- Defusing the Demandingness Objection: Unreliable Intuitions.Matthew Braddock - 2013 - Journal of Social Philosophy 44 (2):169-191.Dogged resistance to demanding moral views frequently takes the form of The Demandingness Objection. Premise (1): Moral view V demands too much of us. Premise (2): If a moral view demands too much of us, then it is mistaken. Conclusion: Therefore, moral view V is mistaken. Objections of this form harass major theories in normative ethics as well as prominent moral views in applied ethics and political philosophy. The present paper does the following: (i) it clarifies and distinguishes between various (...)
This is a very interesting paper. I am in agreement with the basic premise, namely, that we should be suspicious of moral intuitions which are highly contingent or "flippable". However, I have one or two questions about the argument.
In one section, you're dealing with the problem of "typing" mechanisms. The point, as I understand it, is to show that your argument defeats demandingness intuitions but does not defeat other moral intuitions (such as those concerning the wrongness of slavery). You say:
Here, you seem to be claiming that the fact that racist beliefs are false gives us warrant to "type" the mechanisms narrowly in this case, i.e. to say that racist and non-racist beliefs are somehow produced differently. On the face of it, this just seems like the wrong kind of consideration: we need sociological/psychological/developmental description here, not truth-conditions. After all, it is a familiar fact that just about exactly the same sorts of mechanism can produce true and false beliefs.
A more direct version of the problem is this. Take two general moral beliefs, M and ~M, and assume the bivalence named in your premise (10). If a sufficient condition for their having been produced by different socialization-mechanisms is that one of them is false, then you've actually provided an argument which protects demandingness intuitions, since, on this view, every intuition will be "non-flippable": its contrary will have been produced by a different type.
Anyway, one of the things I really like about this paper is that it anticipates objections that I naturally had as I was reading it, so thanks for being so responsive within the text. My view, however, is that one of these responses doesn't work, and that you will end up debunking a lot of our more 'sacred' intuitions. Best regards,
Hi Nick,Your comments much appreciated.
Your main comment focuses on the Argument from Socialization Effects (pp. 177-182). So let me briefly characterize the argument. Observation: many of us find some moral views to be intuitively too demanding. Why? Empirically plausible to suppose that our socialization has much to do with it. This suggests that if we had a slightly different socialization--i.e. if our parents and culture had expected more from us in terms of sacriﬁcial giving--we would have probably arrived at different intuitions about the demandingness of some moral views. In other words, our demandingness intuitions are subject to "socialization effects." This highly contingent character or "flippability" of our demandingness intuitions suggests that they are unreliable.
The flippability claim is understood more precisely this way: in some nearby worlds different instantiations of the same socialization process types that are instantiated and lead to our demandingness intuitions in the actual world and close worlds just like it would lead us to intuitions that directly oppose these intuitions. Since the opposing sets of demandingness intuitions in the respective worlds cannot both be right—these moral principles either are too demanding or not (i.e., are mistaken or not)—we have been led astray in some nearby worlds.
So then, on a fairly standard characterization of unreliability where a belief is unreliably formed if and only if it is formed by a process that does not produce mostly true throughout nearby worlds, we have good reason for thinking our demandingness intuitions are unreliably formed.
The anticipated worry--your worry, too--is whether the argument implies a rather sweeping moral skepticism. I argue No, given the *sort of contingency* it specifies.
"Importantly, the argument does not imply global moral skepticism, since many of our moral beliefs are not highly contingent or easily “ﬂippable” in the sense that in nearby worlds where different instantiations of our belief-forming processes produce the beliefs, we would arrive at beliefs that directly oppose the beliefs at which we arrive in the actual world and in close worlds just like the actual world." (p. 179)
"For example, consider my belief that slavery is morally wrong. It is plausible
Of course, the point about typing depends on how narrowly we should type belief-forming processes:
"If a quite general or expansive typing is the correct typing—for example, if a socialization-based belief-formation process in a racist society is of the same type as that in an anti-racist society—then the point in question would not hold. However, if some fairly narrow typing is the correct typing, then the point would hold: it in fact would have taken a different type of socialization process to get me to believe that slavery is morally permissible. It is important to recognize that the point would hold even if we cannot identify the correct typing among the range of narrow typings available. So to reinforce the point, we need no clear solution to the generality problem. Rather, what is necessary is motivation of the range of narrower typings over the range of general and expansive typings that would count socialization-based processes in a racist society as being of the same type as processes in an anti-racist society. With a narrower range of typings motivated, then with more conﬁdence could we say that my belief that slavery is wrong is not easily ﬂippable. Hence, with more conﬁdence could we say that The Argument from Socialization Effects implies no sweeping moral skepticism."It is pointed out that motivations for the narrower range of typings are available in current discussions of the generality problem:
"In recent epistemological discussions, reliabilists and others who deal with the generality problem—broadly, the problem of how to individuate belief-forming process types—almost invariably favor rather narrow speciﬁcations of processes, even though they do not agree on how exactly we should select among the range of narrow speciﬁcations available..." (p. 180)Now let's get to the example in the passage you quote in your comments: what if my mother were a racist bigot who socialized me to believe that enslaving others is morally permissible? Would that socialization process--which centrally involves testimonial mechanisms--be of the same *type* as the actual belief-forming process in which my anti-racist liberal mother socialized me to believe in human dignity, moral equality, and the wrongness of slavery? I argue No, since such testimonial mechanisms are plausibly typed more narrowly than that.
You may ask: what is the basis of the narrow typing of testimonial processes? I quickly motivate narrow typings this way (at the end of the passage is the quote you cite):
"Plausibly, the socialization processes would centrally involve testimonial mechanisms or processes. And plausibly, testimonial processes should be typed fairly narrowly: for example, typed to some extent according to the domain of testimony, the testiﬁer, and the testiﬁer’s epistemic situation. Examples can motivate narrow typings....The testiﬁer matters: the reliability of my parent’s testimony should not necessarily be impugned by the unreliability of some other parent’s testimony. And the testiﬁer’s epistemic situation matters: the reliability of my parent’s testimony in the actual world should not necessarily be impugned by the unreliability of her testimony were her epistemic situation very different—for example, were her relevant foundational background beliefs false. Given these considerations, how generally should we type the testimonial process behind my moral belief that slavery is wrong? The reliability of (say) my mother’s anti-racist moral testimony in the actual world should not necessarily be impugned by the unreliability of her moral testimony were she a racist bigot, for her epistemic situation (i.e., her foundational moral beliefs) in the latter case would be radically mistaken. The two types of testimonial processes, then, are plausibly distinct." (p. 181)Without resolving the generality problem, my (quick) suggestion in the passage you quote is that my mother's foundational moral beliefs are radically different in these two cases, and so her moral testimony is these two cases would plausibly be of different types. I agree with you that it's not the truth or falsity of the foundational moral beliefs that makes the typing difference; rather, it's the fact that her system of moral beliefs is so different. (I presume foundational racist beliefs would tend to "infect" someone's system of moral beliefs more generally). For example, the reliability of my mother's moral testimony is not impugned by the fact that she would have testified to me differently were she a neo-Nazi. This is so because her testimony would be of a different type and it would be of a different type because the testifier's system of beliefs would be radically different.
Of course, this elaboration was not spelled out in the two sentences you quote! So I understand how one could gather from that quote an apparent quick inference from truth conditions to typing. My infelicitous words: "radically mistaken." It would have been clearer for me to say "radically different" in the passage. Thanks again for your encouraging comments and this opportunity to respond.
Reply to Matthew Braddock
Thanks again for the chance to discuss this really interesting issue. I hope you take these comments (from a random stranger on the internet) in the constructive spirit in which they are intended!
The basic question, as I see it, is this: can you make out what it means for a system of beliefs to be "radically different" in a way which does not make any reference to the truth or falsity of those beliefs? If you can't, then the suspicion must remain that it is truth-values that are doing the work. In your response here, you allude to this distinction, but you don't quite say what it is, and I must admit that I don't quite know what it is. By way of illustration, consider two sets of foundational beliefs:
B1: Race R is made up of people who are intrinsically inferior to others.
B2: R is made up of people who deserve fewer political rights than everyone else.
B3: R is made up of people who are disgusting and morally corrupt...(etc)
B4: Race R is made up of people who are intrinsically equal to others.
B5: R is made up of people who deserve the same political rights as everyone else.
B6: R is made up of people who are neither disgusting nor morally corrupt... (etc)
Extend these beliefs further until you end up with an egalitarian and a vicious bigot. The difficulty is that beliefs B1-B3 are about precisely the same set of people and concern precisely the same property-ascriptions as B4-B6. They are indeed different, but only truth-conditionally.
Anyway, if you've got the time, I'd love to hear more about the distinction between "radically different" and "radically mistaken". It strikes me that something like this distinction must be made and defended if we are to have any hope of sorting out reliable from unreliable moral intuitions. Thanks again,
Hi Nick, thanks for your useful (very constructive) comments. Quick response.
Of course, no attempt to solve the generality problem will be ventured here. But no such attempt is necessary to advance my argument, as I mention in the paper, as long as narrow typings of testimonial processes are adequately motivated.
To your question, my initial thought is this: two sets of beliefs are different if they have different contents. Two sets of beliefs can have different contents even if both sets of beliefs have the same truth values.
For example, there are a variety of racists who hold different foundational moral beliefs (e.g. about which race(s) possess higher moral status than other races), which lead them to different systems of moral belief. Such systems of beliefs can be radically mistaken (same truth value) and radically different from each other. For concreteness sake, a neo-Nazi and a KKK member may have very different but equally mistaken systems of moral belief.
So let’s return to your original question concerning the individuation of testimony-based belief-forming processes. What can plausibly distinguish between various types of racist testimony, I suggest, is not that some racists have true systems of belief and others have false systems—they both have false systems—but rather that their respective systems of beliefs are radically different. So in this example, it’s not truth values that determine the typing of testimonial processes, it’s the radically different contents of the systems of belief. The same goes with my example of my mother’s (actual) egalitarian testimony and her (counterfactual) testimony were she a racist bigot. I suggest that the two instances of testimony would plausibly be of a different type, since the contents of my mother’s foundational moral beliefs would be radically different in the two cases.
Perhaps you don't share this judgment, perhaps this is where we differ. Do you think that the testimony of my mother in both cases would be of the same type? If so, do you have a view (or leaning) on how generally testimonial processes should be typed? As I mentioned, most who work on the generality problem favor a narrow sort of typing (though they disagree on how to select among the range of narrow typings available). So I see my typing judgment here as cohering with that trend of work.
Now let me step back. How does this argumentative thread fit into the broader argument? I suggest that as long as belief-forming processes are typed fairly narrowly, my Argument from Socialization Effects does not have unpalatably sweeping skeptical implications. Hence, it is kept local.
Reply to Matthew Braddock
Hi again Matt,
The only sense I can make of a "type" of socialization is by thinking of it as a distinct social-psychological process. I suspect that we could generate a range of typings based on the kinds of mechanisms and social contexts that are in play when a person is socialized. So, here's a suggestion based on some Philosophy of Language. Take:
B2: R is made up of people who deserve fewer political rights than everyone else.
B5: R is not made up of people who deserve fewer political rights as everyone else.
I was suggesting that two pairwise sets of moral beliefs like this might differ only in their truth-values, but you are suggesting that they might differ in their "contents" as well. I think that this means that, for you, the meaning of a moral sentence is not reducible to its truth-values (i.e. you don't have a truth-conditional account of meaning, at least not here). This is interesting: if you were to be, say, an expressivist, you could say that the sentences have different "contents" (i.e. meanings) because they are expressions of radically different emotional/affective sensibilities (i.e. B2 arises out of fear and disgust, B5 arises out of less troublesome affective states). You could then "type" socialization-processes according to these psychological facts about socializers. There would still be hurdles, but this would be a start.
I think something like this must be your route. If you stick to a standard descriptivist account of the contents of moral sentences, where such sentences simply ascribe properties to things in the world, then B2 and B5 will remain indistinguishable except via their truth-values, and you will have to type the pairwise sets together. Anyway, I'll let you have the last word on this, as you've already been more than generous with your summer time.
Good stuff, Nick.
My Argument from Socialization Effects intends to stay neutral on the question of moral semantics. But is it actually neutral? I’m not convinced that the typing of socialization-based belief-forming processes hinges on the question of descriptivist vs. expressivist semantics. Perhaps this is so because I’m not convinced that the content of egalitarian and racist systems of moral belief is adequately captured by your B1-B3 and B4-B6, where racist systems of moral belief are characterized as the simple negations of egalitarian systems (at least in the case of your B2 and B5). Perhaps this issue could be sidestepped and progress made if we just focused on a different example. After all, there are lots of potential examples.
So instead of the moral intuition that slavery is morally wrong, consider the moral intuition that adultery
is wrong. Plausibly, this intuition is also
not easily flippable (at least in my case), since in nearby worlds where the same socialization
process types behind this intuition are instantiated and all else is held ﬁxed, it is plausible to suppose that I would not form the belief that adultery
is morally permissible (i.e. not wrong). The formation of that
intuition would have required a quite different socialization
experience—more precisely, the instantiation of different socialization
process types—such as being socialized by parents with a very different code of
sexual ethics, a different conception of marriage, even a different overall worldview. What could be a plausible basis for
distinguishing the (actual) anti-adultery testimonial process from the (counterfactual) pro-adultery testimonial process? My suggestion
would be the difference in the content of the testifiers' system of beliefs. Does this suggestion hinge on
the question of descriptivism/expressivism?
I don’t think so.
But is the claim which you find objectionable--i.e. my claim regarding grounds for typing--even essential to the larger Argument from Socialization Effects (pp. 177-182)? I don't think so. Here's why. The Argument does not essentially depend on the question of what makes it the case that a particular socialization process should be distinguished from another socialization process. If you plug in different grounds for narrow typing than the grounds I quickly suggest, the argument still goes through and avoids the objection that it implies a sweeping moral skepticism. What suffices for the cogency of the argument, as I tried to stress in the paper, is the claim that socialization processes (like all belief-forming processes) should be typed fairly narrowly. And this claim, in my view, is independently plausible, coheres well with the philosophical literature on the generality problem, and coheres well with typical descriptions of belief-forming processes given in the social and psychological sciences. I'm actually more confident in this modest claim concerning the plausibility of narrow typings than I am in the non-essential claim to which you are objecting, i.e. the claim that the contents of a testifier's beliefs play a role in distinguishing between different types of testimonial belief-forming processes.
All this is to say: even if the sort of objection you’re pressing were successful—even if you showed that I have not identified a plausible basis for distinguishing between various types of testimonial processes—it would not show that the Argument from Socialization Effects implies a sweeping moral skepticism. To show that would require more work. So I still find the Argument from Socialization Effects to impugn our demandingness intuitions (and other highly contingent moral intuitions) without impugning all (or an unpalatable range) of our moral intuitions.
Stimulating discussion, Nick! All to my benefit. (I don't mind if it continues or if it rests here).