I. Introduction
Moral responsibility, understood in the backward-looking, retributive sense associated with basic desert,Footnote 1 is generally thought to require free will. However, philosophers remain deeply divided on what free will is, what kind of agents could have it, and the conditions under which it could be exercised. One enduring point of dispute is how to answer this seemingly simple question: Is it possible for humans with free will to exist when determinism is true? Indeed, the so-called “problem of free will and determinism” is often characterized as the problem of settling whether the answer to this coexistence question is “yes” or “no.”
In asking whether it is possible for a free human agent to exist when determinism is true, the coexistence question asks, in effect, whether the truth of determinism is correlated with the absence of free human agents. So, for ease of reference, let us refer to the challenge of answering the coexistence question as the correlation problem of free will and determinism. A new question arises for those who answer “no” to the coexistence question, namely: Why is it impossible for a normal human to exercise free will when determinism is true, that is, what accounts for the correlation between determinism and the lack of free human agents? This question captures a distinct explanation problem of free will and determinism. A free-will theorist looking to solve the explanation problem must do two key things. First, she must identify at least one necessary condition on free will which cannot be satisfied by a normal human when determinism is true. Second, she must spell out what keeps a normal human from satisfying that condition when determinism is true. Only by meeting both of these challenges does the free-will theorist provide a metaphysical story that explains the correlation between the truth of determinism and the absence of free human agents.
Since the distinction between the correlation and explanation problems of determinism is an unfamiliar one, some readers may be skeptical of its philosophical significance. In most discussions of the problem of free will and determinism, it is simply assumed that theorists who answer “no” to the coexistence question must accept—irrespective of their preferred characterization of the unmet condition—that determinism itself plays a role in keeping humans from exercising free will. From this perspective, the only explanatory question left open by a “no” reply to the coexistence question is how to flesh out the necessary condition that cannot be met by a normal human when determinism is true. It is unsurprising, then, that most work on the explanation problem addresses only the “unmet condition” part of the problem.Footnote 2 Still, it is a fallacy to infer from the mere fact that two phenomena are correlated that this correlation holds because there is a direct causal relation between the two. Likewise, we should be wary of inferring from the (purported) fact that the truth of determinism is correlated with an absence of free human agents that determinism itself is an essential part of the explanation for this correlation.
In this essay, I offer a new approach to the explanation problem which addresses the question “What robs humans of free will when determinism is true?” as a metaphysically substantive and interesting question in its own right. Using a familiar form of abductive reasoning, I argue that philosophers who grant that it is impossible for humans to act freely when determinism is true may, in virtue of accepting different accounts of free will, be committed to substantively different accounts of what undermines free will when determinism is true. In broad strokes, the abductive approach taken here starts by identifying a dependent variable, namely one’s status as a free agent, and specifying its candidate values as free or unfree. Next, an independent variable is selected and its values are specified. Using this method, we can assess whether any change in the value of the independent variable affects the value of the dependent variable, that is, changes one’s status as a free agent from unfree to free. This means that we can use the abductive strategy to identify the best explanation for the lack of free human agents in worlds in which determinism is true. The main upshot of this abductively-driven investigation is that different theories of free will imply substantively different views about what undermines free will when determinism is true and, by extension, whether the very same freedom-undermining feature is also a fundamental feature of possible worlds in which determinism is false.Footnote 3
This essay proceeds as follows. I begin by spelling out one popular notion of determinism and then place the type of laws that determinism describes in a larger “leeway-based” taxonomy of laws of nature. Using this taxonomy, I provide a more careful characterization of the correlation and explanation problems of free will and determinism. Next, I apply the abductive strategy to three distinct theories of free will: (i) an “agent-causal” account suggested by Derk Pereboom, (ii) an “event-causal” account promoted by Robert Kane, and (iii) an impossibilist “ultimate sourcehood” account from Galen Strawson. I argue that each yields a distinct metaphysical explanation for the lack of free human agents in worlds in which determinism is true. If this is right, it means that some broadly libertarian-friendly accounts of free will (that is, accounts which assert that free will is metaphysically possible, but only when determinism is false) must reject the traditional incompatibilist view that deterministic laws are metaphysically incompatible with—undermine, rule out, destroy—human free will.Footnote 4 I close by suggesting some ways that the abductive strategy might be used as a tool in future work on free will and moral responsibility.
II. A Leeway-Based Taxonomy of the Laws of Nature
There are many doctrines that go by the name “determinism,” and there is increasing debate about which of these doctrines should be used when characterizing the problem of free will and determinism. William James was among the first to use the term “determinism” in the context of the free will debate, explaining it in his seminal “Dilemma of Determinism” as follows:
What does determinism profess? It professes that those parts of the universe already laid down absolutely appoint and decree what the other parts shall be. The future has no ambiguous possibilities hidden in its womb; the part we call the present is compatible with only one totality. Any other future complement than the one fixed from eternity is impossible. The whole is in each and every part, and welds it with the rest into an absolute unity, an iron block, in which there can be no equivocation or shadow of turning.Footnote 5
With no mention of causation, determination, or laws of nature, James describes a world in which one unique future is determined—unconditionally fixed, settled, guaranteed—by factors that were in place before that future unfolds. The doctrine of predetermination that James describes, call it “Jamesian determinism,” can be traced back to at least the Stoic view that all things are subject to a naturalistic, global “fate.”Footnote 6
If we were to translate the metaphorical language James uses to describe the doctrine of determinism into more contemporary language, we might restate his description of a world with “no ambiguities hidden in its womb” as the claim that the laws of nature that obtain in the world rule out all actual-sequence alternative possibilities. Although many rival conceptions of determinism are available, variants of Jamesian determinism continue to play a central role in the contemporary free will debate. This is, it seems, not because philosophers continue to believe that this radical doctrine of predetermination might be true (they generally don’t), but rather because it is rhetorically useful to use a doctrine that describes worlds that permit absolutely no actual-sequence leeway.
According to Jamesian determinism, there is something present at each timeslice that fixes what will happen next, such that there is, at any arbitrary moment in time, only one future that may be added to the past. In the free will literature, we generally refer to this future-fixing “something” with the phrase “laws of nature.”Footnote 7 In order to set the stage for the abductive strategy explored below, we need to clarify what these laws would have to be like in order to rule out all actual-sequence leeway in the way that James describes, and which types of actual-sequence leeway might be available if other sorts of laws were to obtain. Toward that end, I will classify laws of nature based on three characteristics: (1) strength, (2) probability, and (3) potency.
Let us first consider the strength of the laws of the nature. Within the free will debate, it is often assumed that the laws of nature are such that they, given the facts of the past, constrain what can happen in the future. Famously, though, some philosophers (for instance, proponents of the Humean Regularity theory) deny that the laws of nature fix the future in any metaphysically robust way. Following John Perry, let us say that strong laws fix the future; weak laws do not. Footnote 8 While strong laws (together with facts of the past) fix or somehow constrain what unfolds in the future, weak laws place no restriction whatsoever on how the future will unfold. Indeed, weak laws are themselves determined by what has unfolded in the past, rather than the other way around. Following standard practice, I will limit the present discussion to strong laws of nature (though the abductive strategy may be used to address issues related to weak laws as wellFootnote 9).
Strong laws may be subdivided based on considerations of objective probability. For present purposes, let us classify strong laws as deterministic when the laws support (underwrite, guarantee, and so on) that every event has an objective antecedent probability of either 1 or 0. By contrast, let us classify strong laws as probabilistic when the laws assure that all events have an objective probability on the spectrum between 0 and 1, with at least some events having an objective probability of neither 0 nor 1.Footnote 10 Both deterministic strong laws and probabilistic strong laws fix the future (in virtue of their strength), but only deterministic strong laws fix one unique future.Footnote 11
Finally, strong laws may differ in what we might call potency.Footnote 12 Jamesian determinism describes a world in which one unique future is completely and unalterably assured to occur given the facts of the past and the laws (that is, the elements of nature that account for the evolution of the world). To borrow an apt phrase from Saul Smilansky, the future is just an “unfolding of the arbitrarily given.”Footnote 13 In effect, Jamesian determinism says that there is literally no condition on which the unique future fixed by the deterministic strong laws (given the facts of the past) would fail to occur. Jamesian determinism, then, must not be conflated with the more modest thesis that the laws fix a unique future given the past, assuming that nothing gets in the way of the laws carrying out their work. This is worth noting because descriptions of strong laws of nature typically include (covertly or overtly) a ceteris paribus clause that specifies the conditions under which strong laws might fail to account for the evolution of the physical word. For example, one might say that deterministic strong laws fix the future assuming the causal closure of the physical system in which the laws hold. Given such a clause, a system may be “opened” by something (such as a divine intervention by God) without any laws being broken or violated. However, if a system with deterministic strong laws is opened in this way, then the unique future that would have obtained if the system had remained closed will not occur; a future that is distinct from the one determined by the laws and initial state of the universe will occur instead. Likewise, one might say that a set of deterministic strong laws ensures one unique future on the condition that the physical world continues to exist. Given this condition, it is possible that deterministic strong laws obtain in a universe and yet that universe ceases to exist (for example, because it is destroyed by God) before the unique future determined by the laws and the past completely unfolds. When deterministic strong laws have such conditions, it is technically false that the laws, given the past, ensure that one unique future will come pass.
Jamesian determinism, by contrast, has no ceteris paribus clause.Footnote 14 This doctrine describes deterministic strong laws that unconditionally bring about one unique future given the past; the laws make it the case that one unique future will unfold, full stop. Jamesian determinism, that is, describes laws that cannot be broken, suspended, circumvented, changed, or the like. Even more radically, this doctrine implies that the laws ensure the future existence of the physical world, not merely what the future will be like on the condition that the physical world continues to exist. To track the potency distinction between deterministic strong laws, let us say that unconditional deterministic strong laws unconditionally fix the future, but conditional deterministic strong laws do not.Footnote 15 Notably, probabilistic strong laws may likewise differ in potency. Unconditional probabilistic strong laws would be sufficiently potent to rule out any changes to the objective probabilities affixed to future events by the laws, but conditional probabilistic strong laws might, for example, allow God to intervene and change the probabilities given to future events based on the laws and past alone.
By characterizing laws in terms of strength, probability, and potency, we can arrange laws of nature on a spectrum based on the type(s) of actual-sequence leeway they permit.Footnote 16 Unconditional deterministic strong laws, which ensure a total lack of actual-sequence leeway, mark one extreme on this spectrum; weak laws, which place no constraint whatsoever on the actual-sequence leeway available in the world, reside at the opposite end of the spectrum.Footnote 17 With these leeway-based distinctions in hand, we are ready to flesh out the abductive approach to the explanation problem of free will and determinism.
III. The Explanation Problem of Free Will and Determinism: An Abductive Approach
For the sake of argument, let us characterize the problem of free will and determinism as arising from concerns about the relationship between human free will and unconditional deterministic strong laws of nature.Footnote 18 We can now refine our statement of the correlation problem as the challenge of answering this coexistence question: Is it metaphysically possible for a normal human to exercise free will when unconditional deterministic strong laws obtain?Footnote 19 For ease of reference, I will call the view of those who answer “yes,” that is, the view that human freedom is compossible with unconditional deterministic strong laws, anthropocentric compossibilism; I will call the view of those who answer “no” anthropocentric incompossibilism.Footnote 20 We may now state the question at the core of the explanation problem of free will and determinism more carefully as well, namely: Why is it metaphysically impossible for a normal human to exercise free will when unconditional deterministic strong laws obtain? This explanation problem arises for those who hold that anthropocentric incompossibilism is true.
In this section, I will assume that anthropocentric incompossibilism is true, and focus on how abductive reasoning might be used to answer the challenge of identifying precisely what does the freedom-undermining work when unconditional deterministic strong laws obtain. In addressing this explanatory challenge, we are, in effect, looking for competitors to standard anthropocentric incompatibilism, the explanatory view that unconditional deterministic strong laws are metaphysically incompatible with—preclude, undermine, destroy, or stand in some similar antagonistic relationship to—human free will. Put another way, anthropocentric incompatibilists hold that necessarily, if unconditional deterministic strong laws obtain, then no human exercises free will at least in part because the laws are deterministic. To identify the alternatives to the explanation proposed by anthropocentric incompatibilism, I will use a familiar form of reasoning by inference to the best explanation: I will first identify a dependent variable and an independent variable (which has more than one value), and then I will change the value of the independent variable and assess the effect on the value of the dependent variable.
For purposes of our investigation, the dependent variable of interest is one’s status as a free agent, where the allowed values are free and unfree. Choosing the right independent variables is a bit trickier. Once again, the explanatory question we are looking to answer is this: Why is it metaphysically impossible for a normal human to exercise free will when unconditional deterministic strong laws obtain? The scope restrictions on this question (in italics) conversationally imply that unconditional deterministic strong laws and being human are somehow relevant to free will. Indeed, many philosophers have suggested that free will requires at least some sort of actual-sequence leeway, but no such leeway is permitted by unconditional determinsitic strong laws. As such, it seems potentially fruitful to select actual-sequence leeway permitted by the laws as an independent variable. With the leeway-based taxonomy in hand, we have several different values for this variable. As such, we can easily take up the task of assessing whether any change in its value changes the value of our freedom-status dependent variable.
It is less obvious which independent variable would be best for assessing the relation between one’s having the property of being human and one’s status as a free agent. Certainly, one’s having or lacking certain bits of human DNA does not seem likely to be a freedom-relevant feature of an agent, so “type of DNA” is not a very promising independent variable. However, many philosophers have suggested that some special superhuman properties (that is, properties that no normal human could possibly manifest) are required for free will. That being so, we could use the abductive strategy to assess whether there is any superhuman property that an agent might have such that this superhuman agent could exercise free will though a normal human could not. Values of this independent variable would include such features as being immortal, being eternal,Footnote 21 having the ability to travel back in time,Footnote 22 and other godlike properties. Although we do not have time here to consider the full range of values of the two independent variables I have selected, my application of the abductive strategy will make it clear enough how one would go about doing so in the future. The goal here is to apply the abductive strategy enough to demonstrate its usefulness in clarifying and perhaps adjudicating the in-house debate among anthropocentric incompossibilists about what makes anthropocentric incompossibilism true.
To begin, let us consider the view that there is some type of libertarian-friendly “law-overriding” condition on free will. For example, Pereboom holds that an agent could act freely in a world with strong laws by bringing about an event that would not have been produced by the strong laws of nature and facts of the past alone; in effect, he suggests that one must be able to perform a miracle relative to the laws of nature in order to exercise free will when strong laws obtain.Footnote 23 Pereboom’s commitment to the plausibility of the law-overriding condition has waned over time, but he remains sufficiently committed to the plausibility of this condition that he appeals to it in order to reject Galen Strawson’s view that it is metaphysically impossible for someone (human or otherwise) to exercise free will.Footnote 24
Given that Pereboom is often classified alongside Strawson as a free will impossibilist, it is worth explaining how Pereboom’s commitment to the law-overriding condition separates his position from Strawson’s. In the broadest strokes, Strawson contends that our intuitive standard of freedom-relevant sourcehood is so demanding that it is impossible for anyone or anything to satisfy it. Strawson spells out the sourcehood condition as a “starting point” condition:
In order for one to be truly or ultimately responsible [i.e., responsible in a basic-desert sense] for how one is in such a way that one can be truly responsible for what one does, something impossible has to be true: there has to be, and cannot be, a starting point in the series of acts of bringing it about that one has a certain nature; a starting point that constitutes an act of ultimate self-origination.Footnote 25
In very broad strokes, Strawson holds that free will is impossible because—irrespective of the laws of nature—there is an ultimate sourcehood condition on free will that no possible being could satisfy. Pereboom agrees that there is a very demanding necessary sourcehood condition on free will, but he contends that a law-overriding agent could be the freedom-relevant source of her actions without being the ultimate “starting point” of her actions (in Strawson’s sense of that phrase).Footnote 26 That is, Pereboom rejects impossibilism even though he holds that it is metaphysically impossible for normal humans to exercise free will. In short, Pereboom and Strawson are both impossibilists, but of slightly different sorts. Pereboom is an anthropocentric impossibilist; he is not an impossibilist tout court. By contrast, Strawson is a full-blooded impossibilist.
At first blush, the difference between Pereboom’s anthropocentric impossibilism and Strawson’s (unqualified) impossibilism may seem trivial—especially if our central concern is whether or not humans have free will. However, while Strawson and Pereboom agree that it is impossible for humans to exercise free will, their respective characterizations of the source condition on free will commit them to very different metaphysical views about why humans lack free will when determinism is true. To bring their disagreement into the light, let us assume for the sake of argument that Pereboom’s proposal is right: there is a necessary law-overriding condition on free will and at least one possible agent satisifes it (together with all other necessary conditions on free will). Given this assumption, let us ask: What is the best explanation for the lack of human free action when unconditional deterministic strong laws obtain?
As Pereboom rightly points out, there is good (though not conclusive) empirical evidence for thinking that we, as normal humans, do not satisfy his proposed law-overriding condition.Footnote 27 He suggests two ways of accounting for this failure. First, it might be that the strong laws of our world admit of law-overriding, but normal humans lack the intrinsic properties required to override them (in the same sense that a normal cat lacks the properties required to do calculus). Second, it might be that normal humans have the requisite properties to override strong laws, but the potency of the laws in our universe keeps us from exercising our law-overriding potential.Footnote 28 These are importantly different suggestions, so let us address them separately.
On the first proposed view of human nature, humans fail the law-overriding condition in virtue of their intrinsic properties; we, due to our nature, are subject to the laws. On this view of human nature, no change in environmental factors would improve our prospects for satisfying the law-overriding requirement. More precisely, no mere change in actual-sequence leeway available in the world would change a human’s status from unfree to free. To confirm this, let us consider Norm, a normal human living in a world with unconditional deterministic strong laws. If, keeping all else fixed, we were to change the laws of nature in Norm’s universe to conditional deterministic strong laws, this change alone would not change his status as a free agent (just as moving a cat from a world with unconditional deterministic strong laws to a world with conditional deterministic strong laws would not improve its mathematical reasoning). Likewise, we would not change Norm’s status from unfree to free merely by changing the laws of his universe to unconditional probabilistic strong laws or, from there, changing the laws to conditional probabilistic strong laws. In short, assuming that law-overriding is required for free will, no change in the actual-sequence leeway permitted by the laws changes Norm from an unfree agent to a free one. If this is true of Norm, then it is true of any normal human. In sum, when we assume that free will has a necessary law-overriding condition and that humans are (by their nature) subject to the laws, abductive reasoning supports the anti-incompatibilist conclusion that necessarily, normal humans cannot satisfy the necessary law-overriding condition on free will because of their intrinsic properties, irrespective of the laws of nature.Footnote 29
Notably, it follows logically from the thesis that necessarily, normal humans cannot satisfy the necessary law-overriding condition on free will irrespective of the laws of nature that the following conditional proposition is true: Necessarily, if strong unconditional deterministic laws obtain, then humans do not exercise free will. However, this material conditional is completely silent on what undermines free will; it expresses mere anthropocentric incompossibilism. This highlights that we must take care when using logical entailment claims to characterize competing metaphysical views about the relationship between free will and the laws of nature, as they may be misleading (and generally ill-suited for use) in explanatory contexts.
Interestingly, if we do not place an anthropocentric restriction on the scope of possible agents under consideration, one may appeal to Pereboom’s law-overriding condition to argue that unconditional deterministic strong laws pose a threat to free will. To see this, recall the initial Norm scenario in which Norm is a normal human living in a world with unconditional deterministic strong laws. Now, consider a new case in which we replace Norm with Super-Norm, an agent who satisfies all the necessary conditions on free will that a normal human can satisfy and, in addition, has the sort of superhuman properties required to override strong laws of nature. Would Super-Norm have free will even though Norm does not? Even if we accept the law-overriding condition, we must answer “no.” Although Super-Norm has the properties required for law-overriding agency, we did not change the background assumption that the strong laws of nature in his universe are unconditional. Unconditional strong laws are, by definition, impossible to override. As such, anyone living in a world with unconditional deterministic strong laws is subject to the laws of nature and, so, fails the law-overriding condition and, in turn, cannot exercise free will.Footnote 30 But if we change Super-Norm’s environment just slightly, replacing the unconditional deterministic strong laws of his universe with conditional deterministic strong laws (keeping all else fixed), we could thereby give him the type of actual-sequence leeway he needs in order to exercise his law-overriding powers and, by extension, his free will—even though the laws are still deterministic.Footnote 31 Again, it seems that the law-overriding condition leads us to the anti-incompatibilist conclusion that whether the laws are deterministic or not is irrelevant to free will.
To confirm the conclusion that whether strong laws are deterministic or probabilistic is irrelevant to free will on the assumption that freedom has a law-overriding condition, we can change the unconditional deterministic strong laws of Super-Norm’s universe to unconditional probabilistic strong laws. A world with unconditional probabilistic strong laws permits a distinct sort of actual-sequence leeway, but everyone in a world with unconditional probabilistic strong laws is still subject to the laws in precisely the same sense as they would be if unconditional deterministic strong laws obtained. It is impossible for anyone to satisfy the law-overriding condition on free will in a world with unconditional strong laws of any kind.Footnote 32 However, if we keep everything else about the previous Super-Norm case fixed, except that we change the original unconditional probabilistic strong laws to conditional probabilistic strong laws, we can restore the special sort of leeway required for Super-Norm to exercise free will. Again, it seems to be the potency of the laws, not the probabilities they assign to future events, that matters to free will on Pereboom’s view.
In sum, when we assume that there is a law-overriding condition on free will, the change from deterministic strong laws to probabilistic strong laws (values of our independent variable) does not change one’s status as a free agent (the value of the dependent variable): however, the change from conditional strong laws to unconditional strong laws does change the freedom-status of those who have the requisite properties for overriding the laws. This means that the distinction between conditional strong laws and unconditional strong laws is freedom-relevant to agents who have the properties required to override overridable laws. However, the distinction between deterministic strong laws and probabilistic strong laws is not a freedom-relevant divide for law-overriding agents. As such, for those possible beings who possess all of the intrinsic properties needed to satisfy the necessary conditions on free will, free will is metaphysically incompatible with (undermined by) unconditional deterministic strong laws because of their potency, but not even in part because of the probability assignments they impose on the world. Once again, the law-overriding condition gives us reason to accept the unrestricted incompossibilist view that it is impossible for anyone (including humans) to exercise free will when unconditional deterministic strong laws obtain. However, the law-overriding condition leads us to reject the traditional incompatibilist view that deterministic strong laws undermine free will at least partly in virtue of being deterministic.
For illustrative purposes, let us now return to Pereboom’s second view of human nature, according to which normal humans have the requisite intrinsic properties to override laws. On this assumption, normal humans are not relevantly different from Super-Norm. Applying the lessons from the discussion of Super-Norm above, it is clear that the conjunction of Pereboom’s second view of human nature and his law-overriding condition implies that anthropocentric incompatibilism is false: unconditional deterministic strong laws keep normal humans from acting freely in virtue of their potency, not in virtue of their being deterministic.Footnote 33
If we have correctly drawn out the implications of the combination of Pereboom’s preferred solution to the correlation problem and his acceptance of the law-overriding condition, then it seems that Pereboom’s view of free will is in tension with his distinctive explanatory view, hard incompatibilism. Roughly, hard incompatibilism is the anthropocentric view that deterministic strong laws undermine—or, more precisely, deterministic causation by factors beyond one’s control undermines—human free will in one way, while non-deterministic laws undermine human free will in a different, distinctive way.Footnote 34 As such, anthropocentric incompatibilism is a defining tenet of hard incompatibilism. In light of the abductive arguments above, however, it seems that so long as Pereboom accepts the law-overriding condition on free will, he must reject anthropocentric incompatibilism and, by extension, hard incompatibilism.Footnote 35
Of course, Pereboom’s law-overriding condition represents just one of many incompossibilism-friendly views about free will. For example, the event-causal libertarian, Robert Kane, proposes a different theory of human freedom. Kane’s view is very complex, so I cannot provide a proper summary within the confines here; but the fine-grained details of his account are not important for present purposes. Roughly, Kane argues that free will requires “self-forming actions” (SFAs), and these require probabilistic strong laws. SFAs take place only when an agent is deeply torn between doing an action A and an incompatible action B. More specifically, Kane contends that a person exercises free will when choosing A over B (or vice versa) only if that person’s experience of indecision is reflected in the objective physical probabilities of those outcomes, such that there is a physical probability of 0.5 that he will do A and 0.5 that he will do B. Let us proceed on the assumption that free will requires (at least one of) these SFAs.
At first glance, it may seem obvious that Kane’s view of free will supports anthropocentric incompatibilism.Footnote 36 Assuming Kane’s view, a normal human may exercise free will when probabilistic strong laws obtain, irrespective of whether the laws are conditional or unconditional. A person may perform an SFA in a world with conditional probabilistic strong laws, and (keeping everything else fixed) a person might just as well perform an SFA in a universe with unconditional probabilistic strong laws. To put the point another way, Kane’s view allows that people can exercise free will even when they are subject to strong laws of nature. So, given Kane’s view, changing the value of the “leeway” independent variable from conditional probabilistic strong laws to unconditional probabilistic strong laws does not change the freedom-status of the agent, that is, the value of the dependent variable. Changing the leeway variable from probabilistic strong laws to deterministic strong laws, however, does affect the value of the dependent variable, changing it from free to unfree. Kane, then, seems to embrace a view of free will that implies the truth of both anthropocentric incompossibilism and anthropocentric incompatibilism.
However, there is one complication worth considering before concluding that Kane—or a proponent of his proposed necessary conditions on free will—must embrace anthropocentric incompatibilism. Arguably, godlike beings might exploit the potency of conditional deterministic strong laws in order to create just the sort of probabilistic conditions prior to a decision that Kane claims are required for free will.Footnote 37 On this view, moving a person from a world with unconditional deterministic strong laws to a world with conditional deterministic strong laws (keeping all else fixed) could change a person’s status from free to unfree. As such, even an event-causal libertarian such as Kane may have reason to accept anthropocentric incompossibilism and yet reject the incompatibilist claim that deterministic strong laws undermine free will in virtue of being deterministic. Kane’s view arguably supports the view that laws of nature pose a categorical threat to free will only when they are the toxic combination of deterministic, unconditional, and strong.
Of course, both of the incompossibilism-friendly conditions on free will discussed so far have been subject to sustained and serious criticism. What if it turns out that every broadly libertarian-friendly theory of free will (that is, every theory that implies that both free-will possibilism and anthropocentric incompossibilism are true) is untenable? Or, to put the question more precisely: What is the best explanation for the lack of free human agents in worlds with unconditional strong deterministic laws if it turns out that incompossibilism is true, but libertarianism is false?
To assume that libertarianism fails is, in effect, to assume that there is no type of actual-sequence leeway that allows one to overcome whatever it is that keeps people from acting freely when unconditional deterministic strong laws obtain. As such, the failure of libertarianism might tempt us to embrace the common slogan “determinism hurts and indeterminism doesn’t help.” However, it seems that we now have reason to reject this slogan, for the abductive strategy implies that its exact opposite is true: if no sort of actual-sequence leeway “helps” an agent overcome the problem (whatever it is) in worlds without any actual-sequence leeway, then it is not the lack of leeway in such worlds that “hurts.” More technically, if there is no way to set things up so that, keeping all else fixed, changing the value of the “leeway” independent variable to something other than unconditional deterministic strong laws changes the value of the “freedom-status” dependent variable from unfree to free, then the abductive inference we should draw is that actual-sequence leeway (and the laws and/or other factors which constrain it) is irrelevant to free will.Footnote 38 In short, if free will cannot exist in a possible world with actual-sequence leeway, then (against the background assumption that incompossibilism is true) the abductive reasoning we have been exploring seems to imply that actual-sequence leeway is irrelevant to free will. If this is right, it seems a libertarian-friendly solution to the “What is the unmet condition on free will?” portion of the explanation problem is required to motivate an incompatibilist solution to the “What keeps that condition from being met?” part of the explanation problem. In other words, the abuctive approach seems to show that the plausibility of (anthropocentric) incompatibilism hangs on whether or not there is a successful libertarian-friendly account of (human) free will.
Even if we ultimately conclude that free will is impossible and actual-sequence leeway is irrelevant to free will, we would still face the explanation problem of free will and determinism. That is, even a free-will impossibilist who rejects incompatibilism must identify which necessary condition on free will goes unsatisfied in a world with unconditional deterministic strong laws, and what keeps that condition from being satisfied. For example, recall Strawson’s “starting point” source condition on free will. Strawson contends that no sort of actual-sequence leeway helps a person to satisfy this (or any other) necessary condition on free will.Footnote 39 If he is right, the abductive approach supports Strawson’s contention that his solution to the “unmet condition” portion of the explanation problem implies that strong laws of nature pose no threat to free will and, so, incompatibilism is false.Footnote 40 However, assuming that the laws of nature are irrelevant to free will, it is not entirely clear why Strawson’s starting point condition is impossible to satisfy. Is it because no one can be a causa sui, that is, a person who self-creates ex nihilo? It seems not, for to say that a causa sui would (counterpossibly) satisfy a necessary source condition on free will does not provide an adequate metaphysical explanation for why people who are not causa sui cannot satisfy the requisite source condition. Is it because people who are not causa sui are subject to a freedom-undermining type of “constitutive” or “endowment” luck, roughly luck in the way that one is mentally constituted at the time one performs an action?Footnote 41 Perhaps, although one might worry that such “luck” explanations do not get to the heart of the matter, even if they help us to grasp the basic metaphysical problem.Footnote 42 These are interesting questions and concerns that have, in my view, not received their fair share of attention. Perhaps further development of the abductive strategy will help us to identify and assess candidate replies.
IV. Conclusion
In this essay, I used a familiar form of abductive reasoning to shed new light on the traditional problem of free will and determinism. I began by teasing apart the correlation problem from the explanation problem of free will and determinism, which allowed us to focus on the explanation problem in its own right. By applying abductive reasoning to this explanation problem, I exposed the tight theoretical connection between one’s view about which necessary condition on free will is unmet when determinism is true and precisely what keeps that condition from being met. The abductive approach taken here gives us a higher resolution image of the debate, allowing us to see that philosophers who agree that free will and deterministic laws cannot coexist may disagree about why agents (of a certain sort) do not act freely when deterministic strong laws obtain. Most notably, it appears that only a narrow subset of libertarians have the resources to defend the incompatibilist position that deterministic strong laws (at least partly in virtue of being deterministic) undermine free will, while other libertarians must positively reject the incompatibilist explanation for the incompossibility of human free will and deterministic laws.
In sum, it seems that the abductive strategy, coupled with the leeway-based taxonomy of laws, promises to be of interest to free will theorists of all stripes. For libertarians, it provides a tool that can be used to develop a more accurate statement of the metaphysical commitments of their preferred libertarian view and a clearer picture of how it differs from its libertarian-friendly rivals. For impossibilists, it provides a way to help shift the conversation away from the laws of nature toward other candidate metaphysical explanations for the truth of impossibilism. Finally, it helps compossibilists to appreciate that even if libertarians have been wrong in claiming that free will requires actual-sequence leeway, it may be that there is some other metaphysical justification for thinking that anthropocentric incompossibilism is true. If compossibilists are interested in demonstrating the resilience of compossibilism, they should be open to steel-manning their impossibilist rivals by helping impossibilists identify and develop non-libertarian/anti-incompatibilist worries about compossibilism. Perhaps, then, there is hope that the abductive strategy outlined here may, with further development, be used to push the metaphysical side of the free will and moral responsibility debate in new and interesting directions.
