Incompatibilism and the garden of forking paths

Let (leeway) incompatibilism be the thesis that causal determinism is incompatible with the freedom to do otherwise. Several prominent authors have claimed that incompatibilism alone can capture, or at least best captures, the intuitive appeal behind Jorge Luis Borges’s famous “Garden of Forking Paths” metaphor. The thought, briefly, is this: the “single path” leading up to one’s present decision represents the past; the fork-ing paths that one must decide between represent those possible futures consistent with the past and the laws of nature. But if determinism is true, there is only one possible future consistent with the past and the laws and, hence, only one path to choose from. That is, if determinism is true, then we are not free to do otherwise. In this paper, I argue that this understanding of the Garden of Forking Paths faces a number of problems and ought to be rejected even by incompatibilists. I then present an alternative understanding that not only avoids these problems but still supports incompatibilism. Finally, I consider how various versions of (leeway) compatibil-ism fit with the Garden of Forking Paths as well as the broader question of whether metaphors, however

intuitive, have any dialectical force in the debates over freedom.
The "Garden of Forking Paths," taken from Jorge Luis Borges (1941Borges ( /2018, is one of the most intuitive metaphors for how we ordinarily think of our lives and agency. You know the picture: while there is but a single path behind us, there are many before us. The path we take now will determine which paths are available further on, just as the choices we've made previously have brought us to our present predicament. And though it may be difficult, if not impossible, to tell where the paths before us might ultimately lead, we cannot go backward; we must continue on, doing our best to pick the path that will lead us to where we want to be. Given its intuitive appeal, it's no surprise that several philosophers have used this metaphor in an attempt to better understand the nature of agency, particularly that of the freedom to do otherwise. Authors such as Peter van Inwagen (1983Inwagen ( , 1990, John Martin Fischer (1994), Laura Ekstrom (2000), and Robert Kane (2005), among others, all use the Garden of Forking Paths to motivate, to varying degrees, (leeway) incompatibilism, i.e., the view that causal determinism is incompatible with the freedom to do otherwise. Roughly, the idea is this. The single path behind us is best understood as corresponding to the temporal past: as containing all those events that have occurred prior, and hence "led up," to our present decision. The paths before us, being connected to the path we are currently on, are best understood as corresponding to those possible temporal futures that are consistent with the past. But not merely that. Just as there are rules for how we can travel from one path to another-there's no jumping from one path to another, say-so there are rules for which possible temporal futures are allowed, namely, the laws of nature. If a temporal future is consistent with the past but not the laws, then it is not a future available to us. So, the paths before us correspond to those possible futures consistent with the conjunction of the past and the laws. Let's call this the "standard incompatibilist interpretation" of the Garden of Forking Paths.
It's easy to see how incompatibilism follows. If causal determinism is true, then, at the very least, the laws operating on the past determine a unique future. That would mean there's only one path available to us at any given moment, which corresponds to the claim that we are never free to do otherwise than what we actually do. Hence, causal determinism is incompatible with the freedom to do otherwise. Peter van Inwagen (1990) puts it succinctly: To say that one has free will is to say that when one decides among forks in the road of time. . . , one is at least sometimes able to take more than one of the forks. . .
. . . Determinism is the thesis that it is true at every moment that the way things then are determines a unique future, that only one of the alternative futures that may exist relative to a given moment is a physically possible continuation of a state of things at that moment. . . Thus, according to determinism, although it may often seem to us that we confront a sheaf of possible futures (like this) what we really confront is something like this Here the dotted lines represent futures that are not physically possible continuations of the present, and the single solid line represents the future that the laws of nature permit. (pp. 277-79) 1 So, if this is indeed the best interpretation of the Garden of Forking Paths, this intuitive metaphor supports a highly controversial view in philosophy, something noteworthy in and of itself.
Alas, this is not the best interpretation of the Garden of Forking Paths, or so I'll argue. In what follows, I'll present two general problems for this interpretation and then offer a more promising alternative that still leads to incompatibilism. I'll then consider (leeway) compatibilist-friendly revisions and note the relative strengths and weaknesses of these alternatives. 2 Finally, I'll turn to the larger question of whether the Garden of Forking Paths, however it is best interpreted, can provide any insight into the nature of freedom.

WEEDS IN THE (INCOMPATIBILIST'S) GARDEN
The first general problem with the standard incompatibilist interpretation is that it includes too much. Consider the single path behind us. Intuitively, this path represents whatever it is that led us to our current decision, whether that be previous decisions we have made, influences on those decisions, or just happenstance. But here's the point: the path that has led you to your current decision need not overlap exactly with the path that has led me to my current decision. As a dramatic example-there'll be lots of those in what's to come-suppose you are facing a decision here on earth while I am facing a similar decision but in a galaxy far away, so far away that the events of your life couldn't possibly have affected the events of my life, nor vice versa. Should we think that the path you have been on looks exactly like the path I've been on? Presumably not. But the standard incompatibilist interpretation implies just that since the temporal past for us is exactly the same. Indeed, since there is only one temporal past, relative to now, the standard incompatibilist interpretation implies that the single path leading up to the present moment is exactly the same for everyone. That seems like too much. 3 The same problem arises regarding the paths before us. Intuitively, those paths before us correspond only to those possible futures we have some say over. If one path before me is sunny and another is shady, say, then it is up to me, to at least some degree, whether I find myself on a sunny or shady path. The standard incompatibilist interpretation is at odds with this claim. Suppose there's some objective chance, given the past and laws, that the sun explode tomorrow (perhaps due to strange quantum phenomena). This interpretation would suggest that there are (at least) two paths before me: one where the sun explodes tomorrow and one where it doesn't. If I am supposed to have any say at all over which path I end up on, then this interpretation implies that I have some say over whether I find myself on a path with an exploding sun. That looks like a problematic result. 4 Or to return to our earlier example: the paths facing you here on earth need not-arguably should not-be the same as those facing me in a galaxy far away since I have no say whatsoever over what path you end up on (nor vice versa). But since the set of possible futures consistent with the temporal past and laws is exactly the same for all agents, everyone faces the same paths before them under the standard incompatibilist interpretation. Again, that seems like too much.
The root of the problem is that the standard incompatibilist interpretation invokes notions that are invariant across all agents. We all share the same temporal past; the same laws obtain for all of us; the set of futures consistent with the temporal past and laws is absolute. In contrast, the intuitive understanding of the Garden of Forking Paths is such that the paths vary greatly from agent to agent-that the garden looks quite different for each of us.
It's worth addressing an objection here (and perhaps offering a confession). The authors mentioned above claim, at most, that a path is available to you only if it is consistent with the temporal past and laws; no one claims its being so is a sufficient condition for it being available to you. Something similar could be said about the path behind us: that a path is behind you only if it is at least part of the temporal past. With this in mind, the problems just raised disappear. Sure, the objection goes, the path behind us shouldn't be identified with all of the temporal past, but it is clearly part of it; and sure, the paths before us shouldn't be identified with all possible futures consistent with the laws and past, but they are surely a subset of those futures. And insofar as that's all the incompatibilist needs to get the argument going, what's the problem?
I concede the objection: the standard incompatibilist interpretation offered above is a misrepresentation. But the misrepresentation was intentional, for it helps to highlight just how shallow the standard incompatibilist "interpretation" is. One would think that a minimal condition on any interpretation of the Garden of Forking Paths is that the interpretation respect how the paths in the garden, both those behind and in front, vary from agent to agent. Surely, the agent-relativity of the Garden of Forking Paths is a basic feature of the metaphor. If an interpretation of a metaphor doesn't even attempt to capture basic features of that metaphor, it's hard to see how the metaphor, however intuitive, provides any motivation for the claims made by that interpretation. Indeed, it's hard to see how it is an "interpretation" of that metaphor at all. So, in my view, once the standard incompatibilist "interpretation" is properly understood, it's a non-starter.
However, if this line of reasoning is found to be unpersuasive, I offer up another response: the standard incompatibilist interpretation faces another general problem, namely, it includes too little. That is, there seem to be counterexamples to even the minimal claims that a path is behind you only if it is part of the temporal past, and that a path is before you only if it is consistent with the past and laws. The easiest way to see this is to consider some even wilder examples. First, a version of the (in)famous Grandfather Paradox: Tim travels back in time and approaches a young Grandfather with the intent to kill him and thereby save all the lives Grandfather will one day ruin. But at the last moment, Tim has a change of heart and doesn't even attempt to kill Grandfather.
At the moment when Tim exits his time machine and approaches Grandfather, is he able to kill him? That is, in addition to the path he actually takes-the one where he has a change of heart and walks away-is there another path available, one where he kills Grandfather? Admittedly, it's not entirely obvious, but there's a compelling case for a negative answer. After all, if he had killed Grandfather, Tim would never had been born, in which case he wouldn't have been faced with the decision to begin with. Had he chosen the path where he kills Grandfather, Tim wouldn't have found himself on any path at all. That seems like a good reason to think Tim cannot kill Grandfather. 5 If so, then the standard incompatibilist interpretation includes too little. That's because, relative to the time when Tim first appears with his time machine, there would seem to be futures consistent with the past and the laws where he kills Grandfather. Imagine a path where he kills Grandfather, but then someone harvests Grandfather's genetic material and delivers it to Grandmother. Or imagine one where Tim forces Grandfather at gun point to first visit the fertility clinic in Grandmother's neighborhood and only then pulls the trigger. These would seem to be paths consistent with the conjunction of the past (relative to Tim's arrival) and the laws but where Tim nonetheless kills Grandfather. Of course, the future would have been different, had these paths obtained. Grandfather wouldn't have met Grandmother, or ruined so many lives, for example. But that's precisely the point: it is the future which seems to be constraining Tim's freedom, not the past.
In terms of the Garden, this means that the path which has "led up" to Tim's decision of whether to (try to) kill Grandfather ought to include not just parts of the temporal past, but parts of the temporal future. Grandfather's meeting Grandmother, or certainly Grandfather's ruining so many lives, seem to be events which have "brought" Tim to his current predicament, even if those events take place in the temporal future. More generally, the single path behind us is richer than the standard incompatibilist interpretation suggests.
Next, consider a supernatural case: suppose God has given you a great gift, namely, you are to decide whether gravity will continue to operate in the way that it has up until now. You are torn. On the one hand, you know how fine-tuned the laws supposedly need to be in order to support intelligent life. Altering those laws could be disastrous. On the other hand, it would be fun (and convenient) to leap over tall buildings in a single bound. You decide to err on the side of caution and let gravity remain as it has. 6 Intuitively, at the moment of your decision, you have some (indirect) control over the laws: it is up to you, to some degree, whether you find yourself in a world with the actual laws of gravity or not. In terms of the Garden, that means some of the paths before you include different laws of nature. Obviously, the standard incompatibilist interpretation doesn't allow for this. If a path has different laws, it is automatically unavailable to you. So, in this case, the set of paths before you includes paths the standard incompatibilist interpretation disallows.
I'm sure many will object to these cases. It's highly controversial whether time travel is metaphysically possible; cases involving God altering the laws are even more fantastical. And even if both cases are metaphysically possible, they clearly don't apply to us. We lack access both to time machines and a divinity that is willing to change the laws for us. So, at best, these cases are irrelevant counterexamples. At worst, they are not counterexamples at all.
I am willing to grant these points because I think they are somewhat orthogonal. The point of these cases is to show there's something deeper about freedom that the standard incompatibilist interpretation is missing. The fact that we ought to interpret the Garden of Forking Paths differently for time travelers or divinities should make us wonder: what is it about those agents that changes the layout of the Garden? If we can answer that question, we'll presumably gain some insight into what features are relevant to freedom. Or, to put the point another way: if we instead had an interpretation of the Garden of Forking Paths that applied not only to us, but also time travelers and divinities, that interpretation would presumably be tracking factors more central to freedom than those tracked by the standard incompatibilist interpretation. So, even though we are neither time travelers nor divinities, and even if such cases are impossible, they still seem to point us toward something deeper about freedom. Now we can see how the standard incompatibilist interpretation includes too little. If we were time travelers, the single path behind us ought to include parts of the temporal future as well, such as Grandfather's terrible deeds; if we were divinities or approached by the right ones, the many paths before us ought to include ones where the laws of nature are different. Of course, we find ourself in neither situation-such situations may even be absolutely impossible-but that's not the point. Rather, focusing on this problem promises to shed light on the nature of freedom. Let's see if we can direct some of that light to the garden.

PULLING THE WEEDS
Now that we've seen the shortcomings of the standard incompatibilist interpretation, I'll offer an alternative, one that both avoids the general problems just raised and still implies incompatibilism.
Let's start with the single path behind us. We noted that the standard incompatibilist interpretation includes both too much-it needn't include the entire temporal past-and too little-for time travelers, say, it ought to include parts of the future as well. It turns out there is a relatively simple fix for both problems, one that turns on what it means for the path to "lead up" to our current decision. Even though the temporal past is the same for all of us, only some of the past is causally responsible (directly or ancestrally) for getting us to our current decisions. If we are living galaxies apart, the events that brought us-in a causal sense-to our current decisions will vary considerably. Depending on how far back we are willing to follow the causal chain (an issue which we will touch on later) there may be some overlap, but the overlap needn't be perfect.
Similar comments apply to the case of time travelers. Even if the crimes of Grandfather occur in the future, relative to when Tim exits his time machine, they are clearly part of what brought Tim-in a causal sense-to his current decision. Not all of the future necessarily has a causal bearing on Tim's decision, but plainly some of it does. Those parts of the future that do ought to be included in the single path that brought Tim to his current decision.
So, the suggestion is this: we should not understand the single path behind us as being comprised of all of the temporal past (or even a part of it), but rather as being comprised of all of the causal past: as consisting of all of those events which are causes, direct or ancestral, of the agent's current decision (or "choice-situation," if one prefers). Since the events that brought each of us to our current decisions needn't be the same, our paths will look different. And if somehow parts of the future have brought us to our current decisions, as in the case of time travel, our paths will include those parts of the future as well. More generally, the path behind us is not about time; it's about causation. 7 Let's now turn to the paths before us. I've argued that the standard incompatibilist interpretation includes both too much-not every possible future consistent with the past and laws is an available path-and too little-for divinities, say, possible futures with different laws ought to be available, too. Again, focusing on causation provides a relatively simple fix. Whether the sun explodes tomorrow or not, say, has nothing to do with any of my choices. My choices are completely causally inefficacious with regard to that event. The opposite holds in the case where a divinity offers to change the laws tomorrow. In that case, my choices do have some (indirect) causal bearing on the event (or fact) in question. Absent an intervening divinity, my choices don't, of course, but in this fantastical case, they do.
The suggestion, then, is this: we should not understand the paths before us being comprised of every possible future consistent with the path behind us and the laws, but rather as being comprised of every possible future consistent with the path behind us and that our choices make a causal difference to. Somewhat more precisely: an event (or fact), e, is on one path before us but not another at the time of our decision just in case both e's occurrence and non-occurrence is consistent with the events that brought us to that decision and our decision makes some causal difference to e's occurrence (or obtaining). What does it mean for an event to make a causal difference to another event? This is a delicate issue, but something along the following lines seems promising: event e 1 makes a causal difference to event e 2 if (i) were e 1 to (not) occur, then e 2 would also (not) occur, and (ii) this counterfactual relation obtains at least partly because e 1 is (or would be) a cause, direct or ancestral, of e 2 . 8 If it is a fact, say, that the sun will explode tomorrow, then, since my choices make no causal difference to that fact, every path before me ought to include the sun's future explosion. Meanwhile, if a divinity offers me a chance to change the laws, then, since my choices do make a causal difference regarding which laws obtain, some of the paths before me ought to include different laws. 9 Those look like the right results, but only because, again, this interpretation focuses on the causal rather than the temporal future.
Let's call this interpretation the "causal incompatibilist interpretation." Here's the justification for the "incompatibilist" label: since we in fact lack any causal power over the laws, every path before us has the same laws as our own. But if causal determinism is true, then our causal pastsall those events that brought us, directly or ancestrally, to our current decision-together with the laws entail a unique future. That is, there is only one path before us consistent with the conjunction of our causal pasts and the laws. Hence, if causal determinism is true, none of us are ever free to do otherwise.
To sum up the argument so far: the standard incompatibilist interpretation both includes too much and too little. By shifting away from time and the laws toward causation, we arrive at an interpretation that not only includes the right amount, but also delivers incompatibilism. This strongly suggests that incompatibilists ought to abandon the standard incompatibilist interpretation in favor of the causal incompatibilist interpretation. 10 In what remains, I'd like to address two further questions. First, can the (leeway) compatibilist offer an alternative interpretation, one that has as much appeal as the causal incompatibilist interpretation? Second, should we care whether a view of freedom fits well with the Garden of Forking Paths?

COMPATIBILIST INTERPRETATIONS
Hopefully by now, I've convinced you that the Garden of Forking Paths metaphor cries out to be understood in terms of causation. When we say that the single path behind us has "led" us to our current decision, or that which path we choose now will "lead" us to further forks in the Garden, it is almost irresistible to understand this "leading" as causal. Therefore, I think it would be a mistake for the compatibilist to stray away from causation and focus exclusively on some other notion. This lesson can help highlight what many find relatively attractive or unattractive about various versions of compatibilism. We'll start with two versions of compatibilism that seem to run afoul of a causal interpretation of the Garden of Forking Paths. We'll then turn to two versions that are much more promising. LAW According to "local miracle compatibilism," 11 agents like us are sometimes free to perform actions, even should their doing so require the laws to have been different. 12 Suppose that I decided to have a donut rather than an omelet this morning and that my decision was causally determined by my causal past and the laws. According to local miracle compatibilism, I could have decided to have an omelet this morning even though, had I done so, the laws would have to have been different-that a few particles would have to have swerved in another direction, or a few additional neurons would have to have fired, such that, given the circumstances, the laws would have to have been different. Now consider what this would imply for the Garden of Forking Paths. If I could have chosen to have the omelet, then, at the time of my decision, there was a path before me with a different set of laws. As I suggested above, that strongly suggests that, at the time of my decision, it was up to me which set of laws obtain. More carefully, I take this to be a central feature of the Garden of Forking Paths: if one path before me has feature F and another doesn't, then it is up to me, at least to some degree, whether I find myself on a path with feature F or not. To use our earlier example, if one path is sunny and another shady, then it is up to me whether I find myself on a sunny path or a shady path. Combined with local miracle compatibilism, this would imply that it is up to me, at least to some degree, whether I find myself in a world with the actual laws or not. That looks problematic.
It's worth spending more time on the problem here. If I choose the sunny path over the shady one, my decision does not make the path sunny-it is not up to me whatsoever how much sunlight reaches that path. Rather, all that is up to me is whether I find myself on a sunny path. By analogy, the problem for local miracle compatibilism is not that, if I were to do otherwise, my decision would thereby make the laws different. I'm willing to concede that local miracle compatibilism implies no such thing. 13 The problem is instead that, just as the fact that I find myself on a sunny path rather than a shady one is partly explained by my choices, so the fact that I find myself in a world with the actual laws rather than slightly different ones is also partly explained by my choices. That implication, while perhaps not as troubling as the claim that I can make the laws different, still seems troubling enough. 14 Next, consider "multiple pasts compatibilism," 15 according to which agents are free to perform actions, even should their doing so require the past to be different. 16 For instance, it may be true that, had I eaten an omelet instead of the donut this morning, the past would have to have been different-say, I would have had to have not deprived myself of sweets last night. But that is no obstacle to my being free this morning to choose the omelet instead.
Turning again to the Garden of Forking Paths, the multiple pasts compatibilist will admit that the fact that I deprived myself of sweets last night causally contributed to my decision to eat the donut. Nonetheless, she should insist that that fact is not on the "single path" which led up to that decision. After all, a basic feature of the Garden of Forking Paths is that whichever path I choose before me, the path behind me stays the same. But, by hypothesis, had I chosen the omelet, I wouldn't have deprived myself of sweets last night. So, that fact cannot be part of the path. More generally, it seems as if the multiple pasts compatibilist should claim that the single path which leads up to the agent's decision is comprised of all those events that causally contribute to the agent's decision (or "choice-situation") but that are counterfactually independent on what the agent actually decides. The path that leads up to my decision to eat the donut rather than an omelet, for instance, may include some causal contributors, like the fact that I was in a rush this morning, say, so long as my actual decision makes no counterfactual difference to my being in a rush. But any causal contributors that my actual decision does make a counterfactual difference to, like my having deprived myself of sweets last night, are not included on the single path. Most simply, the "single path" is the agent's counterfactually independent causal past.
Once we place this emphasis on the difference between the counterfactually dependent and independent causal past, though, we run into problems. Consider Tim and Grandfather again. I claimed that some of the future, relative to when Tim is deciding whether to (try to) kill Grandfather is part of the single path that brought Tim to that decision-that the fact that Grandfather ruined so many lives, for instance, is part of the single path. That's because those events are causally responsible, directly or ancestrally, for Tim's being faced with that decision in the first place. However, many of the future events that brought Tim to that decision would seem to be counterfactually dependent on Tim's not killing Grandfather-Grandfather certainly wouldn't have ruined so many lives, had Tim killed him. So, if counterfactual dependence makes all the difference, it would appear as if the multiple pasts compatibilist cannot say that those events are part of the single path that led Tim to that decision. That seems like a strike against this interpretation. Now, the multiple pasts compatibilist could respond by claiming that, while all counterfactually independent parts of one's causal past are part of the single path, some but not all counterfactually dependent parts of one's causal past are. In general, those parts of an agent's causal past which are counterfactually dependent on the agent's decision-such as my having deprived myself of sweets last night-are not part of the "single path" which leads up to the agent's decision, but there are exceptions. Perhaps Grandfather's ruining so many lives is one of those exceptions.
Of course, we'll want to hear more about how to distinguish between those counterfactually dependent parts that are part of the single path and those that aren't. But let me offer a more general worry for the multiple pasts compatibilist interpretation: that while perhaps not incoherent, it nonetheless seems ad hoc. The multiple pasts compatibilist draws a line between causal contributors that are counterfactually dependent on our choices and those that aren't, and understands the single path in light of this distinction. But why? What is the independent motivation for placing so much emphasis on this distinction? The situation only becomes worse if the multiple pasts compatibilist admits that, in cases like time travel, parts of the counterfactually dependent causal past are also included in the single path. Without a deeper story, this interpretation seems undermotivated.
It seems to me that the best the multiple pasts compatibilist can say is something like this: "We take for granted that the freedom to do otherwise is compatible with determinism. But that means, given our view of counterfactuals, parts of our causal pasts would have been different had we done otherwise. So, we ought to understand the Garden of Forking Paths in light of this (and perhaps with some tinkering to accommodate time travel)." The problem with this line of reasoning is obvious in this context: we are trying to use the Garden of Forking Paths to arrive at a view of freedom, not the other way around. This line of reasoning gets things exactly backwards.
So, in my view, two of the most popular versions of compatibilism, local miracle and multiple pasts compatibilism, do not provide promising interpretations of the Garden of Forking Paths. However, there are less popular versions of compatibilism that do, or so I will argue now.
First, consider so-called "contextualist" versions of compatibilism, as explored by John Hawthorne (2001) and defended by Ann Whittle (2021). Just as ascriptions of knowledge may be true relative to one context but false relative to another, so ascriptions of freedom may be true relative to one context but false relative to another, contextualist compatibilists claim. For instance, Hawthorne offers the following: S does x freely only if S's action is free from causal explainers beyond S's control-Psst!-apart from those causal explainers that we are properly ignoring. (2001, p. 68) Whether a causal explainer is one that we are "properly ignoring" is context-dependent. In one context, we may say that Jones freely decided to drink another beer, properly ignoring minute details about Jones's neural makeup at that moment, complexities of the environment he is in, his family's history of alcoholism, etc.; in another context, one where those causal explainers cannot be properly ignored, we may say that his decision was not free. Since there are contexts where we may properly ignore parts of an agent's causal past, there would seem to be contexts where ascriptions of freedom to that agent are true, even if determinism holds.
In terms of the Garden of Forking Paths, the idea would be this: the single path that has led up to our current decision is comprised of those events that are causally responsible for our current decision and contextually salient. When I am deciding what to eat for breakfast, there may be an impossibly long list of events that brought me to that decision (or "choice-situation"), but only some of them are salient. The fact that I denied myself sweets last night, or that I'm in a rush, say, may be salient with regard to my decision in certain contexts; the fact that a few particles swerved in a certain direction, or that some additional neurons did not fire, is not salient in a typical context, even if such facts somehow causally contributed to my decision. The paths before us, then, are comprised of those futures consistent with the causally salient past and that our choices make a causal difference to.
Next, consider "higher-level" views of freedom, like that defended by Christian List (2019). According to such views, free will is a "higher-level" phenomenon, one that is not reducible to "lower-level" phenomena, such as particle physics. Just as it might seem mistaken to look for causes of recent economic inflation by investigating the quantum level, say, so it would be mistaken to look for free will there too. Rather, according to List, causal determinism in physics, being a hypothesis about lower-level phenomena, is no direct threat to free will. It is only determinism at the right level-the psychological or agential level-that would pose a threat to free will. And crucially, it is possible for determinism to obtain at the level of particle physics without it obtaining at the level of psychology or agency. Hence, determinism at lower-levels is compatible with the freedom to do otherwise.
Turning to the Garden of Forking Paths, this view would suggest that the single path behind us consists of all those events at the psychological or agential level that brought us to the decision in question. When I am deciding what to eat for breakfast, my hunger is certainly part of the single path, but the microscopic events that "underwrite" my hunger (or the atomic-level description of my hunger) are not part of the path. Even events that are not contextually salient-say, events that occurred thousands of years ago-may be part of the single path that has led up to my current decision, so long as the events are at the right level, roughly, the level of psychology or agency. The paths before us, then, are comprised of those possible futures consistent with the level-appropriate causal past and that our choices make a difference to.
Of course, some compatibilists will be unsatisfied with both the contextualist and higher-level views since they admit that, if we are in the right context, or if determinism holds at the right level, then we have no free will (respectively). That may be seen as conceding too much to the incompatibilist. I don't have anything new to say to these objections. I merely wish to point out that the intuitive features of the Garden of Forking Paths seem respected by these versions of compatibilism.
In summary: while some compatibilist views, such as local miracle compatibilism and multiple pasts compatibilism, don't seem to fit very well with the Garden of Forking Paths metaphor, other compatibilist views, such as contextualist compatibilism and higher-level compatibilism, do. It is an interesting question whether the causal incompatibilist interpretation has any significant advantage over a contextualist or higher-level compatibilist interpretation (or vice versa), but I will not address it here other than to say that I think the answer is probably "no." (If anything, I'm inclined to think that the contextualist interpretation is the most promising of the bunch.) Instead, I will conclude by trying to answer another question: should any view of freedom be concerned with how well it fits with the picture of the Garden of Forking Paths? There's certainly something right about these comments. The point is perhaps clearest in cases where there is reason to think that our intuitive pictures of the world don't (or needn't) provide reliable representations of the world. Think about different intuitive pictures of time. Standard relativistic physics tells us that time does not "tick away" at an invariant rate, but rather "stretches" and "compresses" at extremely high speeds or near massive objects. Hence, many of our intuitive pictures of time are inaccurate here. But this does not seem to be a strike against standard relativistic physics whatsoever since we have good reason to think that our intuitions are unreliable in this domain: since we only deal with medium-sized, relatively slow objects, there is no evolutionary advantage to being "in tune" with relativistic phenomena. There is no reason to pay any mind to our intuitions here. However, compare that with the picture of time "flowing" or having a "preferred" direction. If there is a direction to time, one that reduced to or tracked features like causation and change, it would presumably be a large evolutionary advantage to have a quick and intuitive grip on this, even if our grip isn't perfectly tight. After all, our survival would seem to depend on being able to track such features. Of course, that's not to suggest our intuitions here are infallible, and plenty of authors have argued that there is no preferred direction to time. But that's the point: if one is going to argue that our intuitive picture of time as an "ever-flowing river" 17 is mistaken, one had better have some powerful arguments.

CONCLUSION: METAPHORS AND ARGUMENTS
So, the suggestion is this. In domains where our intuitions are unreliable, or we have reason to suspect so, the fact that a certain view doesn't fit with an intuitive picture or metaphor is no strike against the view-pictures aren't arguments! But if our intuitions are reliable, or we have reason to suspect so, then pictures can serve as arguments. Or, more carefully: the fact that a view doesn't respect central features of an intuitive metaphor or picture is a consideration against the view.
With regard to the Garden of Forking Paths, the question then is whether our intuitions about agency and choice are reliable. For obvious reasons, I can't hope to even sketch an answer one way or the other, but allow me one point in closing. Most (if not all) local miracle and multiple pasts compatibilists believe that we in fact are free to do otherwise on some occasions. That would presumably mean that at least some of our intuitions about agency and choice are reliable to some degree. If so, these compatibilists owe us a story as to why the intuitions undergirding the Garden of Forking Paths are not reliable, at least if the foregoing arguments are correct. That's not an impossible task-in my view, there are some promising things to say here. 18 The point is merely that the typical local miracle and multiple pasts compatibilist faces a dilemma: either they must show that, contrary to the arguments above, their view does fit well with the Garden of Forking Paths metaphor, or they must show that the intuitions behind the metaphor are unreliable despite other intuitions about free will being reliable. They must either prune the garden or burn it down. 19

A C K N O W L E D G M E N T S
Open access funding enabled and organized by Projekt DEAL.

E N D N O T E S
1 The figures, while approximating van Inwagen's, are my own. 2 I set aside "source" in/compatibilist views-views according to which the freedom relevant to moral responsibility doesn't require the ability to do otherwise-other than to say this. The standard motivation for such views comes from so-called "Frankfurt-style cases" according to which an agent is morally responsible for some action despite the fact that the agent couldn't have done otherwise. But it is an infamously difficult issue whether the agents in such cases really couldn't have done otherwise and, thus, it would benefit the typical source in/compatibilist to have an interpretation of the Garden that vindicated that claim. 3 A related worry has to do with the possibility of time travel. If we were able to travel to the past, it would seem that certain options would be open to us that were, nonetheless, inconsistent with the actual past. The standard incompatibilist interpretation obviously denies this. For more on this point, see Swenson (2016), Law (2020), Wasserman (2022), and Law & Wasserman (2022). 4 Thank you to Carolina Sartorio for raising this case. 5 The literature on the Grandfather Paradox is decently large. The classic piece is Lewis (1976). For a nice contemporary overview, see Wasserman (2018, ch. 4). For more on this rationale in particular, see Law & Wasserman (2022). 6 This case is inspired by Cutter (2017). 7 For similar suggestions, albeit not explicitly in the context of the Garden of Forking Paths metaphor, see Rea (2015) and Wasserman (2022). Reflecting on cases of time travel, even Fischer (2021) has claimed that "fixity follows causation," which I take to be amenable to the suggestion here. More generally, I take the idea here to be consistent with the recent trend of thinking about the relation between causation and freedom rather than mere determinism and freedom. See Sartorio (2015) for an especially lucid expression of this idea, one that has greatly influenced my own thinking. 8 For an alternative take on the "difference making" relation, see Sartorio (2013Sartorio ( , 2016. 9 Some might wonder about overdetermination and preemption cases. Suppose Suzy throws a rock, shattering the window, but Billy either threw a rock too or would have had Suzy not thrown it. Intuitively, it seems as if there was no path available to Suzy where the window doesn't shatter-she only had a say over whether she was a cause of the window's shattering. The definition of "causal difference" invoked here aims to capture that. Thank you to Carolina Sartorio for raising this point. 10 There is an interesting asymmetry in the "causal incompatibilist interpretation," as I have presented it here, namely, that whereas the path behind us is comprised of all of those events which contributed to our present decision-even those causes which are not counterfactually connected to our present decision-, the paths before us are comprised of all of those events that our present decision causally contributes to and that are