In Building Better Beings: A Theory of Moral Responsibility (2013), Manuel Vargas develops and defends what he characterizes as a revisionist view of moral responsibility. It is revisionist in the sense that the prescriptive account it offers concerning how we ought to conceive of moral responsibility is one that openly admits to being at odds in significant ways with how people ordinarily think of moral responsibility. The book is divided into two main parts. The first part of the book is aimed at motivating revisionism in general. The second part of the book develops and defends Vargas’s particular revisionist account of moral responsibility, the agency cultivation model, according to which praising and blaming practices are justified in virtue of their effectiveness in developing a certain kind of agency, one that is concerned with and sensitive to moral considerations. Here, I will focus primarily on the first part of the book.

1 The case for revisionism

Vargas argues that experimental data, together with certain other considerations, provides strong support for the view that a significant portion of ordinary thinking about free will and moral responsibility involves commitments to incompatibilism (p. 48).Footnote 1 This, in itself, does not show that we should embrace an incompatibilist account of moral responsibility as being correct, of course. But given what Vargas calls the principle of philosophical conservation, according to which “we ought to abandon our standing commitments only as a last result,” the fact that ordinary thinking about free will and moral responsibility involves significant incompatibilist commitments implies that our first project should be to try to develop a satisfactory incompatibilist account (p. 73). Only once this project proves unsuccessful should we proceed to consider compatibilist alternatives. As it turns out, Vargas argues, “there are good reasons to doubt the prospects of libertarianism as a theory about what we ought to think about moral responsibility, all things considered” (p. 53).

Vargas maintains that there are two standards that any promising theory of moral responsibility will satisfy: the standard of naturalistic compatibility and the standard of naturalistic plausibility. As Vargas uses the term, “naturalistic compatibility refers to coherence with the known facts, along with compatibility with two further things: (1) the principle of the causal closure of the physical; and (2) acceptance of a principle of methodological minimalism,” according to which we are willing to add entities and powers to our ontology only when doing so is “warranted in light of their explanatory power, coherence with other commitments, predictive utility, and so on” (pp. 56–57). In recent years, it has not been terribly uncommon to see those in the sciences who address free will and moral responsibility argue that libertarian free will is incompatible with a naturalistic view of the world on the grounds that it would require such things as the existence of immaterial souls, that certain mental actions (such as decisions) have no (causal) connection whatsoever to the neural activity in our brains, regular violations of the laws of nature, or other “magical” phenomena. As Vargas is quick to point out, however, libertarianism admits of different types. All libertarians hold both that we sometimes act freely and that acting freely requires that one’s action is not causally determined, but there is variation regarding whether and how actions must be caused in order to be free. Whereas noncausal libertarians maintain that free actions involve mental events that are not (or at least need not be) caused at all, event-causal libertarians conceive of free actions as being caused by prior events but only indeterministically so. According to agent-causal libertarians, while free action involves indeterministic causation, it is causation of an event by an agent (a substance) and is not reducible to causation between events. Vargas readily grants that versions of each type of libertarianism can, and most contemporary versions do, satisfy the standard of naturalistic compatibility as he characterizes it—including agent-causal views that require “irreducible agential powers [that] emerge at sufficiently complex levels of organization” (p. 57).

Here, I will just pause to note that Vargas’s treatment of this point about the various versions of libertarianism and their ability to satisfy the standard of naturalistic compatibility is indicative not only of a deep familiarity with the vast literature concerning free will and moral responsibility but of an intellectual honesty and philosophical charity that is evidenced throughout the book. One nice consequence of this is that reading through the book feels much more like following the author on a sincere intellectual exploration of an issue about which he is genuinely concerned than it does being presented with a philosophical sales pitch for the theory he ultimately wants to advance. It’s refreshing, and it makes for a very enjoyable read. Okay, now back to why he is an evil revisionist and must be stopped!

Vargas argues that the trouble for libertarianism comes when we apply the standard of naturalistic plausibility. The general idea behind this standard, he says, is that “it is one thing for an account to be consistent with the facts and it is another for it to be a plausible theory, given what we know” (p. 59). Satisfying the standard of naturalistic plausibility requires not only satisfying the standard of naturalistic compatibility but that “there are truth-relevant considerations that speak in its favor when it postulates requirements that exceed the known facts or the widely accepted ontologies of our current scientific understanding” (p. 59, italics omitted). Vargas distinguishes two versions of the standard of naturalistic plausibility: a threshold version and a scalar version. Libertarianism fails to satisfy either version of this standard, he argues, and so should be rejected.

According to the threshold version of this standard, an account will be naturalistically plausible only if there is some “positive truth-relevant evidence for the account” (p. 60). Libertarian accounts are committed not just to the world’s being indeterministic but that indeterminism shows up at particular times and that our brains operate in particular kinds of ways. Such an account’s merely being consistent with what we know about the operations of the brain, Vargas maintains, does not give us any reason to adopt such an account. To illustrate the point, he asks us to consider belief in extra-sensory perception (ESP). Even if ESP is not strictly ruled out by current brain science, this does not mean that we should consider it plausible. For that, “we would need some independent piece of evidence for its plausibility” (p. 63). “Similarly,” he says, “what the libertarian needs is an account of why we should be optimistic that libertarianism will be vindicated by science,” such as “solid empirical data that showed that indeterminism is present in exactly all the right spots and few, if any, of the wrong ones” (pp. 62–63). As it stands, however, he thinks “it is striking that nothing in the brain sciences would lead an independent observer to conclude that we must be libertarian agents” (p. 62). Thus, the threshold version of the standard of naturalistic plausibility requires that we “[reject] libertarianism on the grounds of there being no truth-relevant evidence in favor of the conclusion that our agency is of that sort” (p. 64).

How a theory fares on the scalar version of the standard of naturalistic plausibility is a function of how demanding it is in comparison to alternative accounts. Vargas writes:

We do not know where science will lead us and it takes a puzzling sort of confidence to simply assume that future discoveries will vindicate the more demanding theory. They might—or they might not. For us, the issue is what we have reason to think will be more likely. The growth of human knowledge has frequently been unkind to the products of the philosophical imagination. All things being equal, it seems a bad idea to bet on the truth of the more demanding theory. (pp. 66–67)

Since libertarianism requires that the world is indeterministic at particular times and places leading up to free actions, it is more demanding than compatibilist alternatives, thus making it “comparatively implausible” (p. 66). Vargas concludes that “libertarianism requires us to accept a comparatively implausible view without evidence” and so fails to satisfy either version of the standard of naturalistic plausibility (p. 72). As a result, we have good reason to reject libertarianism and look elsewhere for a view of moral responsibility.

In what follows, I offer a couple reasons for thinking that perhaps Vargas has been a bit too quick in rejecting what he himself takes to be the common sense view of moral responsibility—especially given his view that “we ought to abandon our standing commitments only as a last resort” and the crucial role that this principle plays in his own argument for why, once we exclude libertarianism, we ought to opt for revisionism over eliminativism (p. 73).

2 Meeting the standard of naturalistic plausibility

First, consider the claim that libertarianism fails to satisfy the standard of naturalistic plausibility. As we have seen, on the threshold version of this standard, an account will be plausible only if there are truth-relevant considerations that speak in its favor (p. 60).Footnote 2 On the face of it, however, it looks like there are truth-relevant considerations that speak in favor of libertarianism—namely, considerations concerning desert. On the basis of our experience in the world, we naturally come to form judgments not only that people sometimes do things that are morally good or morally bad but that they deserve certain kinds of treatment for the things they have done and so are morally responsible for their actions in a way that grounds or allows for that desert. The notion of desert at work here is what Derk Pereboom refers to as basic desert and what Galen Strawson (2002) describes as “desert of such a kind that it can exist only if punishment and reward can be fair or just without having any pragmatic justification, or indeed any justification that appeals to the notion of distributive justice” (p. 452). When combined with various incompatibilist arguments that desert-entailing moral responsibility requires the falsity of causal determinism, the apparent fact that people sometimes deserve praise and blame, punishment and reward, in this basic, non-consequentialist way would seem to provide us with evidence in favor of the truth of libertarianism.

Vargas never explicitly addresses whether considerations of this sort are able to provide libertarianism with the support needed to satisfy the threshold version of the standard of naturalistic plausibility. Later in the book, however, he expresses skepticism about the suggestion that the ordinary folk concept at work in desert judgments is as substantive as the notion of basic desert characterized above, substantive enough to preclude its being generated by consequentialist considerations. In Vargas’s view, when we judge that a person deserves blame for something she has done “what we are mainly committing ourselves to is the idea that [she] has done something wrong, and that in light of that violation, blaming is called for. In making this judgment, we need not have any view one way or another about the particular details of why that blaming is called for” (p. 253). “There is little evidence,” he says, “that ordinary desert ascriptions are especially deep in any uniform way to support the ‘basic desert as folk desert’ reading” (p. 256).

I certainly agree with Vargas that it would be nice to have better experimental data to lean on here. Not having any to offer at the moment, I will simply register my agreement with Pereboom and Strawson and note that, if it isn’t the case that ordinary judgments about desert are about the basic, non-consequentialist kind of desert Pereboom and Strawson describe, it seems utterly mysterious why anyone—let alone a significant number of philosophers—would think that desert and moral responsibility are incompatible with determinism. Even if the world turned out to be causally deterministic, agents could certainly still be such that their actions say something about the quality of their wills, that they could be subjected to aretaic judgements, that they are the kind of being of whom it make sense (at least in principle) to demand that they either justify or modify their attitudes, and that their behavior, character, and responsiveness to reasons are subject to being influenced by praising and blaming behavior of various sorts. And perhaps, following Vargas, there is even a consequentialist sense in which such agents can be said to deserve certain kinds of response (though, personally, I balk at using the term in this way). My point here is that, rather than offering support for compatibilism, the fact that all these things are not only compatible with determinism but are so obviously compatible with determinism seems like a strong indication that this isn’t (sufficient for) what the folk and all those incompatibilist philosophers are talking about when they talk about moral responsibility.

With respect to the scalar version of the standard of naturalistic plausibility, I am happy to grant that, to the extent that libertarianism requires that indeterminism shows up in the world at particular times and places leading up to free action, it is more demanding than accounts without such requirements. And perhaps, following Vargas, there is a sense in which this makes it reasonable to say that, all else being equal—that is, setting aside all considerations other than demandingness—libertarianism is comparatively less plausible than competing views. Even so, it is not clear that this gives us much, if any, reason to opt for an alternative account. After all, although they do not all require that the mental processes leading up to free action are indeterministic, most accounts (compatibilist and incompatibilist alike) require that free actions are caused in certain ways rather than others: that they are causally related to an agent’s beliefs, desires, or recognition of reasons; that they are the product of a mechanism that is suitably reasons-responsive; that they are caused by an agent (qua substance), even if deterministically so; or that they are caused in some other particular way. Any account of this sort will be more demanding and so, all else being equal, comparatively less plausible than a kind of non-causal account that has no requirements on whether free actions are caused at all, let alone how they are caused. Very few, it seems, have thought that this provides us with good reason to opt for a non-causal account of free action. Presumably, this is because all else is not equal and other considerations, such as the importance of control and reasons explanations, provide good grounds for favoring causal accounts over alternatives. Assuming we sometimes deserve punishment and reward in the way characterized above, if some of the arguments for incompatibilism are successful, the same will be true of libertarianism.

3 Questioning the standards of naturalistic compatibility and plausibility

Thus far I have offered some reasons for thinking that libertarianism does not fare nearly as badly on the standard of naturalistic plausibility as Vargas makes out (and perhaps not badly not all). In what remains, I want to suggest that, regardless of how libertarianism fares on them, neither the standard of naturalistic plausibility nor the standard of naturalistic compatibility are obviously relevant to determining what account of moral responsibility we ought to accept. Here, it is important to distinguish three kinds of views: (1) views about what moral responsibility is (i.e., views about what it is or means to be morally responsible); (2) views about what moral responsibility requires [i.e., views about what is required in order to be that which is characterized in (1)]; and (3) views about whether we are morally responsible (i.e., views about whether the conditions identified in (2) are satisfied in the world by us).

Now, whereas there does not appear to be any difficulty in separating (1) and (2) from (3), since it is common to give an account of what something is by offering a conceptual analysis of it in terms of necessary and sufficient conditions, marking the difference between (1) and (2) can be tricky. Still, it seems that there is a real distinction to be made here. Perhaps one way to see this is to consider the nature of the possible disagreements that exist between people like John Martin Fischer (a compatibilist, at least concerning moral responsibility), Derk Pereboom (a hard incompatibilist), and me (a libertarian). It may well be (and I suspect it is) that all three of us basically agree about (1) what moral responsibility is (or what it is to be morally responsible)—namely, that, among other things, it is capable of grounding ascriptions of basic desert. Pereboom and I might also agree (and I suspect we do) about (2) what is required in order to be morally responsible in this desert-entailing way, though we disagree about (3) whether we are morally responsible in this desert-entailing kind of way. And while Fischer and I might agree about (3) whether we are morally responsible in a desert-entailing way, we disagree about (2) what this requires.

Importantly, the only one of these disagreements that marks a difference in theory of moral responsibility is the disagreement that Pereboom and I have with Fischer. Only (1) and (2) are part of an account of moral responsibility. It is difficult to see, however, how the standards of naturalistic compatibility and naturalistic plausibility are relevant to evaluating (1) and (2). Whatever positions one takes with respect to (1) and (2) will be compatible with the truth (or falsity) of naturalism, as well as with whatever the known facts of science turn out to be. And since (1) and (2) do not posit the existence of any entities or powers, the standard of naturalistic plausibility will always be satisfied as well. Even a theory of moral responsibility that holds that being morally responsible requires the existence of Cartesian souls and violations of the causal closure of the physical will satisfy the standards of naturalistic compatibility and plausibility. It is only when this theory is joined with the view that we are in fact morally responsible that it is even possible for these standards to fail to be satisfied. As just noted, however, (3) views about whether we are in fact morally responsible are not part of an account of moral responsibility. While the standards of naturalistic compatibility and plausibility might be relevant to evaluating (3) views about whether we in fact satisfy the conditions of a given account of moral responsibility—and so might help us decide between positions like libertarianism and hard incompatibilism—it’s not clear that they are able to show that we should adopt one account of moral responsibility over another—at least not directly.

One way that these standards could play this kind of role is if we held that, all else being equal, an account of moral responsibility on which (3) the view that we are morally responsible satisfies the standards of naturalistic compatibility and plausibility is more plausible than an account of moral responsibility on which (3) the view that we are morally responsible fails to satisfy these standards. It may well be that this is precisely what Vargas has in mind. Note, however, that this kind of move makes sense only on the assumption that our confidence in (3) our view that we are morally responsible exceeds our confidence in (1) our view about what moral responsibility is (or what it is or means to be morally responsible). This seems very strange.

Certainly, there are some cases in which something like this is natural. For instance, there doesn’t appear to be anything odd about one’s being surer that water exists than one is about what water is. There are other cases, however, in which this seems very odd indeed. Imagine someone telling you, “I’m not sure what miracles are, but I’m confident they occur!” The beliefs at issue in (3) are concerned not merely with whether certain observable, rigidly designatable things exist or occur but with whether they admit of a certain normative classification. Thus, being more confident in one’s view that people are morally responsible than one is about what it is or means to be morally responsible is much more like the miracles case than the water case—unless Vargas is right about the minimal conceptual content of folk judgments about desert and moral responsibility. If, as Vargas maintains, ordinary judgments that an agent deserves to blamed (or possibly punished) for something she has done involve judging merely that “the agent has done something wrong and that something about that wrongdoing licenses blame” without “any view one way or another about the particular details about why that blaming is called for”—or even, presumably, about whether the blame is licensed on consequentialist or non-consequentialist grounds—then it is a bit more plausible that one could be more confident in (3) one’s view that people are morally responsible than in (1) one’s view about what moral responsibility is (p. 253).

In the end, then, the success of Vargas’s revisionist approach seems to significantly depend on the role that basic desert plays in ordinary judgments of blame and the folk conception of responsibility. Assuming some incompatibilist arguments are successful in showing that basic desert is incompatible with determinism, the more that ordinary judgments involve ascriptions of basic desert, the more reason we have for thinking libertarianism satisfies the standard of naturalistic plausibility and is true. Furthermore, the more central basic desert is to the ordinary conception of moral responsibility, the more reason we have to opt for eliminativism over revisionism in the event that libertarianism turns out to be implausible.