Cuypers on Externalist Autonomy

In the latest issue of Philosophical Studies, Stephen E. Cuypers has written an excellent critique of externalist compatibilist autonomy and of Mele’s view in particular (“The Trouble With Externalist Compatibilist Autonomy”, Philosophical Studies 129:171-196). One recent development in my own view (the evolution of which never ceases to surprise me) is the embracing of an element with which I’ve always sympathized: internalism about moral responsibility. I just do not think there can ever be a rational basis for judging an agent as being morally responsibility, or not, for choices depending upon factors over which the agent had no awareness whatsoever—however convenient such a view might be for soft compatibilists who would like to distinguish between CNC controlled agents and “mere determinism.” Experimental philosophers, such as Josh Knobe, may compile data showing that the folk are not internalists—indeed, I suspect they are externalists in multiple ways—but I would regard these as widespread performance errors, no different than the countless other varieties of human irrationality (for a list of such irrationalities, see the Wikipedia article “

List of cognitive biases”).

Related to the question of internalism/externalism is the question of whether a view on freedom or autonomy should be historical. In the literature it seems common to distinguish between hard compatibilists—who are more willing to endorse a bullet-biting historical view—and soft compatibilists—who give historical factors more consideration. Harry Frankfurt has given perhaps the most eloquent statement of the hard compatibilist theme:

“A manipulator may succeed, through his interventions, in providing a person not merely with particular feelings and thoughts but with a new character. That person is then morally responsible for the choices and the conduct to which having this character leads. We are inevitably fashioned and sustained, after all, by circumstances over which we have no control. The causes to which we are subject may also change us radically, without thereby bringing it about that we are not morally responsible agents. It is irrelevant whether those causes are operating by virtue of the natural forces that shape our environment or whether they operate through the deliberate manipulative designs of other human agents (2002).”

I have had more difficulty in identifying soft compatibilists. Mele’s emphasis upon historical considerations has suggested him as perhaps the most prominent soft compatibilist, if he is one (I understand that he defends agnosticism about whether compatibilism or libertarianism secures human autonomy). Once one begins to make this distinction between soft and hard compatibilism, however, one realizes that these views are not discrete but rather help constitute a spectrum of more or less historical views. Thus, I doubt that even the most time-slice hard compatibilist could attribute moral responsibility to a murderer who came into existence one second before the murder, loaded into the murdering position (I welcome any suggestions for defenders of this view). Similarly, it seems to me that many or most defenders of human freedom (or autonomy) are willing to attribute moral responsibility to a murderer even if the relevant CNC control is not immediately local but ultimately distant. For example, it seems that many or most defenders of human freedom (or autonomy) would be willing to attribute moral responsibility to a murderer even if God (or whoever) designed the entire trajectory of the universe before it came into existence: Frankfurt, Fischer, Watson, Dennett and (if I understand his view correctly) Mele—even if Frankfurt, Watson and Dennett are willing to tolerate somewhat more local manipulation than Fischer and Mele are. Paul Russell seems to call this latter problem the notion of manipulation “at the horizon” (see his contribution to Kane’s Oxford Handbook). Gary Watson reserves this latter notion of ultimately distant manipulation for the term hard compatibilism and concluded his contribution to the Journal of Ethics issue in honor of Harry Frankfurt: “The philosophical alternatives for those who take freedom seriously (as I think we all must, in practice) are hard.”

If internalism about moral responsibility is right, then some free will deniers and compatibilists are left in the awkward position of agreeing with each other about something essential. From the similarity of “merely determined” agents to agents whose entire lives were designed by God (or whoever), deniers conclude that the design or manipulation argument succeeds. Hard compatibilists extrapolate in the opposite direction, to reach a conclusion that is superficially counter-intuitive, and perhaps only slightly more palatable: from the similarity of “merely determined” agents to agents whose entire lives were designed by God (or whoever), hard compatibilists conclude that such agents must also be morally responsible for their actions.

In defense of internalism, Stephen E. Cuypers has written “The Trouble With Externalist Compatibilist Autonomy.” In particular, Cuypers argues that Mele’s sufficient criteria for autonomy cannot satisfy his condition H:

“(H) The account must (a) be free of nonhistorical or internalist presuppositions; the account should remain ‘‘pure’’, i.e., untainted by self-identification or endorsement by the agent himself. It must (b) draw a principled distinction between an appropriate causal history and an inappropriate one; given that everything has a causal history on the assumption of determinism, causal origin and genesis can only make a difference, if causal histories can be sorted out in two separate sets: the set of the ‘‘right’’ (or legitimate) histories and that of the ‘‘wrong’’ (or illegitimate) ones. And, it must (c) be compatible with the truth of determinism, for the externalism under consideration is a variety of compatibilism.”

Cuypers further notes the problems with Mele’s 1, according to which authonomous action entails either a lack of unsheddable pro-attitudes (“Yet, neither rresistibility nor unsheddability seems to be a historical property.”) or a lack of pro-attitudes which bypassed the agent’s rational control faculties (but “[n]either the origin nor the genesis or formation-process of the greater part of my proattitudes (and anti-attitudes) falls under my control-capacities.”). In response, Mele might note that the criteria he offers were only intended to be sufficient and not necessary, for autonomy. But although this suggests a distinction, it is not clear how the distinction would be relevant. For example, if Mele’s sufficient criteria for autonomy are merely sufficient but nevertheless problematic, one might wish to know what other sufficient criteria apply. Mele may also note that his historical requirement for autonomy according to which manipulation cannot bypass an agent’s control faculties only applies to unsheddable pro-attitudes. The historical requirement does not need to apply to sheddable pro-attitudes because such manipulation is not necessarily problematic—the agent can just shed the implanted values. Cuypers might respond that, even if loosening the criteria for autonomous possession of sheddable pro-attitudes has a certain utility, it still remains an impurely externalist view and so violates his principle H. But perhaps Mele does not feel that H is a test worth surviving. Furthermore, I suspect that Mele’s distinction between sheddable and unsheddable pro-attitudes may face the following additional problems: (i) it seems to rely upon that most contention notion of “can” or “ability” and (ii) many or most of our most cherished pro-attitudes are unsheddable and yet have also bypassed our rational faculties of control.

Cuypers illustrates this last problem by noting the problem of young children and education. Such children cannot rationally evaluate the pro-attitudes that their environment gives them (and some of these pro-attitudes are biologically inherent in virtually all humans) because they have not developed the initial pro-attitudes with which to evaluate later ones. Although Mele and Cuypers seem to limit their discussion to the subject of autonomy, Cuypers notes that David Zimmerman criticizes Fischer’s view for substantially the same reasons (indeed, I doubt whether distinguishing between autonomy and moral responsibility has any utility in this area of philosophy). Zimmerman calls this “the puzzle of naturalized self-creation in real time.”

The distinction between autonomy and moral responsibility might also explain one curious feature of Cuypers’ great article: its apparent failure to even consider the possibility that people just lack freedom or autonomy. For example, Cuypers writes that “I will draw the over-all pessimistic conclusion that no party deals with this problem satisfactorily.” But there is one minority view, at least in the free will and moral responsibility literature if not the autonomy literature, which remains perfectly consistent with all of Cuypers’ arguments: free will denial. Later in the article, Cuyper seems to dismiss this possibility rather quickly:

“Still, Mele could bite the bullet and he might just claim that many more of our pro-attitudes than we normally suppose are inauthentic and thus nonautonomous. Apart from the counterintuitive ring of this reply, it strikes me as a hopeless retreat to a highly theoretical stance, or even worse, a suspicious last resort to the denial of autonomy, bordering one or other post-modernist free-wheeling Weltanschauung, such as, for example, that we all live inauthentic lives or that nonautonomy is our fate. This escape route, despairing as it is, would also lead to internal inconsistency, for Mele himself explicitly rejects nonautonomism: what he calls ‘‘agnostic autonomism’’—the disjunction of compatibilist belief in autonomy and libertarianism—is more credible than the view that no human being is autonomous (pp. 237–254; 2002, pp. 543–546).”

Finally, it is worth noting that, once Cuypers removes the unsheddable question from Mele’s criteria for autonomy (because it violates Cuypers’ principle H), these criteria reduce to a requirement for no pro-attitude—sheddable or unsheddable—to have bypassed an agent’s rational control faculties. This is a view which Mele would never defend but which comes close, I suspect, to representing the view of the untutored folk. Indeed, the notion that no pro-attitude has byspassed an agent’s rational control faculties seems to be a useful explication of what Galen Strawson (and Nietzsche) meant by the term “causa sui.” In defending such a view, people seem to make the mistake of failing to sufficiently distinguish the rational from the evaluative portions of one’s mind. One cannot rationally evaluate all pro-attitudes because the existence of pro-attitudes is a necessary requirement for rational control. In tracing the history of chosen pro-attitudes from earlier pro-attitudes, eventually one must meet pro-attitudes which were just given. Agents which have no attitudes, values or desires—if they are agents—cannot choose them ex-nihilo. Yet these initial pro-attitudes constrain the entire life trajectory of the agent and this is what gives the manipulation or design arguments such force.

Comments

[Neil]

One minor point, Kip. I doubt that folk externalism is a performance error. It's controversial how many of the biases are the result of such errors (if any). but this one isn't. It might be in conflict with other folk views, and it might be wrong (I think it is too). But there is no competency which is misapplied to yield externalism.

[Kip Werking]

Thanks, Neil.

Perhaps I misunderstand what makes something a "performance error" in this context. When you say "it might be wrong (I think it is too)", that seems to express all that I thought was necessary. But you also say "there is no competency which is misapplied to yield externalism." It's not clear what the difference is? Can you suggest any reading material to help explain this?

[Neil]

Hi Kip,

I'll try, even though I'm in transit on a longhaul flight. As I understand it, a performance error is an error in misapplying a basic competency. The distinction comes from linguistics, I think. The idea is that some kinds of mstakes are the result of something preventing a basic competency from being applied. So ignoring base rates or conjunction fallacies are, arguably, performance errors: when it's pointed out to people that they depart from the correct application of some competency, they retract the judgments. I'm pretty sure that Stein's book *Without Good Reason* discussess this question.