Variantism

John Doris, Joshua Knobe and Robert Woolfolk’s (hereafter DKW) new paper on responsibility has been posted over at the Garden of Forking Paths. Most of the experimental work will be familiar to regular readers; the point of the paper is to argue for a particular view about responsibility on the basis of this work. DKW argue that philosophical approaches to responsibility are usually committed to two claims: invariantism and conservatism. Invariantism is the view that there is a single set of responsibility attribution criteria; conservatism is the view that the correct theory of moral responsibility will systematize folk attributions of responsibility reasonably well, so that folk attributions constrain, relatively tightly, the correct attribution criteria. DKW argue, on the basis of the experimental evidence, that in fact invariantism and conservatism conflict: folk attributions of responsibility are variantist.

DKW amass a large amount of evidence for the claim that folk morality is variantist. First, they argue, folk morality is neither compatibilist nor incompatibilist, but both at once: the folk have compatibilist intuitions with regard to concrete cases, but incompatibilist responses at a more theoretical level. Second, folk attributions of responsibility are sensitive not only to the psychological relations between agents and acts, but also to the valence of the action and the seriousness of its consequences. Thus, for instance, the folk hold people responsible for producing foreseen side effects when these outcomes are negative, but not when they positive, and they hold agents more responsible for negligent actions that produce seriously bad outcomes than for the same negligent actions when they produce trivially bad outcomes (in all, DKW identify four asymmetries in responsibility attribution).

Now, we can take issue with the details of the experimental paradigms in one or other of these experiments (as a matter of fact, I do). But it is very likely that DKW are right: there really is a lot of variation in folk attributions of responsibility. If that’s right, we can’t have both invariantism and conservatism. DKW argue that we must either become invariantist revisionists, or variantist conservatives (I note that there is a third option available – variantist revisionism – but it seems rather unattractive). They argue that we ought to plump for the latter.

As I think they will admit, their suggestion is a little too underdeveloped to be properly assessable. Indeed, it faces a dilemma, of which they’re well aware: on the one hand, too much variation, in the absence of a principle explaining and justifying the variation, and the resulting mess will better motivate scepticism than variantist conservatism; on the other hand, too much unification by principles and the result will be an invariantism. The success of their theory depends upon finding a middle way between these horns.

Here I want very briefly to raise a methodological issue. Conservative variantism is a possible view just in case conservatism is a possible view: that is, that folk attributions are stable enough for us to be able to say how the folk would respond to this or that case. But is conservatism possible? I’m worried that it isn’t: that in fact DKW only think it is because the studies they cite have limited their attention to a narrow range of questions.

I’m too lazy to check, but I suspect that these studies used a between-subjects design. Why? Because a within-subjects design would yield different results: subjects would become aware of a prima facie inconsistency between their responses, and would therefore be motivated to make them internally inconsistent. If that’s correct (and it is an empirical hypothesis) then how subjects judge these cases depends upon the nature of the study design. So the methodological question is this: what is the rationale for limiting ourselves to the between-subjects design when we want to uncover how the folk attribute responsibility? A within-subjects design doesn’t obviously contaminate judgments. If I’m right that both kinds of ways of probing folk morality are valid, and that they yield different responses, then conservatism isn’t on the table at all. And if that’s right, then the evidence DKW cite is evidence for revisionism or for scepticism.

Comments

[Joshua Knobe]

Neil,

These are excellent points, and I think the suggestions you offer provide a very good way to investigate this issue further.

In thinking about these questions, though, it seems important to keep in mind that there is a continuum of possible views. On one end is an extreme conservativism that says that everything about the way people ordinarily make moral responsibility judgments is completely correct and should not be changed in any way. On the other end is an extreme revisionism which says that the way people ordinarily make judgments of moral responsibility is radically mistaken and that we ought to replace it with a system that applies a single invariant standard in all cases.

Our paper wasn't intended as an argument for extreme conservativism. Rather, we tried to show that the only way to introduce a single invariant standard was to make certain radical revisions in the ordinary approach to moral responsibility judgment. This kind of radical revision, we suggested, does not seem warranted.

[jonathan weinberg]

Joshua, I'm not sure y'all really have the means to argue for your intended claim that _radical_ revisions would be required -- basically, because I don't see where you've got a good metric for radicalness in place. Not all revisions are of equal importance, after all, since not all things being changed were equally important to begin with. And all you've got at this point is the result that lots of _psychological effects_ will have to be ignored by an invariantist. But does running roughshod over those effects amount to proposing "radical revisions in the ordinary approach to moral responsibility judgments"? I just don't know, because I don't know how large a role these various effects play in our real, everyday moral cognition. Issues of ecological validity really do come into play here in a way that y'all just don't take seriously enough -- in particular, in your mss. you make a sort of 'well, at least we're better than armchair philosophizing!' move in response to these worries, but that only works against one very particular kind of opponent. And it doesn't work against one who says, e.g., "we need to be conservative about all the ordinary moral cognition _that matters_; but investigations like the ones you cite don't distinguish between aspects of our cognition that matter and aspects that don't; so your arguments fail against properly-construed conservatism".

I think that the most relevant section of the paper, with regard to this worry, is on p. 44 of the mss, concerning the invariantist's piecemeal rejection of various of these effects by means of Vargas' normative adequacy condition. And I'd note that, at that point, you don't sound like you're arguing for the victory of variantism, but for at best a kind of pro-tem evidential parity. But you're trying to overturn, really, the entirety of tradition here -- it doesn't take much theoretical conservatism at all to imply that you've got to do a lot better than fight the invariantist to that kind of draw! Without addressing the question of which revisions would count for how much, you can't establish your right to call these revisions "radical". And without that, there's really not an argument against invariantism here.

(Having said all that: overall I really like the paper, and I do think that just putting variantism into play is a very important philosophical result!)

[Neil]

Given that I think that wide reflective equilibrium is itself intuitive, I think variantism isn't in the running at all. There is no view that systematizes all the intuitions that the folk find compelling. The view I'm concerned to refute - which isn't to say I can refute it - is scepticism. I have a view that I think does a fairly good job of systematising the most important intuitions; Manuel has another, and there are others out there. The sceptic wins if they can make a compelling case that some of these (incompatible) views are as good as one another, in terms of revision costs, and that there is no other grounds to choose between them.

[Yentz Mahogany]

"Thus, for instance, the folk hold people responsible for producing foreseen side effects when these outcomes are negative, but not when they positive". There's likely a reason for that, or at least a possible explanation. The word, "responsible", has largely to do with what the agent can possibly be held blameworthy for. Entirely different ideas are in play for notions of praiseworthiness (though perhaps, arguably, they ought not be).

[Neil]

Yentz,

The paper canvasses that explanation: there is, as they say, a well-known negativity bias. Unfortunately, for the invariantist conservative, we get precisely the opposite pattern of responses to some other kinds of cases.

[Yentz Mahogany]

Bah, thanks for that. I'll read the whole thing when I have time.

My first concern is that "invariantism" is a complex idea. It seems to me that it is composed both of what has been referred to as the "constancy assumption" (the idea that values hold the same relative weight in moral judgment across contexts) and the universality criterion (that genuinely moral concerns not be associative duties, but apply universally), though restricted in scope to the domain of responsibility. One aspect may be more troubling than the other. (If this has also been treated in the paper, I apologize, and may be ignored suitably.)

Second, I wonder, to what extent does the abandonment of conservatism and invariantism in this context lead to the abandonment of notions of "common sense morality" or "folk morality"? When people appeal to moral "common sense", it is often in the context of general rules which are, well, common. I myself am skeptical of the notion of common sense, but I wonder if the authors are.

[john m. doris]

Jonathan:

I am puzzled by your remarks on "ecological validity." Since some of your work has deployed social psychological methods akin to those in DKW, you apparently do not -- unless you have repudiated your previous efforts -- think that *all* social psychology experiments of the sort at issue fail standards of ecological validity. What is it about the studies referenced in DKW (a good many of which appeared in refereed psychology venues) that makes worries about ecological validity especially acute?

jmd

[Jonathan Weinberg]

(Sorry it took me so long to reply, John -- much end-of-term craziness, plus two conferences and the OPC!)

I'm not concerned about the ecological validity of the basic type of task involved -- i.e., the evaluation of hypotheticals about fictional characters & situations. I could imagine a hard-core embodied/embedded/situated kinda theorist worrying about that, but it does seem to me that such hypotheticals do play a significant role in our quotidian moral practices.

Or, rather, that a class of such hypotheticals do so. Most such hypotheticals that we regularly entertain, moot, dispute, and judge are what we might call "metaphysically light": they can be sketched with just a few ordinary physical & psychological facts, about what was done & the state of mind of the relevant persons. But the Nichols & Knobe cases are "metaphysically heavy" -- they involve rather large-scale suppositions about the fundamental causal structure of the worlds in which their scenarios unfold. And outside of a philosophy classroom and a rarified class of sci-fi stories, I don't think you'll find too many such metaphysically heavy cases being entertained, mooted, disputed, or judged.

So my concern is about the ecological validity of the materials, i.e., the use of metaphysically heavy cases. On the whole, we're not used to thinking about them, and much more to the point, they just don't seem to play any role whatsoever in our ordinary moral practices. So, if a would-be invariantist conservative wants to give up on those cases (I guess they'd best be a compatabilist invariantist conservative), then they can very plausibly contend that they just are not calling for a _radical_ revision. The stuff that needs revising out just wasn't doing anything important to begin with.