Attributionism

A quick follow up to the discussion of the question "What is moral responsibility?" My paper on the preferability of a volitionist account of MR to an attributionist account has just been published in The Journal of Ethics and Social Philosophy, which (for those of you who don't know) is a web-based, peer-reviewed journal attempting to establish itself as a leader in the field.

The paper is here. Hope some of you find it interesting.

Comments

[Kip Werking]

Looks good, Neil. I look forward to reading it. :)

[John Fischer]

Yes, it looks great, but I am used to reading things horizontally, not vertically. I suppose it is better (for me at least) than Hebrew...

Cheers!

[Kip Werking]

I skimmed over the paper. I was happy to see a discussion of hard compatibilism (defined as the compatibility of free will with one's life being designed), as a consequence of attributionist theories, as well as the attributionist concern that:

"Perhaps Scanlon thinks that that way an infinite regress lies; acts of choice could only ground our responsibility if we were responsible for our ability to choose, and such a demand is in principle unsatisfiable. On this view, only an ens causa sui could be responsible."

This seems to echo what I wrote in an earlier comment here at the Garden: "I suspect that the road to historical compatibilism leads to TNR..." There is an interesting polarization here, which I have described before: hard compatibilism, on the one hand, which doesn't quite do justice to intuitions about TNR and self-creation, and being-causa-sui, which is so metaphysically extravagant (indeed, logically impossible), as to seem to be an unreasonable demand. Sophisticated hard compatibilists, recognizing the latter problem, content themselves with the former view of moral responsibility.

Of course, these views have not been shown to be exhaustive (although, I've come to think of them as the best alternatives), and Neil defends a middle-ground. To be honest, the definitions of attributionist and volitionist views of moral responsibility seem much too vague to me, to work with. Perhaps I am just missing something. Neil cites Wolf's work as such a legitimate possibility:

"Suppose, with Wolf (1990), that responsibility requires normative competence. Whether or not agents possess that competence is, typically, not up to them. Instead, it is a matter of luck: Good upbringing, native capacity and so on, is required, and none of these is (initially, at least) within the control of the agent."

I don't find Wolf's account persuasive, but I'm not sure that bottoming-out-with-character versus bottoming-out-with-competence is the proper of way of characterizing this problem. If I understand Neil correctly, he feels that by adding certain epistemic conditions to Wolf's competence account, he achieves something superior to Scanlon's view:

"The volitionist account allows us to give the right responses to these kinds of cases: If the agent possessed the relevant kind of control over their competence, where “relevant control” includes the satisfaction of relatively stringent epistemic conditions, then she is responsible for her competence or her lack thereof."

But as soon as Neil does this, Scanlon's concern about being causa-sui arises. Indeed, one might say that Neil has added a bit of "sugar" to his account, to make it more tasteful, such that, if the regress bottoms-out-at-competence, the agent can have, apparently, some control of whether they are competent. Thus the regress bottoms out with the agent in control. This is precisely what Galen Strawson's Basic Argument shows cannot happen, and so Strawson and Levy must have two different accounts of what this explanatory regress is, and how one can satisfy it.

But this brings me to my point: why can't Scanlon *also* add "sugar" to his account, so that an agent might, apparently, have some control over their character, such that the regress ends with the agent in control? In other words: what has escaped me is the fundamental difference between character and competence, such that the move Levy makes is available to his view, but no-similar-move is available to Scanlon? It seems to me that such a move is available to both of them, that either of them can adopt more historical or more time-slice views. Furthermore, Scanlon declines to go down this road, not because bottom-out-at-character views limit him in this way, but because he recognizes the futility of doing so. Those who follow Levy's strategy are chasing something they cannot reach.

It seems to be that neither view escapes this vicious regress. One can add all of the "sugar" (such as detailed accounts of epistemic conditions) one wants, but ultimately agents are not causa-sui (and the regress is properly characterized by Strawson), and furthermore their entire lives could be the product of design.

One final note: the discussion of apology and responsibility did not quite convince me. The area is too fuzzy to make any definitive statements, but I felt that there was at least a mixture of responsibility and non-responsibility in such cases as being-so-busy-one-steps-on-another's-foot. We might say roughly that, in all possible universes, the number X of universes where the person steps on the others foot is small. But by noting this I'm not sure they are signalling lack of responsibility, or just lack of responsibility, so much as regret, guilt, and the unlikelihood of repeat mistakes.

Such discussions of X bring me to a tangent: considerations of perfect-agents, or angels (Galen Strawson calls them Epictecians, I think), where X is not just small but zero, and the sense in which such people *lack* free will. For example, I find it interesting that Fischer (i) agrees with Strawson about subjectivism and (ii) finds strong reason-responsiveness too demanding to be critera for moral responsibility. If Fischer agrees with Strawson about subjectivism, does he also agree with him about the sense in which perfect agents have *less* responsibility? Does he feel the pull of that intuition at all, and if so, how does he account for it? It seems to me that compatibilists will feel tempted to disguise this intuition, if it exists, by focusing on cases where agents aren't perfect, in which they are only moderately reason responsive.

[Mark Smeltzer]

Kip,

Fischer does argue that strong reasons-responsiveness is too demanding, but I don't recall anything in his works that suggest he believes that agents exhibiting SRR would be less responsible than agents exhibiting MRR.

In other words, Fischer seems to be establishing MRR as the floor, not the ceiling.

Also, could you clarify which Strawson you are referring to when you say that, "I find it interesting that Fischer (i) agrees with Strawson about subjectivism"?

[Kip Werking]

Mark, I know that Fischer doesn't explicitly acknowledge any problem with SRR. Your phrase about ceilings and floors captures this nicely. My intention was just to speculate about whether Fischer would find SRR troublesome in this sense. This is so because I would *expect*, if there is any truth to Strawson's (about subjectivism, I'm referring to the younger Strawson) claim about perfect agents, that hard compatibilists would notice some tension on this issue.

In other words, just as speculation, might a compatibilist feel tempted to disguise the problem (if it is a problem) of perfect agents by focusing upon imperfect agents?

[Neil]

Kip,

I *think* (hope) you've misread me. I'm not claiming that for any competent (in Wolf's sense) agent, that agent is responsible for her competence. I'm claiming that there are cases - probably rather rare cases - in which agents are responsible for their competence or lack thereof. I'm thinking of cases like this: I want to kill someone, but I don't want to go to jail. I reason that if I am insane when I kill them, I won't be responsible. So I take the insanity pill, which predictably puts me in a state of insane rage, and kill my target. I have so arranged things that the target will be the only person present while the effects last. Now, I think that when I am on trial, my insanity plea won't cut much ice. The court will say that I was responsible for my lack of competence. I argue that attributionism can't make much sense of this case. But that's not what matters here. What matters here is that this kind of case is unusual (you will have noticed). I'm not claiming, nor am I committed to the claim (indeed I reject the claim) that agents are responsible only if they are responsible for their competence, nor am I claiming that agents are generally responsible for their competence. Indeed, I have a paper in which I argue that agents are generally not responsible for their characters.

[Kip Werking]

Neil, thanks for pointing this out. I apologize for the less-than-delicate reading of your article. I have misread it (my understanding was based, in part, upon the quote about Wolf's view and competence, above). Having read the example you give here (I tend to work better with examples), I think your point makes good sense: if attibutionists are willing to accept a certain bite to their view (as described in your example about the manipulator who permanently changes one's character), then this leaves them vulnerable to strange examples such as the one you just cited, whereby agents attempt to escape blame by rendering themselves incompetent. If I understand you correctly, attributionists are vulnerable to both of these sorts of scenarios (manipulator and self-induced incompetence).

One concern of my remains: doesn't your paper argue for a fundamental difference between bottoming-out-at-character and bottoming-out-at-competence views, such that the latter are superior to the former?

I hope I haven't continued to misread your paper. Part of the problem, I suspect, is that it remains fuzzy to me, exactly what attributionism is.

[Neil]

Attributionism is the view that an agent is responsible for an action if that action is properly attributable to her. That is, it reflects her identity as a practical agent, reflects her sense of what matters and what does not (best argument for attributionism - one that neither Scanlon nor Smith actually makes - is that it gives us a natural way of accounting for responsibility in Frankfurt-style cases). One problem with the view is that it is unable to incorporate historical considerations convincingly: Phineas Gage is responsible for his actions, immediately it becomes obvious that his character change is permanent.

I don't have a quick argument against bottoming out in character: the paper is the argument, because 'responsibility bottoms out in character' is the essential attributionist claim. The summary is, I suppose, that responsibility requires that we know what we're doing, and whether or not our action reflects our character is simply irrelevant to that question.

[Kip Werking]

Thanks, Neil. I think I'm starting to better understand. I'll have to give your paper another read. :)

[Mark Smeltzer]

Kip,

My interpretation of Fischer puts him squarely in the camp that essentially views responsibility as a derivative of a capacity for self-expression, which is by nature opposed to views that derive the notion of responsibility from a capacity for self-definition (which seems to be indicative of G. Strawson's view).

If this categorization is correct, it is hard for me to see why Fischer (or anyone else in the self-expression camp) would have trouble attributing robust responsibility to agents that are SRR. (Even if I have pegged Fischer incorrectly, this can be applied to those of us who do belong to this camp. I suppose John could speak for himself on this matter.)

[John Fischer]

I sometimes prefer it when others speak for me, esp. when they do it better than I can. I haven't studied this thread carefully, but my view is an agent is morally responsible insofar as his or her behavior issues from his own, appropriately reasons-responsive mechanism. SRR is an even more demanding criterion than MRR, and MRR (moderate reasons-responsiveness, as laid out in RESPONSBILITY AND CONTROL) is enough. Both components, mechanism ownership and MRR, need to be present.

In my view, these components--constituting guidance control--are the freedom-relevant conditions for the application of the concept, "moral responsibility." The value of acting freely and being responsible is the value of aritistic self-expression.

[Kip Werking]

Mark and John,

I want to be careful here: the interpretation you have given of John's view (Mark has explained it twice and John has explained it once now) is not something I intended to challenge. Indeed, that is exactly what I took John's position to be after just having finished Responsibility and Control. It may help to reconsider what I originally wrote:

"If Fischer agrees with Strawson about subjectivism, does he also agree with him about the sense in which perfect agents have *less* responsibility? Does he feel the pull of that intuition at all, and if so, how does he account for it? It seems to me that compatibilists will feel tempted to disguise this intuition, if it exists, by focusing on cases where agents aren't perfect, in which they are only moderately reason responsive."

If we were just going by what Fischer has already written, then the obvious answer would be no, he doesn't consider perfect agents (or SRR) problematic, and he doesn't feel the pull of this intuition (which Strawson) describes at all. However, in that case, just what was Galen Strawson talking about? I, for one, feel the pull of this intuition, and the reason I feel it is that the programmed nature of perfect agents, or people with SRR, is especially striking is such cases. Computers can perform flawlessly, and if a human were to perform flawlessly in the moral realm, she or he may start to seem more like a computer, in a way that may undermine our ascriptions of moral responsibility.

So, with these expectations, I was speculating about how hard compatibilists such as Fischer would integrate such an intuition into their view. Alternatively, they may simply deny that it exists, and this seems to be what Fischer does. I may, however, be mistaken. :)

[John Fischer]

Hmmm. I hadn't really considered "perfect agents". Anyway, I think that the intuition you speak of may come from other problems with computers' being agents, not specifically their (putative) SRR. (That's supposing a computer can have reasons, which is contentious.) My view requires BOTH mechanism-ownership and RR.

But I'm afraid I haven't really thought through these issues--sorry that I don't have time to do them justice right now.

[Mark Smeltzer]

Kip,

Discussion of Fischer's view aside, does it strike you as odd that there are those of us who agree with Locke (or at least Ayer's interpretation of Locke) and his conception of perfect agency? The ideal Lockean agent is SRR *and* tracks objective value. I agree with Locke regarding this concept and the ideal it presents.

You seem to be approaching the question from the self-definition point of view -- placing great conceptual value on the capacity for an agent to change and re-define itself. As a result, you see the agent exhibiting SRR as inferior because it seems to be "forced" (somehow) or locked into the responses it experiences. Consequently, one might be tempted to say that SRR entails a degraded or lesser form of responsibility than does MRR, which seems to offer the agent some "wiggle room."

Rather than to address this incompatibilist intuition directly, I'd like to paint an alternative picture. Imagine a person whose self-knowledge is complete (supposing that is even possible to imagine!) and as a result, he knows two practical things: 1) exactly who he (in the neuter sense of course) is, and 2) exactly what to do in various situations in response to various reasons (given his full self-knowledge).

If you understand the person I've depicted, then you'll see that this person wouldn't ever have to deliberate in the fashion that you and I are accustomed to doing. He will simply know what he has to do in every possibly situation in virtue of that fact that he knows exactly who he is. However, that does NOT mean he is omni-competent. When presented with new normative contexts (contexts in which there are distinct correct and incorrect behaviors/actions), he will go about the business of learning and obtaining competence in that area, and in any given instance he will take the consummate body of knowledge obtained to that point and act accordingly -- which is to say that this person will still make mistakes, even moral mistakes (so this person is not an ideal Lockean agent).

The person I'm describing still sounds like a *person* to me (he will still feel, experience, and reason as you and I), even though I cannot relate to the type of thought life that this kind of person would enjoy. This person doesn't sound at all like a computer (what in the heck would it mean for a computer to have complete self-knowledge?). Moreover, it is plausible to think that as people age, some will obtain higher degrees of self-knowledge than others, and some of these may even toe the line between MRR and SRR as a result.

I think there are two ways to respond to this case.

The first is to deny that the strength with which one responds to variations in reasons would increase proportionally with one’s degree of self-knowledge. However, I think that is patently false. (If there is disagreement on this point, I may choose not to pursue this line of inquiry because I don’t want to spend a lot of time defending this idea just yet.)

The second is to deny that such an agent is possible. In response to this claim, I think the consideration above sheds some light on why this response fails: we have an idea of what it means to increase one’s self-knowledge, and that this happens over time. So, given a person who is able to learn about themselves in a reliable and consistent manner will eventually approach something close to SRR, given enough time. So, this would be the difference between an agent with perfectly strong RR and an agent with just strong RR; the degree of difference owing to the (presumably) ever existent epistemic gap for the agent without perfect self-knowledge. Now one has the burden of establishing the relevant difference between the SRR agent with perfect self-knowledge and the SRR agent with near perfect self-knowledge. Why should we think the one is less responsible for their actions than the other?

Lastly, one could respond to this response by stipulating that the person who develops SRR as a result of obtaining near perfect self-knowledge is in a process of becoming less responsible. However, this response is likely to fail for two reasons: 1) on the face it is simply a rejection of the semi-compatibilist position in general, and 2) if it is not simply the rejection of the semi-compatibilist position, then one must explain what the MRR agent lost in becoming SRR on these grounds, and I think that challenge will ultimately prove insurmountable.

That said, I think the best option for the dissenter is to formulate arguments in favor of rejecting the semi-compatibilist position simplicter; as opposed hurling what Dennet might call “bogeymen” at us. There may be other ways of explaining the relevant conditions that could make an agent SRR which would make an agent less responsible, and it may be worthwhile to discuss those in another context, but I would agree that any formulation of SRR that does negatively impact the quality of responsibility enjoyed by the agent would be most undesirable.

[Eddy Nahmias]

Mark, your thought experiment sounds similar to one I have developed in a paper that should be coming out in Philosophical Studies soon, "Close Calls and the Confident Agent." My confident agent does deliberate but, after deliberating, she always comes to a confident decision about what she really wants to do. Hence, she never faces the "close calls" many libertarians (e.g. van Inwagen and Kane) require for free will to exist (all free actions must be traceable to a close call choice for them). I try to describe her in such a way that she is intuitively free and responsible to pump the intuition that close calls are not necessary for free will or moral responsibility. I'll have the paper posted here.

In my dissertation and papers in progress, I emphasize the role of self-knowledge for free will. I'd love to hear your (and others') thoughts on this idea.

[Kip Werking]

Ok, Neil, I reread the paper. I think in my rush to show how not-being-causa-sui would be problematic for both bottoming-out-at-character *and* bottoming-out-at-competence views, I attributed (!) a bottoming-out-at-compentence view to you.

Now that I've reread your paper, I see that you do not necessarily defend such a view. Instead you note bottoming-out-at-competence views just to to argue against the attributionist's claim that: "from the fact that we cannot have chosen to possess the capacity to make choices, that our responsibility does not bottom out in our acts of choice." You make this argument in the context of a larger defense of volitionism (which emerges through your critique of attributionism).

I think that, once one recognizes that you don't actually defend the view I criticize above, that the thrust of my comment remains: is the hardness of attributionism specific to bottoming-out-at-character views? Is the attributionist's emphasis upon character what causes the dilemma, or might the volitionist's emphasis upon choices be just as problematic? This is the fascinating question which you ask:

"Perhaps Scanlon thinks that that way an infinite regress lies; acts of choice could only ground our responsibility if we were responsible for our ability to choose, and such a demand is in principle unsatisfiable."

Perhaps there is something illusory or unsatisfying about the volitionist's bottoming-out-on-choices. I don't pretend to know what this would be, but it remains suspicious how intelligent persons such as Scanlon find the bite of attrituionism preferrable the bite (in their eyes) of volitionism.