“Philosophical doubts about moral responsibility,” writes William FitzPatrick, “have typically been rooted in worries about free agency in the face of causal determinism, culminating in familiar metaphysical arguments against the very possibility of moral responsibility” (589).  So it is noteworthy that a novel argument for moral responsibility skepticism (hereafter ‘MRS’), recently advanced by Michael Zimmerman and Gideon Rosen, does not rest on controvertible metaphysical claims such as the defense of an incompatibilist determinism.  FitzPatrick’s article presents a critical assessment of this new argument.  While he applauds it for drawing our attention to difficult problems surrounding the theory and practice of our attributions of moral responsibility, FitzPatrick concludes that the argument’s skeptical conclusion is too strong.

This new argument for MRS (FitzPatrick focuses his discussion on Rosen’s 2004 presentation, which I have summarized here) is distinctive insofar as it is an epistemological argument, focusing on the warrant for our attributions of moral responsibility, rather than a metaphysical argument concerning the very possibility of moral responsibility itself.  Rosen’s argument is two-staged.  First, he argues that for any agent’s action to qualify as blameworthy, an instance of genuine akrasia must appear at some point in the “causal etiology” of that action.  Second, Rosen argues that we never have good grounds for judging an action to be akratic.

FitzPatrick (correctly, in my assessment) quickly dispatches with the second, “epistemological” stage of Rosen’s argument.  This stage amounts to little more than the assertion that reliable attributions of akrasia are difficult to make.  According to Rosen, we have scant grounds for such attributions, even in the first-personal case.  This claim rests on an appeal to the general “opacity of the mind”, together with the observation that it is often difficult to distinguish cases of genuine akrasia from those of “an imposter – ordinary weakness of will” (593-4).  But as FitzPatrick capably demonstrates, our grounds for akrasia-attributions are not nearly as scant as Rosen seems to suppose.  To begin with, consider cases of first-person attributions.  FitzPatrick points out that the experience of guilt and shame, sometimes felt even at the time of performing an action, can serve as a form of evidence (admittedly defeasible) that our present action is akratic.  Likewise, we need not think that second- and third-personal attributions of akrasia face insurmountable epistemic problems.  For consider the evidence provided by testimony: “people are sometimes honest with themselves about having acted akratically… and are honest with others about it, too” (596).  So the fact that people will sometimes flat-out tell you they have acted akratically is sufficient to undermine Rosen’s claims regarding the alleged inscrutability of akrasia.  Furthermore, FitzPatrick argues, certain forms of circumstantial evidence might warrant these attributions.  Inference to the best explanation, for instance, might demand that we regard a person’s action as akratic, even if that person testifies to the contrary: “Sometimes the alternative of ignorance is just so implausible that the principle of charity requires the attribution of akrasia” (598).  Likewise, a person’s efforts to conceal her morally incorrect behavior – if later uncovered – often provide “good evidence that she knew what she was doing was wrong,” and thus that she was acting akratically (598).  FitzPatrick concludes that there are “plenty of real cases – involving both our own actions and those of others – in which we can know, by any reasonable epistemic standards, that an action involves … akrasia” (599).

FitzPatrick takes little solace in his quick dispatch of the second stage of Rosen’s argument, however, as he reckons that the more forceful challenge lies in its first stage, wherein Rosen defends a tight link between moral culpability and akrasia.  Thus, FitzPatrick devotes the bulk of his article to examining and refuting Rosen’s contentions there, to the effect that assignments of moral responsibility for the performance of wrong actions must always terminate with the identification of an akratic act.

Let me first set out Rosen’s argument at this stage, borrowing liberally from my earlier review of Rosen’s 2004 article, “Skepticism About Moral Responsibility.”  Imagine Sally performs wrong action A.  There are two relevant possibilities here.  The first is that Sally knows that A is a wrong action.  If so, we have a clear case of akrasia, and clear grounds for an attribution of moral responsibility.  The second possibility is that Sally does not know that A is a wrong action.  In this case, we say that Sally acts “from ignorance” – either “circumstantial ignorance” (wherein Sally’s failure to know that A is wrong results from her ignorance of some matter of fact) or “normative ignorance” (wherein her failure to know that A is wrong results from her ignorance of some normative or moral truth).  Does Sally’s ignorance – whether circumstantial or moral – excuse her from responsibility for her wrong action?  Only if she bears no responsibility for her ignorance itself.  If Sally is not to blame for her ignorance, we regard this ignorance as a mitigating factor, and exempt Sally from judgments of responsibility for performing wrong action A.  If Sally is responsible for her ignorance, however, we say that she is “culpably ignorant”, and we do not regard her ignorance of A’s wrongness as an excuse.

When, then, is Sally morally responsible for her ignorance that A is wrong?  When she is morally responsible for her failure to discharge one of her “procedural epistemic obligations” – the procedures “any reasonably prudent person in [her] circumstances would have done in order to see to it that [she] was adequately informed” (Rosen 2004, p. 301).  And when is Sally responsible for her failure to discharge one of these procedural epistemic obligations?  Here is the crux of Rosen’s argument: Sally is thus responsible only when her failure to discharge these epistemic obligations is itself a clear instance of akrasia.  Were Sally (non-culpably) unaware that she stood under any such procedural epistemic obligation, her failure to discharge this epistemic duty – and thereby her resulting ignorance regarding A’s wrongness – would be non-culpable.

We now see why Rosen believes that for any putative case of moral responsibility for a wrong action, an instance of genuine akrasia must appear at some point in the “causal etiology” of that action.  We can also see why, even if we reject the second stage of Rosen’s argument for full-blown MRS, the sub-conclusion of this first stage poses an important challenge: for as FitzPatrick puts it, the “striking result” is that Rosen’s akrasia-invoking condition often “will not be met in many cases where we normally attribute responsibility, particularly in cases involving basic normative ignorance” (599).

Thus FitzPatrick defends an alternative condition for culpable ignorance, “CI”, which is weak enough to include (rather than to exclude, as Rosen’s does) many of the central cases in which we typically wish to attribute responsibility for normative ignorance.  FitzPatrick formulates CI thusly: “Ignorance, whether circumstantial or normative, is culpable if the agent could reasonably have been expected to take measures that would have corrected or avoided it, given his or her capabilities and the opportunities provided by the social context, but failed to do so either due to akrasia or due to the culpable, nonakratic exercise of such vices as overconfidence, arrogance, dismissiveness, laziness, dogmatism, incuriosity, self-indulgence, contempt, and so on” (609).  Among the central cases which CI (but not Rosen’s condition) countenances as examples of morally-culpable ignorance (cases to which FitzPatrick devotes considerable discussion, so as to motivate his view) are both fictional and non-fictional ones.    Among the former, there is the case of the character Mr. Potter from It’s a Wonderful Life.  Among the latter, there is President George W. Bush.  (FitzPatrick confesses that his motivation for formulating and defending CI may have been his recognition that Rosen’s criterion could serve – wrongly, in FitzPatrick’s estimation – to exculpate President Bush from responsibility for many of his actions while in office.)

Gideon Rosen’s contention that moral responsibility for wrong actions only arises as a direct or indirect result of clear-eyed akrasia is clearly articulated, forcefully argued, and – one feels – incorrect.  William FitzPatrick’s rival criterion for responsibility-generating culpable normative ignorance, CI, is unwieldy and imprecise in formulation; it is vague and relies heavily on anecdote in its defense; and yet it is – one feels – largely correct.  Thus the dialectic here is as it ever was between philosophical skeptics and their critics.  Neither Rosen, with his elegant and compelling argument for a radical and counter-intuitive conclusion, nor FitzPatrick, with his less-dazzling defense of a more-plausible position, clearly wins the day in this exchange.   Fortunately, though, inasmuch as it comes away from this contest with a vastly-improved understanding of the issues and options at hand, their audience does emerge as the clear winner.

Typically, skepticism about attributions of moral responsibility bears two features: (i) it relies upon metaphysical claims regarding the impossibility of free agency in a deterministic universe; and (ii) it encompasses all cases of moral responsibility, good and bad – judgments of blameworthiness and praiseworthiness alike. Thus it is noteworthy that Gideon Rosen offers us an argument for moral responsibility skepticism (hereafter ‘MRS’) bereft of either of these standard features.

In essence, the argument is two-staged. First, Rosen argues that for any agent’s action to qualify as blameworthy, an instance of genuine akrasia must appear at some point in the “causal etiology” of that action. Second, Rosen argues that we never have good grounds for judging an action to be akratic. Let us consider each stage in turn.

1. The first argument is, in my judgment, quite compelling. Imagine Sally performs wrong action A. There are two relevant possibilities here. The first is that Sally knows that A is a wrong action. If so, we have a clear case of akrasia, and clear grounds for an attribution of moral responsibility (in the form of a judgment of blameworthiness). The second possibility is that Sally does not know that A is a wrong action. In this case, we say that Sally acts “from ignorance” – either “circumstantial ignorance” (wherein Sally’s failure to know that A is wrong results from her ignorance of some matter of fact) or “normative ignorance” (wherein her failure to know that A is wrong results from her ignorance of some normative or moral truth). Does Sally’s ignorance – whether circumstantial or moral – excuse her from responsibility for her wrong action? Only if she bears no responsibility for her ignorance itself. If Sally is not to blame for her ignorance, we regard this ignorance as a mitigating factor, and exempt Sally from judgments of responsibility for performing wrong action A. If Sally is responsible for her ignorance, however, we say that she is “culpably ignorant”, and we do not regard her ignorance of A’s wrongness as an excuse.

When, then, is Sally morally responsible for her ignorance that A is wrong? When she is morally responsible for her failure to discharge one of her “procedural epistemic obligations” – the procedures “any reasonably prudent person in [her] circumstances would have done in order to see to it that [she] was adequately informed” (301). (Such requirements are “obligations to do (or to refrain from doing) certain things: to ask certain questions, to take careful notes, to stop and think, to focus one’s attention in a certain direction, etc.” (301).) And when is Sally responsible for her failure to discharge one of these procedural epistemic obligations? Here is the crux of Rosen’s argument: Sally is thus responsible only when her failure to discharge these epistemic obligations is itself a clear instance of akrasia. Were Sally (non-culpably) unaware that she stood under any such procedural epistemic obligation, her failure to discharge this epistemic duty – and thereby her resulting ignorance regarding A’s wrongness – would be non-culpable.

Thus, we see that assignments of moral responsibility for the performance of wrong actions must always terminate with the identification of an akratic act – either Sally’s performance of A is itself akratic (Rosen terms this a case of “original responsibility” for A), or Sally’s ignorance of A’s wrongness derives from an akratic failure to discharge one of her procedural epistemic obligations (a case of “derivative responsibility”). Attributions of moral responsibility thus track attributions of akrasia: any culpable wrong action A must either itself be an akratic action, or must stem from an ignorance made culpable by an akratic failure to discharge epistemic obligations, causally ‘upstream’ from the action.

2. Having identified akratic action as a necessary condition of any agent’s bearing moral responsibility for any wrong action A, Rosen now turns to the second stage of his argument. And this stage amounts to little more than the bare assertion that reliable attributions of akrasia are difficult to make. According to Rosen, we have scant grounds for such attributions, even in cases of first-personal attributions of akrasia. This claim rests on an appeal to the general “opacity of the mind” (308), together with the observation that it is often difficult to distinguish cases of genuine akrasia from those of “an imposter – ordinary weakness of will” (309). (In contrast to the akratic agent, who “judges that A is the thing to do, then does something else, retaining his original judgment undiminished,” the moral weakling initially judges “that A is the thing to do, but when the time comes to act, loses confidence in this judgment and ultimately persuades himself … that the preferred alternative is at least reasonable” (309).) To illustrate what he means, Rosen offers reflections on his own reflections: “When I consider my own case and ask whether some weakish act of mine amounts to genuine akrasia as opposed to ordinary moral weakness, I have no trouble identifying tolerably clear cases of the latter; but I confess that I cannot identify clear examples of the former with any confidence” (309). A fortiori, Rosen concludes, we ought to have very little confidence in our second- and third-personal attributions of akrasia.

Rosen’s argument does not depend on controvertible metaphysical theses (for instance: an antecedent defense of an incompatibilist determinism) ‘upstream’, nor does it entail ‘global’ MRS (the sort of skepticism that regards attributions of praiseworthiness to be as ill-founded as assignments of blame) ‘downstream.’ For this reason alone, it is an important contribution. However, Rosen’s overall argument is only as strong as the claim that culpability requires akrasia – that in order to be responsible for performing some wrong action A, one must confidently judge that A is wrong, even as one elects to perform A. Philosophers who question Rosen’s tight linkage between culpability and akrasia will find much in this article that is valuable and instructive, but they will not find here a compelling reason to embrace MRS.

Reviewed by Vaughn Bryan Baltzly
Bowling Green State University