is not fictional determination but real determination. Authors create, bringing things into existence; authors alter what they create, thus causing the things they have created to be other than they were; and authors also have a noteworthy ability to make what was once indeterminate become determinate, even after—in a sense of after to be rendered precise—they have been created. We are, we contend, free in much the same way. It follows that the freewill debate need not be abandoned in favor of psychology so much as theoretically reconstituted to avoid a false proposition that structures and motivates it.
The first, easy step in the theoretical reconstitution requires recognizing that we have become blinkered with respect to the terms in which we have cast our free-will debate. The traditional distinctions in the free-will debate rest on an assumption that is usually not mentioned much less defended; it is, for instance, not acknowledged by Strawson, even though he is someone who has taken a refreshingly broad view of the framework of the debate. In pitting libertarianism against some variety of compatibilism, philosophers understand the existence of libertarian freedom as entailing the falsity of determinism. Although almost universally accepted, such an understanding of the contours of the free-will debate masks an assumption that can be called the absolutist assumption. According to the absolutist assumption, statements of the form ‘A is free’ and ‘an event e is determined’ are semantically complete: they do not need any further specification in order to be true or false.
Despite the prevalence of what can be called absolutist accounts of freedom, it is possible to deny the absolutist assumption and accept instead what we call the authorially parameterized account of freedom. This is the second and much more difficult step in reconstituting the traditional free-will debate. It is, however, a step worth taking, not least because it brings together two fields of inquiry wrongly but understandably thought to be incommensurably alternative approaches to the question of free will, namely the traditional account and the narratological approach. Thus, for instance, in supplanting regulative control with a notion of guidance control, Fischer implicitly concedes that the traditional debate offers only a dead end; there is simply no room for progress. By contrast, we maintain that the narratological approach may be brought to bear directly on the traditional approach. The trick is to deploy an insight of narratology into the belly of the traditional framework. In brief, what must go is the absolutist conception of causation that structures the debate; and what must supplant it is the framework of narrative determination. This occurs when the absolutist conception of causation gives way to an authorially parameterized conception of determination. According to this conception, a statement of the form ‘an event e is determined’ is semantically incomplete. Its completion requires specifying a parameter with respect to which the event is determined; and since parameters can and do change, so too do events, whether those events be merely determinable or fully determinate. Events can and do change, that is, relative to the parameters deployed in their specifications. Because one such framework is precisely temporal, we offer a two-dimensional temporal semantics for assessing the truth values of authorially parameterized statements.
One striking consequence is that if one accepts a parameterized account of freedom, it is at least formally possible to remove the incompatibility between determinism and the existence of libertarian freedom. For if one accepts a parameterized account of freedom, it becomes possible to say that someone is completely determined with respect to one parameter and yet free in a libertarian sense with respect to another. The apparent contradiction between determinism and libertarian freedom is thus eliminated in a way analogous to the way in which the apparent contradiction in saying that a boy is both tall and short is eliminated by shifting contexts of evaluation. Without parameters, it is not easy to see how someone can be both tall and short. But with the appropriate parameters in place, the apparent contradiction dissolves—there is, after all, nothing at all problematic about saying that someone is tall for a boy his age but short for a nationally competitive high jumper.
Helpful in this connection is the work of John McFarlane (Reference McFarlane2005, Reference McFarlane2014, Reference McFarlane2016). We agree with him in thinking that ‘truth value might vary with the context of assessment as well as the context of use’ (McFarlane Reference McFarlane2005: 321), but we do not take up his broader understanding of the relativist's doubly contextual truth predicate, such that an assertion might be ‘true at context of use CU and context of assessment CA’ (McFarlane Reference McFarlane2005: 321). This is in part because we do not mean to concern ourselves with the larger and more daunting topic at the heart of his work, namely the general understanding of relativized truth. Rather, we concern ourselves with the more delimited question of whether an absolutist or a parameterized account of freedom should be adopted with reference to the traditional problem of free will. We argue in favor of the parameterized account; but this account requires an explication of the notion of authorial determination.
Now, there are many different sorts of parameters one might offer in making judgments about determination and freedom. The parameter we think lies at the core of the issue is a temporal parameter (again, Ginet: all one can do ‘is add to the given past’ [Reference Ginet1990: 110]). We can see why Ginet's plausible sounding statement is false once we adopt a temporally parameterized truth predicate in connection with a proper understanding of authorial determination. For this reason, we urge, ascriptions of freedom should be parameterized with respect to the passage of time. Such a parameterization not only makes compatible determinism and the existence of libertarian freedom, but it also provides an intuitively compelling picture of free action, one in broad sympathy to the approach articulated by Fischer (Reference Fischer2009). As we discuss below, such a view requires relinquishing what seems to be Ginet's intuitively compelling thesis about the past, one that Fischer, along with other his fellow participants in the traditional debate, has not been prepared to abandon (though Fischer arguably does make some steps in the right direction by appealing at crucial junctures to the Ockhamist distinction between hard and soft facts). That some such conceptual alteration is necessary, however, is altogether unsurprising, given what seem to be the conceptual difficulties attendant upon articulating a satisfactory account of libertarian freedom. Again, we also agree with Wiggins (Reference Wiggins and Honderich1973), who recognizes that a coherent account of freedom need not be familiar; it need only be coherent. As a first attempt to take our difficult second step, then, we aim to show that the authorially parameterized conception of freedom is at least coherent.
4. Two Kinds of Historical Precedent
Although we believe that our approach is novel, we would like to stress that the general strategy we embrace has instructive historical antecedents. In particular, Immanuel Kant is plausibly understood both as rejecting the absolutist assumption and as embracing a sort of parameterized account of freedom. In the third chapter of the Groundwork of the Metaphysic of Morals, Kant articulates two conceptions of freedom, each of which is strongly libertarian. In its negative sense, freedom, according to Kant, is a type of causality that is ‘able to work independently of determination by alien causes’; and in its positive sense, freedom is nothing ‘but autonomy—that is, the property which will has of being a law to itself’ (Reference Kant, Gregor and Timmermann2012: 114). Still, he courts contradiction by insisting that he does not mean to reject determinism. Kant tries to reconcile libertarian freedom and determinism by introducing parameters—he contends that we are phenomenally determined but noumenally free, which in effect offers two frameworks for ascriptions of freedom (Kant Reference Kant, Guyer and Wood1998: A532/B561-A558/B586). With respect to one, an event is determined and so not free; with respect to the other, an event works ‘independently of alien causes’ (Kant Reference Kant, Gregor and Timmermann2012: 52) and so is evidently not determined.
Kant's use of parameterization is, of course, fraught with interpretive controversy and philosophical difficulty. (For an elegant exploration of the alternatives, see Nelkin [Reference Nelkin2000]). Still, his general strategy is noteworthy. Given the apparent contradiction involved in asserting the existence of libertarian freedom and determinism, parameterization would appear to be the only way of accepting the existence of libertarian freedom in a deterministic world. Moreover, Kant's appeal to the phenomenal/noumenal distinction highlights the first criterion that a parameterized account should meet. Any successful parameterized account should not only specify the parameter in view, but should also give some semantic account of how a parameter functions with respect to truth claims. And although one might reject the distinction between the noumenal and the phenomenal, Kant at least goes to great lengths in his attempt to explain what the distinction is supposed to be.
It is not our intention to enter the nuanced debates about these matters amongst historically orientated commentators; we nonetheless do believe that despite its ingenuity, Kant's theory of freedom fails in two different directions: first metaphysically and then also phenomenologically. Its metaphysical failing stems from Kant's conception of the noumenal world. When Kant says that we are noumenally free, he evidently intends to talk about things as they are in themselves, which are independent of the conditions necessary for experience. Since space and time are two such conditions, my noumenally parameterized acts of will cannot be regarded as located in space in time. So, noumenally free acts do not fall within the spatio-temporal manifold. Phenomenally parameterized acts, by contrast, do fall within the spatio-temporal manifold and thus, like all other such events, are determined.
Here, then, enters an objection to Kant's view of freedom: it is not at all clear how one and the same event can be parameterized as both spatio-temporal and as non-spatio-temporal. Just as no number can be even insofar as it inhabits one context and odd insofar as it inhabits another, so no event can be determined as a spatio-temporal being and indetermined as a non-spatio-temporal being. We accept the view, that is, that an event e is spatio-temporal if and only if it is necessarily spatio-temporal, irrespective of any parameters one might use to characterize the event. Moreover—and more to the point regarding our own preferred parameterization—it is unclear what sense can be made of some human's action existing not as part of the temporal order. Every human action, it would seem, is essentially temporal. Locating freedom in that aspect of an action that is not temporally bound, therefore, cannot deliver an acceptable account of freedom. Hence, even though his use of parameterization is ingenious, and even though he employs a distinction that is not ad hoc but rather a principled part of his broader critical philosophy, Kant fails to give a satisfactory metaphysical account of the way in which we can have libertarian freedom in a deterministic world.
This metaphysical shortcoming has a phenomenological correlate. However the noumenal world is to be understood, it seems plain that we seem to ourselves to be actors in time: we act today, now, in the present, and not last evening or next Tuesday morning. Our plans and deliberations are perforce temporally ordered; our executions are experienced as occurring now, and not as anticipations of what is to come or as regrets or satisfied recollections of what has been achieved in the past. Still less do our intentional actions feel as if they were issuing from a timeless zone or a super-temporal present, if indeed sense can be accorded to such notions. Kant might retort that a free action proceeds from reason, which is a timeless feature of the world, not from any act of temporally specified reasoning, and that our actions do indeed sometimes seem to issue from reason. But here we must admit that we have perhaps an irreconcilable difference with Kant in terms of our respective phenomenological perspectives. Although our actions may seem to be in accordance with reason, it does not seem to us that they issue from a timeless realm of reason. Rather, actions seem to us, at any rate, as inherently temporally located and directed. In this way, then, Kant's parameterization falls short phenomenologically as well.
In this respect, Kant may be usefully contrasted with several of his followers. One finds in particular extended passages in the philosophy of Jean-Paul Sartre that not only take the phenomenology of action seriously, speaking in terms of the creation of an intentional action, but insist that such creation is wholly free. When an intentional agent chooses to act, her choice is not part of the causal order. Her choice is of course tempered by a range of facts as they appear to her in light of her context of acting, or what Sartre calls the facticity of her world situation (Reference Sartre and Richmond2021: 178,190, 200–2). These include her physical characteristics, the actions of others, and the laws of nature—none of which she is positioned to change. Still, when she acts, Sartre insists, her actions are not determined: nothing forces her to react to her facticity in any definite way. And in acting, she creates even herself. She is, from a Sartrean point of view, even responsible for creating her own essence.
All the same, Sartre continues, despite the fact that a person's actions are not determined when she acts, there comes a time as of which her self has been fully defined and her actions have entered into the causal order (Reference Sartre and Richmond2021: 65). As of this time, a person is, in Sartre's terminology, no longer for-herself but rather fully in-itself (Reference Sartre and Richmond2021: 65). Once a being in itself, a person, has entered into the causal order; she has become enmeshed in facticity. (More prosaically, she becomes part of Ginet's given past.) When this happens, her actions can be seen to have emanated from the self that she was destined to create even though the creation itself was a free one. In choosing, she creates the person she has become; by anointing some of her desires as the desires that make her the sort of person she will have become, to wit, the sort of person who accedes to these desires (for chocolate, for justice, for beauty) rather than to those (for vanilla, for self-promotion, for remorseless efficiency).
There is something phenomenologically compelling about the idea that through our choices we freely create a circumstance in which the choice-explaining desires we select immediately explain the choices they motivate. This is perhaps the reason why Sartre, a philosopher so interested in phenomenological descriptions, would have advocated it. While we may have introspective access to many of our desires, we remain, all the same, sufficiently opaque to ourselves that it seems at times that we lack determinate desires that can explain our choices. Fully determinate desires simply do not populate our field of awareness prior to our decisions; instead, they emerge to consciousness in the act of choosing. Nonetheless, people can in retrospect say which desires propelled them to act—for their choices make their desires manifest.
Such quotidian phenomenological facts Sartre expresses in much more declamatory language when he discusses anguish and abandonment in Existentialism is a Humanism. His comments about anguish show the extent to which he views human choice as part of the temporal order: ‘When a man commits himself to anything, fully realizing that he is not only choosing what he will be, but is thereby at the same time a legislator deciding for the whole of mankind—in such a moment a man cannot escape from the sense of complete and profound responsibility’ (Sartre Reference Sartre, Kulka and Macomber2007: 162).
The sense of anguish, which comes for Sartre from choosing for all of humankind, is a sense that occurs at a definite moment in time. Moreover, the freedom that becomes manifest at such a moment, requires us to decide the nature of our being, which prior to any such choice is a mere possibility rather than an established fact: ‘We are left alone, without excuse. That is what I mean when I say that man is condemned to be free. Condemned, because he did not create himself, yet is nevertheless at liberty, and from the moment that he is thrown into this world he is responsible for everything he does . . . That is what “abandonment” implies, that we ourselves decide our being. And with this abandonment goes anguish’ (Sartre Reference Sartre, Kulka and Macomber2007: 48).
One might reasonably contend against Sartre that the phenomenological richness of his approach obscures the fact that the process of creation he describes indefensibly foists an arcane metaphysical framework upon what is after all no more than an adroitly observed experiential process of self-discovery, and so, in the end, that he inaptly dresses an epistemic state of affairs in an alien metaphysical garb.
That is, someone may object that, however unaware of our motivations we may be at the moment of our choosing, fully determinate desires and beliefs nevertheless gurgle along beneath our conscious awareness all the while we deliberate, ultimately propelling us to act in fully deterministic ways, only to be discovered after the decisions that emanate from them have revealed them to us by making them manifest to consciousness. This does nothing to show, however, that when we discover why we acted as we did we also made ourselves who or what we were or are.
Although we again wish to prescind from the details of scholarly exegesis, we do maintain that to the degree that there is something in this complaint, it does not undermine the more fundamental insight, whether authentically Sartrean or not, that there might be a metaphysical correlate to Sartre's phenomenologically adroit point: certain of our decisions make it the case that at least some of the events of our lives, including events of the so-called fixed past, come to have some of the determinate features they in fact have. Whether the general kind of phenomenology depicted by Sartre has a real metaphysical correlate is something that can only be ascertained once a metaphysical account of such retroactive determination is provided.
To be clear, then, our brief forays into these two partial precedents have not been made in the service of illuminating the philosophies of Kant or Sartre, both of which certainly merit minute scrutiny, but rather in an effort to explain and motivate two features of our general approach. First, as we have learned from Kant, possibly one and the same event can have distinct though non-contradictory features even from the standpoint of casual consequence and free agency (again, see Nelkin Reference Nelkin2000: 564–65). Second, as offered by Sartre, some of the phenomenology of freedom might lead us to believe that certain choices can render determinate what was left indeterminate before the time of choosing—even where, as we argue below, before means just what it seems to mean, that certain choices can make events which temporally precede them determinate even after their position on the timeline has passed. Of course, this is sort of contention apt to sound outlandish, at least from one sort of perspective. How could an event be made other than it was? More to the point, how can an event become determinate after it has occurred? Surely, someone might insist, this is on its face incoherent.
We now offer a framework capable of removing this seeming incoherence.
5. A Semantic Framework
The seeming incoherence in embracing both libertarian freedom and determinism can be removed by introducing a temporal framework that allows choice to be parameterized with respect to the passage of time. A semantics for such a temporal structure is not hard to develop and can be described very briefly as follows. First, sentences are assigned sets of times. Intuitively, the times are those at which the sentence is true. Such a semantic framework can be modified to accommodate an expanded temporal structure by assigning sentences functions from times to sets of times. The semantics for choice-sentences thus becomes two-dimensional. Because of its two-dimensionality, the framework requires two different operators relative to which such sentences can be considered true or false. Call them the as-of operator and the at operator.
The at operator is the operator that one would employ in a mono-dimensional framework, and is, in fact, the single parameter presupposed by the stalemated free-will debate. We can (and do) say in terms of the at operator that it is true at 2012 that Obama is president of the United States. The as-of operator, on the other hand, represents the moving now. With the introduction of the as-of operator, one must specify both a time as of which a sentence is being evaluated and a time at which it is being evaluated. So, for instance, one can say that as of 2012 Obama was president at 2010. Of course, it may be the case that the moving now does not make any difference with respect to the times at which certain sentences are true or false. So, it is true as of 2012 that Obama is president at 2012; likewise, it is true as of 2013 that Obama is president at 2012. Nonetheless, by introducing such an operator, there is at least the formal possibility that the moving now changes the times at which certain sentences are true or false.
With such a framework in hand, let us reconsider the way in which authorial parameterization might be fleshed out. Consider a decision of Jean-Paul's about which ice cream to enjoy for dessert. Moreover, suppose that there is some sort of law such that if someone has a desire for x at some time t1, and believes that doing y will allow him to obtain x, then ceteris paribus, that person will do y. Were laws to govern psychology in this way, then whether or not Jean-Paul chooses chocolate ice cream at t2 will depend on whether Jean-Paul has the appropriate desires and beliefs at t1, just prior to his choice.
But now suppose that as of the time of the choice, t1, Jean-Paul does not have a determinate desire for chocolate, though he has decided to have some ice cream. (We assume that this is a perfectly familiar situation: Yes, I will have ice cream. Now, let me see: which sort will I have? Hmmm.) Hence, as of the time of the choice it is false that Jean-Paul has a determinate desire prior to his choice such that together with the laws of nature and the existence of other mental states his desire entails that he choose chocolate. Then, on the assumption that he chooses chocolate, as of times later than the time of the choice it was true at times prior to the choice that he had a determinate desire for chocolate, which desire determined him to choose chocolate. So, Jean-Paul's situation is not merely that he failed to access his already determinate desire; it is not, then, a merely epistemic matter. Instead, his desire was genuinely indeterminate. His circumstance is rather like Albert's, when Albert has decided to paint his bedroom wall, and has narrowed down the alternatives to beige and grey. It was determinate that his wall, which had been dingy white, would gain a new color, though it was not determinate which of two colors that would be. It could not become both at once, of course.
It is now possible to see the way in which parameterizing with respect to the passage of time allows for the compatibility of libertarian freedom and determinism. As with any parameterized view, it is not possible to say whether someone is free or determined simpliciter. Rather, one must specify the parameter in question. Hence, in cases of free action, as of times prior to or simultaneous with an action, a person's choice is not determined, but as of times after the choice it is.
Of course, here someone might object that the introduction of a parameter introduces some sort of conceptual loss or even incoherence—of the sort, for instance, we identified in the Kantian framework. What such an objector wants to know, in effect, is whether or not a person has libertarian freedom full stop, not whether he has it relative to some parameter or other. But such an objection simply betrays an allegiance to the absolutist conception of freedom specified in a mono-dimensional temporal framework that cannot at this point be taken for granted. If determination is not parameter-independent, then neither is action-determination. The objector is in effect demanding to know whether, when all is said and done, the boy is really tall.
6. A Metaphysical Framework
Still, the objector has a point. A formal semantics of the sort generated by a paper distinction between the as-of operator and the at operator does nothing to explain, let alone motivate, any contention to the effect that such a device shows that determination is in fact temporally parameterized. Thus, it provides no reason to conclude that the actual choice situations of actual agents are in fact temporally parameterized.
The objector is so far on firm ground. This is to say, then, that we have to this point argued only that an account of freedom that parameterizes ascriptions of freedom with respect to the passage of time, beyond being phenomenologically adequate, has available to it a perfectly coherent semantic framework. Unfortunately, so much is insufficient. If determination is essentially non-temporally parameterized, then the formal semantics will prove a formal distinction without a material difference.
Is determination essentially mono-dimensional? We think it is not, because temporal parameterization may be authorial—and it is here, precisely, where Fischer's newfangled narratology meets old school determinism. As we have indicted, in some ways Fischer's introduction of hard and soft facts presages our approach. This distinction, as he is aware, goes back at least to Ockham: hard facts about the past are temporally non-relational as regards the present, whereas soft facts are, by contrast, relational about the present. Fischer's deployment of this distinction is, in our view, a step in the right direction, but one that does not take the needed next step into a two-dimensional temporal framework. This is why, presumably, he contends that soft facts are all, in the end, changes in how hard facts are valued or understood in the present. The distinction as drawn is articulated in a mono-dimensional temporal framework.
By contrast, within authorial contexts, temporal parameterization becomes not merely possible in terms of shifts in attitudes or evaluations, but fully actual—as actual, at any rate, as authors are actual—and there really are authors, authors who really author real works. Fairly plainly, the most difficult obstacle to our approach derives from the fact it runs afoul of Ginet's edict, which seems evident enough to most, regarding the fixity of the past. The past is given, we tend to think, and we in the present can hope to do nothing more than add to it, by placing onto the ever-growing pile of past events a few more events and then a few more events once again. The pile gets bigger, but none of the items already in the pile can be made other than it is. The fixity of the past entails that even if it is augmentable, the pile of the past is certainly not alterable.
According to proponents of fixity, what is done is done and there is nothing that anyone can do to change it. This is to say, then, that a proponent of fixity will simply deny that differences in the temporal parameter as of which a sentence is evaluated can yield different sets of times at which the sentence is true. And it must be admitted that it is not at all obvious up front how this can be: we seem by our two-dimensional semantics committed to the possibility that the present can change the past. This surely seems to be the purport of allowing that differences in the time as of a sentence is evaluated can yield differences in the sets of times at which that sentence is true. Possibly, at least, a sentence evaluated as of t1 might take neither T nor F at t1 and yet when later evaluated as of t2 come to take the value T not only at t2 but also, incredibly in may seem, at t1. Yet wait: this will alter the past structure of the universe. How can this be?
Our brief answer is this: authors do it all the time. Consider the case of Anthony Trollope, whose character Miss Mackenzie grew and altered in surprising ways—surprising to Trollope, that is—in the course of his writing the novel bearing her name. Trollope reports that Miss Mackenzie was ‘written with a desire to prove that a novel may be produced without any love’ (Trollope Reference Trollope1883: 150). He did not mean without love on the part of the author, but rather on the part of the characters, so that the novel would present a story wholly free of any depiction of love. Alas, he failed: ‘In order that I might be strong in my purpose, I took for my heroine a very unattractive old maid, who was overwhelmed with money troubles; but even she was in love before the end of the book, and made a romantic marriage with an old man’ (Trollope: Reference Trollope1883: 150). In their Guide to Trollope, Gerould and Gerould note that, consequently, Miss Mackenzie, ‘[o]riginally titled The Modern Griselda, is an amusing example of the way that Trollope's characters, as they developed in his mind, contrived to write their own story—in this instance quite against his will’ (1976: 164). While we do not know, as some of his critics have contended, that Miss Mackenzie gained the attributes of loving and being loved as a result of her own contrivance and ‘quite against [Trollope's] will’ (Gerould and Gerould Reference Gerould and Gerould1976: 146), we are certainly prepared to take Trollope at his word when he reports that the character changed appreciably in the course of his writing the novel, which took place between May 22 and August 18, 1864, before it was published the following year.
Consider, then, the proposition that Miss Mackenzie is loveless. True or false? The correct answer seems to us to be given two-dimensionally. The first dimension, which is governed by the as-of operator, reflects the time during which Trollope wrote, and the second, which is governed by the at operator, reflects the temporal dimension within the novel. By our lights, as of the time Trollope began writing, Mackenzie was loveless at all the times of the novel, but as of the time Trollope finished writing, Mackenzie is loveless at some time t1 but in love and beloved at a later time t2. Hence, in table 1, we present the following truth assignments for the sentence ‘Miss Makenzie is loveless’:
How can the proposition change its truth value at t2 between May 22 and August 18? The answer is simple: willingly or not, as of some time between May 22 and August 18, Trollope changed his mind about whether Miss Mackenzie had this feature at t2. Earlier he ascribed it to her; later he denied it of her.
Did Trollope somehow engage in retrocausation? Well, nothing he did caused any part of the fixed past to be other than it was: it remains true as of May 22 that as of t1 it was T at t1 that Miss Mackenzie was loveless. Still, something he did as of a later time determined that Miss Mackenzie was not loveless at t2, which is just to say that he changed the features of the character. It turns out, contrary to his original authorial intentions, Miss Mackenzie came to be other than loveless all along. The character, that is, was intended to be loveless; instead, she—happily, we think—came to love and be loved. We do not feel the need to take a stand on whether this form of determination is causal determination, but we are skeptical that it needs to be, not least because we believe there are all manner of noncausal determination relations (as does Kim Reference Kim1974).
If someone were to insist that, according to her view of things, the only form of determination relations are temporally mono-dimensional causal relations, we would not quibble with her about words. She may, if she wishes, call Trollope's act of creation retrocausation. But if she then goes on to insist that Trollope could not do what he in fact did on the grounds that causation is mono-dimensional and retrocausation is impossible because it unfolds in a mono-dimensional temporal framework, we would point out she has now begged the question. Moreover, in order to do so, she has simply denied the facts as Trollope reports them. The facts, she assures all, cannot be as Trollope understood them to be. We disagree. We think that in this case we have a clear instance of authorially parameterized determination, which is to say determination requiring a two-dimensional semantics where truth evaluations must be made both as of a time and at a time.
Summing up so far, then, authorial intentions and actions must be parameterized along two temporal dimensions, answering to the semantic framework provided by the as-of operator and the at-operator. Authorially parameterized determination provides not only a possible framework for a semantics of truth in fiction, but, we submit, if not the only then easily the best and least revisionary framework within which actual reports of actual authors make any ready sense.
7. Writing the Stories of Our Lives
Suppose we have come this far in agreement. If we have, then it will seem that in some cases we must advert to a two-dimensional framework for determination of truth values. Still, to allow this much is not yet to agree that all instances of choice are subject to a two-dimensional semantic framework; perhaps some are, including most notably those involved in authorial contexts, but these are arguably special contexts, not readily generalized beyond the realm of fiction.
To the degree that this is so, it remains entirely reasonable for those who have traveled this far with us to want to get off the bus before going any further. Perhaps, one might allow, what we say about authors is true, or true enough, but once shorn of its most fanciful flights, the narratological suggestion that we are authors and our lives are the stories we write is simply not true. At any rate, it is not literally true. Works of fiction are, well, fictions; lived lives are not fictions but rather actual progressions of events in the mono-dimensional framework of the actual, non-fictitious world. When all is said and done, those so disposed will conclude, the story of my life is not my life, but rather just that, a story—the story of a life, and nothing more. The events of my life are not true or false, but are rather occurrences in a world, serving as truth makers for true reports or showing some stories to be accurate and others not; in this sense the series of events constituting my life are not themselves truth evaluable in any ready way. So, the detractor concludes, since our lives are not stories and we are not in any literal way authors, then whatever truth there may be to authorially parameterized causation has no bearing on our lived lives and neither, then, on our problem with free will. Put succinctly, authorially parameterized causation is relevant to the problem of free will only if we are in fact authors, literally authors; but we are not, in fact, authors, except in the most figurative sort of way; so, whatever the two-dimensionalist semantic framework proposed may tell us about authors and their ways, it tells us nothing at all about our own situation as free agents.
We make three responses to this objection that range from the conciliatory to the speculative to the categorical. First, in a conciliatory tone, our goal in this paper has not been to show that a two-dimensionalist framework correctly and indisputably captures our situation as agents. Rather, our primary aim has been to establish a more modest, two-fold thesis. First, we have shown that whatever its relevance to the problem of free will, a two-dimensionalist temporal framework for truth determination can be given a coherent articulation. Second, we contend that it provides the resources for an account of libertarian freedom that at least in principle accommodates determinism. Our goal, in other words, has been to present an account of what human freedom might be. Whether humans in fact have two-dimensional authorial freedom is a matter for another occasion.
That conceded, we do note that even our modest goal has one arresting consequence, namely that once a coherently articulated two-dimensional account of freedom is in view, a proponent of the mono-dimensional approach can no longer simply assume the correctness of her formulation of the problem. This much, we maintain, is already of significance: it shows that the traditional problem of free will is not at all conclusive. Our proposal points, in effect, to a way off the merry-go-round that does not merely change the subject.
In a more speculative vein, when we turn to the more demanding question of whether the world in principle permits of a true temporally parameterized understanding of agency, we point to developments within physics that have led a number of physicists and philosophers to question the assumption that causes always precede effects. Huw Price, for instance, has argued that quantum physics along with certain plausible ontological assumptions entails retrocausation, that is, that later events cause earlier events (Reference Price2008). Price's argument depends on a certain view of causality. But even without any particular view of causality, quantum physics raises the possibility at least that the future determines the past. In his famous and provocative essay ‘Information, Physics, Quantum: The Search for Links’, John Archibald Wheeler, attempting to understand the relationship between the quantum and the classical worlds, articulates a theory of what he calls ‘a participatory universe’ (Reference Wheeler and Hey1999: 311), according to which past events are not fixed as past until encountered: ‘“Measurements and observations”, it has been said, “cannot be fundamental notions in a theory which seeks to discuss the early universe when neither existed”. On this view the past has a status beyond all questions of observer-participancy. It from bit [Wheeler's participatory theory] offers us a different vision . . . The photon that we are going to register tonight from that four billion-year old quasar cannot be said to have had an existence “out there” three billion years ago . . . until we have fixed arrangements at our telescope [and] register tonight's quantum as having passed to the left (or right) of the lens . . .' (Wheeler Reference Wheeler and Hey1999: 320–21; see also Bars and Terning Reference Bars and Terning2010: 67–81; Chen and Goldstein Reference Chen, Goldstein and Ben-Menahem2022: 60–61).
Although they, too, are speculative, Wheeler's ideas about a participatory universe in which present states can alter past states have been given experimental credence since the time he wrote the essay. E. Megidish et al. (Reference Megidish, Halevy, Shacham, Dvir, Dovrat and Eisenberg2013) demonstrate the entanglement of two photons that did not temporally overlap. Hence, the collapse of a later photon can, per their experiment, alter the state of the earlier photon.
We are of course not the first to note the potential relevance of quantum physics to accounts of freedom. Our view is not, however, either of the earlier views advocated by Robert Nozick (Reference Nozick1981) and Robert Kane (Reference Kane1999), who have made quantum indeterminacy a cornerstone of their accounts of libertarian freedom. And we do not wish to enter into a lengthy discussion of all the issues that such a potential linkage raises (see, though Loewer Reference Loewer1996). This is because we are not in any way arguing that quantum indeterminacy provides a locus for libertarian freedom. Rather, we are urging only a more limited point, in response to an objector who insists that we are not, and could not be, authors of our lives in anything but a metaphorical sense.
Our response: we insist that this objection has the force it does because of an implicit acceptance of a mono-dimensional understanding of a semantics for tensed statements, which is easily shown not to be the only possible framework for understanding human action. It is threatened as being actually incorrect by some recent developments within physics. At the very least, those who push this sort of response are, then, on shaky empirical grounds; and certainly, some aspects of quantum theory serve to reinforce—if reinforcements are wanted—the thought that those responding to our proposal in this way cannot simply rely on some presumed a priori justification for mono-dimensionality.
Further, in replying to the objector by appeal to physics in this way, we also mean to circumvent the following seemingly serious objection to our view. Let us say we are authors. ‘So what?’ the objector responds. ‘If determinism is true, then it follows that the physical actions I perform are entirely fixed by the laws of nature together with the past state of the universe. What I will do in five or ten minutes follows from necessity from what has already happened. Of course, I can interpret what I do as a product of my ownership, but that tells us only why I think I am free. It does not show that I am free, and it fails to explain even how I might be free. And, indeed, how might I be? If all my actions are fully explicable in terms of past events together with the laws of nature using a mono-dimensional framework (as per Van Inwagen Reference Van Inwagen1983), what does the fact that we can understand my actions differently using a two-dimensional framework do for us?’
The response to this objection should not be hard to fathom, given where we have come so far. If the two-dimensional semantic framework is correctly applied to the case of human action, then it is false that all my actions are fully explicable in terms of past events together with the laws of nature using a mono-dimensional framework. The easy confidence that a mono-dimensionalist has in this picture cannot be permitted to pass unremarked and undefended. After all, it may be that that the correct resolution of the puzzle of human freedom requires the same type of radical theoretical reconstruction that is needed in order to come to grips with the quantum realm—just as Wiggins (Reference Wiggins and Honderich1973) and Thomas Nagel (Reference Nagel1986) surmised.
We end our response to the objector, then, with a categorical riposte. When Fischer presents himself as an author writing the story of his life, we understand him to mean just what he says. He is not speaking metaphorically, but rather introducing a notion of guidance control into our discourse about free will. We think there is a sense in which he is both right and too timid. We are not authors as Trollope is the author of Miss Mackenzie. We are nonetheless authors, we think, in a distinctive and recognizable sense: our decisions not only add to the given past in a linear way, but also shape and structure that past by giving it new hues and complexions brought about by decisions made subsequent to acts earlier committed; some things about the past become true as of occurences in the present. The two-dimensionalist framework we have provided renders precise the truth conditions implicated in our writing the stories of our lives. When, as of t2, we omit to do what we promised to do at t1, then as of t2 we make it true that at t1 we have made a promise that we will break. As of t2, we make it the case that what we said contributes to the story of our lives in a certain way, a way other than it would have contributed had we made good on our promise. More generally, when our present decisions render past acts coherent by embedding them in an overarching narrative whole, then in a straightforward way we fashion the story of our lives. This fashioning, we maintain, is not metaphorical; and it is not fantastic or even unfamiliar. Nor is it, in any sense we find the least threatening, unfree. It is, rather, we think, what is reasonably and defensibly meant by Fischer when he says, ‘When I act freely, I write a sentence in the story of my life’ (Fischer Reference Fischer2009: 167).
We appreciate that the view proposed in this essay will find a hostile reception among those beholden to a rigid understanding of a cluster of concepts usually thought to generate a problem of free will: a linear model of causation; a mono-dimensional temporal semantics; and a non-paramaterized approach to truth conditions. Our approach rejects these presuppositions and so rejects the framework needed to generate their problem, first by showing these presuppositions are hardly necessary and second by showing, more strongly, that we have no good reason to accept them as true. By urging an approach differing in crucial respects from the framework assumed by those operating within the strictures of an ordinary non-paramaterized temporal semantics, our account presents an alternative in terms of which human freedom makes ready sense. This, we conclude, already provides one powerful consideration in its favor: the ordinary analysis has proven to be an unwitting straightjacketing to those who adopt it, one that renders precise notions that might have been—and we think should have been—rendered precise in other ways. Our way has at least this benefit: it adds to the process of precisification the attractive thought that we are the authors of our own lives. We think, all told, the view defended in this paper is never outlandish or otherwise outré. On the contrary, it is phenomenologically apt; it is metaphysically coherent; and it relies upon a perfectly intelligible semantic framework.
So, is all of this sufficient to demonstrate the correctness of the account of freedom in this paper? Perhaps not. This is not, however, as we have said, our immediate goal: we are not trying to demonstrate that parameterized freedom exists. For what it may be worth, we do believe that we are free, and we further believe that an authorially parameterized approach articulated within a two-dimensional semantics for temporal predicates provides the most fruitful framework for characterizing human freedom. That said, our immediate goal has been comparatively modest: we have provided reason to conclude that the metaphysical presuppositions motivating the problem of free will that has so exercised philosophers of our era are just that, presuppositions and no more. When looked at from the vantage point provided by a narratological framework, they dissolve under their own weight. So, then, we conclude where we began, by endorsing a sentiment of Fischer's, if only by inverting his expression of it: when we write the stories of our lives, we act freely.