Welcome to our NDPR Review Forum on Philip Pettit’s book “The Birth of Ethics: Reconstructing the Role and Nature of Morality” (OUP 2018), recently reviewed in NDPR by David Phillips. Below is a brief description of the book, an excerpt from the review, and an initial response to the review by Philip Pettit. Please feel free to comment on any aspect of the book, review, and discussion below.
From the book blurb:
Philip Pettit’s response:
Defending counterfactual genealogy
David Phillips’ review of the Birth of Ethics (OUP 2018) offers a nice, brisk summary of the line of argument, something for which any author must feel gratitude, and then raises a challenging line of critique. He focuses on the way I rely on a counterfactual genealogy to generate an overall view of our ethical toolkit, in particular the dimensions associated with moral desirability and moral responsibility. While that view is neutral between divides in normative ethics like that between consequentialism and deontology, it is quite committal on matters of moral metaphysics, epistemology and psychology. Phillips raises a question about whether such a genealogy offers an alternative to standard analysis, focusing on desirability concepts.
I use the counterfactual genealogy of money, common among economists, to introduce the methodology adopted. This gives us an account of how bartering practices would plausibly lead in a human society, say Erewhon, to the identification of a favored commodity in exchange: perhaps, gold; and how the upshot would give that commodity a role that would likely elicit a distinctive term and concept among Erewhonians: say, ‘aurum’. It then argues that the natural translation of that term in English would be ‘money’, given the role played by aurum in serving for Erewhonians as a medium of exchange and, consequently, as a metric of prices and a means of storing wealth. And so, it raises a question—in effect, rhetorically—as to whether we should not take ‘money’ amongst us to refer to the same sort of thing and to assign money the same sort of role.
Counterfactual genealogy, as I argue in the book, is common in philosophy too; and this, even beyond the work of people explicitly embracing it such as Edward Craig (1990) and Bernard Williams (2002). David Lewis’s (1969) account of conventions is an example. He tells us about how patterns would likely emerge among creatures with our needs and abilities in response to coordination predicaments, and how those responses would likely aggregate in regularities to which everyone conforms because of expecting others to conform. He suggests that such patterns would give those agents natural referents for a term that we might usefully translate as ‘convention’, in virtue of the important coordinating role they would play in the counterfactual world. And then he asks the rhetorical question as to whether we should not take our word ‘convention’ to refer to the same sort of pattern—setting aside some niceties of ordinary usage—and assign to conventions the same sort of role.
Another example of counterfactual genealogy is illustrated, as I have argued elsewhere (Pettit 2019b), by Herbert Hart’s (2012) account of law. He begins with an imagined regime—call it, Normitania—where some collectively beneficial, individually burdensome norms (primary rules) regulate how people treat one another. And then he identifies a set of responses that would likely be made by Normitanians in response to problems of indeterminacy, change and the like, arguing that they would aggregate in the appearance of norms (secondary rules) for supplementing primary rules in the regulation of behavior. Finally, in a move parallel to those charted in the other cases, he suggests that this development would transform the regime into a law-governed one— Lexitania, we might say—and that its dual system of rules, duly internalized, might well serve as the referent for what we ourselves call the law.
My book seeks to provide the same sort of counterfactual genealogy for concepts in the desirability and responsibility families; I shall focus here, as Phillips does, on the first. It begins with a society, Erewhon, where people otherwise like us use language just for reporting on their environment. It then identifies problems they will face in communicating their own attitudes, say of belief and desire, as they will certainly want to do. These problems will arise, in particular, because giving reports on their attitudes would be cheap and unpersuasive; it would allow them, as with any reports, to explain a miscommunication, even one on which others later relied, by their having misread or changed their minds.
But reporting on a state of affairs, p, so the argument goes, would mean expressing or avowing a belief that p and would communicate that attitude in a way that manifestly foreclosed the misleading-mind excuse. Avowing a belief in this way would be more hazardous and expensive than reporting it, of course, but for that very reason would appeal in many cases to speakers: it would increase the credibility of their words. And by parity of argument, elaborated in the book, avowing certain desires would also appeal on that ground.
How are Erewhonian speakers going to be confident enough, however, to avow beliefs and desires? In the one case, according to the argument, by relying on data they take to be robustly persuasive, in the other by relying on desiderata they take to be robustly attractive. Seeking to prove reliable as interlocutors, they will want to be able to live up to their words and will manifestly back themselves to do so when, in virtue of the presence of such data or desiderata, they feel able to avow corresponding attitudes.
But now consider how you in Erewhon will feel about having avowed a desire to take a certain action in some context, when it turns out that, without an independent change of mind, you are inclined rather to do something else. As a creature engaged in widespread practices of avowal and other forms of commitment—the book argues that this is what makes humans into persons—you are bound to think that in general it would be a failure to go with the experienced rather than the avowed desire. You will fail to live up to your words if you are moved by the impulse or incentive supporting the experienced desire rather than by the desideratum or desiderata supporting the avowed.
The claim in the book at this point is that any desideratum that you see as robustly attractive, and fit for the role of supporting the avowal of a corresponding desire—other things being equal, of course—will have a profile in your way of seeing things that entitles us to describe it as a ‘value’, translating whatever word you might use in Erewhon into an English evaluative term. Like a value, after all, it will appeal in a way that can conflict with inclination and will make an appeal such that you must count it a failure not to prove responsive.
As desires may be formed from the perspective of the enduring self alone, or that of a local or more general group, or even that of humanity itself, so desiderata and values may vary accordingly. And suitable values, so the argument continues, are likely to be identified among you and your compatriots—and to count in our language as moral— insofar as they answer to the perspective you will share with them of putting yourself beyond the complaints of others. Why not treat our evaluative and moral terms, then, as serving us in the same way that the terms you use serve you and your compatriots? Why not take them to have referents analogous to the referents of your Erewhonese concepts?
Doing so would make it intelligible why a world of naturalistic desiderata can offer us referents for evaluative, including morally evaluative, terms and concepts, and thereby deliver truth-makers for the truth-conditional statements they allow us to make. And, paralleling the other stories, it would point us to a plausible account of the role of values: specifically, to an account of the role of robustly attractive desiderata in supporting avowals of desire, and serving thereby in a way we take values to serve us.
David Phillips suggests in an initial objection to this genealogy that as ‘an Erewhon story about the acquisition of religious beliefs’—or of hyper-realist beliefs about numbers— would not naturalize them, neither can my story naturalize ethical beliefs. But this objection is off target. Mine is a story about how ethical concepts gain application in a naturalistic world, not a story about the acquisition of ethical beliefs; this would indeed be consistent, like the stories he imagines, with the concepts involved referring to nonnatural properties. And so, as he quotes me saying, it can make ‘the appearance of those concepts in a predicative role…naturalistically explicable’.
But, to turn to a second, more challenging issue he raises, does the genealogy do this explanatory work only in virtue of surreptitiously relying on an independently accessed, naturalistic analysis of value terms? If it does, then it will be redundant.
In pointing us towards the roles associated with money and convention, laws and values, the narratives sketched do indeed support role-based analyses of those concepts, identifying functional conditions that anything must satisfy to count as an instance of the relevant kind. But those analyses do not have to be independently available; they may become accessible only in wake of the genealogy. The analyses produced certainly have to conform in suitable measure to assumptions widely shared among ordinary users of the terms, if they are not to ‘change the subject’ (Pettit 2019a). They may alert us for the first time, however, to such assumptions or may make us aware for the first time of the pattern they reveal in the assumptions; they may give us a new level of understanding.
But even if a genealogically motivated analysis does not presuppose prior access to that analysis, there is a related question as to whether the genealogical route to analysis is any better than standard alternatives. These would generally require us, first, to identify in independent terms the conditions under which a concept to be analyzed applies and, second, to find a property or whatever—a naturalistic property, if the analysis is naturalistic—that it can be taken to predicate. My own view is that in the sorts of cases considered the alternative routes are less easy to access and, if accessed, less satisfying.
Alternative routes will be less easy to access because the conditions relevant to whether a term or concept T applies to an object O is a function in any such case, not just of the conditions that O observably meets, but equally of conditions that the subjects who use the term satisfy without necessarily noticing: practices they engage in, for example, desires or sensations they undergo. These conditions become salient in a genealogy but may not be easy to discern otherwise. That is surely why the conditions highlighted in 4 Hart’s analysis of law or Lewis’s analysis of convention escaped attention before their genealogies. (My own analysis fits comfortably with (Jackson and Pettit 1995)). Alternative routes to analysis, even if accessed, will be less satisfying than the genealogical route, because the sort of analysis they support in the cases considered will treat potentially unnoticed conditions on the subject’s side in the same way as it treats observable conditions on the side of the object. Thus, it will not present the object under the aspect foregrounded for users of the term, giving to background, possibly unnoticed conditions the same role it gives to foreground, observable conditions. The analysis will not find support, then, in a subject’s sense that, yes, ‘that is just what I have in mind when I employ the term’. A genealogy can do much better because it invites us to enter the background perspective of the practices, desires and sensations of users, simulating how things would present to them.
Some purportedly standard analyses clearly fail on this count and if they are useful, it is only because they point us indirectly to a genealogy. Take the analysis: X is red iff it is intrinsically such as to look red to normal observers in normal conditions. As often remarked, that does not tell us what it is we have in mind about an object when we say it is red. If it is helpful, I would say, that is because it points us to a genealogy: a story as to how suitable observers will identify a similarity class of (red) surfaces in virtue of those surfaces having the effect, perhaps unnoticed by them, of eliciting a certain visual sensation (Pettit 2004). I conclude that while genealogy may not be indispensable, it offers a uniquely accessible and satisfying route to analysis in a prominent range of cases. These, as I argue in the book, include cases relevant to our thoughts about what it is for something to be desirable and what it is for someone to be fit to be held responsible.
References
Craig, E. (1990). Knowledge and the State of Nature. Oxford, Oxford University Press.
Hart, H. L. A. (2012). The Concept of Law, 3rd edition. Oxford, Oxford University Press.
Jackson, F. and P. Pettit (1995). “Moral Functionalism and Moral Motivation.” Philosophical Quarterly 45: 20-40; reprinted in F.Jackson, P.Pettit and M.Smith, 2004, Mind, Morality and Explanation, Oxford, Oxford University Press. Lewis, D. (1969). Convention. Cambridge, Mass., Harvard University Press.
Pettit, P. (2004). “Descriptivism, Rigidified and Anchored.” Philosophical Studies 118: 323-338.
Pettit, P. (2019a). Analyzing Concepts and Allocating Referents. Conceptual Engineering and Conceptual Ethics. A. Burgess, H. Cappelen and D. Plunkett. Oxford, Oxford University Press.
Pettit, P. (2019b). “Social Norms and the Internal Point of View.” Oxford Journal of Legal Studies 39.
Williams, B. (2002). Truth and Truthfulness. Princeton, Princeton University Press.
I just want to pick up on Pettit’s response to the example of religious concepts/beliefs.
Here’s Pettit above:
“David Phillips suggests in an initial objection to this genealogy that as ‘an Erewhon story about the acquisition of religious beliefs’—or of hyper-realist beliefs about numbers— would not naturalize them, neither can my story naturalize ethical beliefs. But this objection is off target. Mine is a story about how ethical concepts gain application in a naturalistic world, not a story about the acquisition of ethical beliefs; this would indeed be consistent, like the stories he imagines, with the concepts involved referring to nonnatural properties. And so, as he quotes me saying, it can make ‘the appearance of those concepts in a predicative role…naturalistically explicable’.”
Fair point, but I’m not sure if this goes to the heart of the matter. Take another example, like Joyce’s discussion of “tapu” discourse in the beginning of The Myth of Morality. Plausibly, there’s a good naturalistic story about how tapu concepts gain application in a naturalistic world AND about the acquisition of tapu beliefs. Neither of these empirical conjectures strikes me as a good basis for naturalizing tapu.
This is a big and somewhat general methodological issue, and as Pettit mentions in his response he’s in good company in making such genealogical arguments, so I totally understand if it’d be more helpful to focus on other matters in this forum.
I just hope to draw out Daniel a bit more in his above remarks. Am I understanding your worry correctly, Daniel, if I think of it like this: Seemingly some possible naturalist understandings of how a purportedly normative term gains application in a naturalist world would be vindicatory of the normative pretensions of the term and others would be undermining of such pretensions. Is your worry that Pettit hasn’t said enough yet to persuade us that his naturalistic understanding of how a purportedly normative term gains application in a naturalist world is of the vindicatory kind?
That’s the concern. But I have to confess that I haven’t read the book so there may be plenty more in it that addresses this point!
I don’t want to hijack the thread here – there’s a good deal more to be said about the issues Daniel raises. But I thought it might be useful to give my initial take on Philip’s response: It is that I find the overall line in the response very plausible. But I think counterfactual genealogy then emerges as less of an independent alternative to reductive analysis than some passages in the book might suggest that it is.
If I’m not going too far, I’m inclined to characterize what Philip says in the response as making counterfactual genealogy a discovery procedure for analyses; and, as he says, where the properties referred to are naturalistic, a discovery procedure for naturalistic analyses. I have no problem at all with the claim that genealogy is a good discovery procedure for analyses; the multiple examples Philip appeals to in the book and nicely emphasizes in the response are persuasive. But if this is what counterfactual genealogy does then it’s not the counterfactual genealogy that certifies that the analyses are naturalistic.
I had read some passages in the book as giving the counterfactual genealogy a further role, as an independent source of naturalistic certification. I was particularly struck by some of what Philip says on pages 22 and 23 in introducing the reconstructive alternative to the standard naturalistic strategy. In one bit I quote in the review, on 22, he says
Second…since the appearance of those concepts in a predicative role is naturalistically explicable, the properties they ascribe – and the properties ascribed by our counterpart concepts – must be naturalistic too; if the concepts ascribed non-natural properties, after all, then those properties would presumably have played a role in explaining how the concepts came into use.
That passage in particular made me think that the counterfactual genealogy by itself was supposed to guarantee that the properties ascribed by the concepts whose appearance it explained would be naturalistic. My sketch of a genealogy of religious concepts was intended to suggest that that wouldn’t work. And I take it Daniel has some initial sympathy with me on that.
I see no problem with the more clearly circumscribed metaethical role Philip sketches for counterfactual genealogy in the response. I wonder then whether there are further interesting implications for the overall project. Here is one possible implication: if our interest is simply in generating naturalistic analyses, it might not matter whether the reconstructive genealogies are historically possible. Suppose we discovered, for example, that something blocks actual groups of humans from moving from avowal to coavowal, so there is no way any society could actually have moved along the path Philip sketches. That might not matter at all if the aim of the counterfactual genealogy is simply to generate plausible analyses. But presumably it would matter a great deal to someone engaging in (the commentator in the volume) Michael Tomasello’s kind of project. So Philip’s project and Michael’s project might turn out to be less similar to each other and thus less apt to be in any kind of competition with each other than we might have thought.
Taking up Daniel’s question first, and David Sobel’s query, here’s how I think of the story I tell about the emergence and application (in Erewhon, my counterfactual world) of the concept of desirability. The story explains both how various options should come to seem desirable (in our terminology) to an agent in Erewhon, whether from a prudential, moral, parochial or other perspective. And it explains why they should come to seem desirable in that sense: why, to draw on our ordinary assumptions about desirability, they should look such that despite being less attractive than alternatives, as any desirable option is liable to be, it would be a manifest failure on an agent’s part not to choose a desirable option over such alternatives.
The latter explanation turns on the fact that if an agent avows the desire for an option on the basis of its having certain desiderata—and more generally, other things equal, if the option has desiderata of that avowal-supporting kind—then it will count as a failure in the agent, qua an avower, not to choose that option: not to choose it, even in face of alternatives that are actually be more attractive. This failure will be significant, given that the practice of avowal and commitment, by the story I tell, is central in the lives of Erewhonians, and is even tied up with their constituting persons. (I am not sure if this is too densely summarized to be intelligible on its own!).
On this account, the predicate ‘is desirable’ will naturally come to apply to various options that people in Erewhon—agents relevantly like us—face. It will do so in virtue of having a property that is itself un-mysterious: that of playing a certain role in the activity of avowal. And yet the desirability it predicates will have the ‘normative’ force that shows up in its potentially conflicting with the property of being attractive, where it will count as a failure (for the member of an avowing species) not to realize the option that displays it.
Explained in this way, the property of desirability (to continue to use our English word) is very different from the property of tapu that Richard Joyce discusses. While his (Polynesian) term applies to perfectly ordinary things, as my (Erewhonian) term applies to ordinary options, there is a big difference between the accounts we give. The cue to whether something counts as tapu, as Richard explains, is not empirically available, consisting in what he describes as its ‘metaphysical uncleanliness’. The due to whether something counts as desirable for Erewhonians consists in whether and how far it bears properties of the kind they rely on in making avowals of desire.
I hope this helps to explain why I think that the sorts of stories (religious and mathematical) introduced by David Phillips in his review of the book are not relevantly like the sort of genealogy that I offer. Richard Joyce’s history of the use of ‘tapu’ is of a kind with those stories and I think that the same response may apply in all such cases. Running out of time now: I hope to respond to David’s comment on the role of genealogy relative to analysis in a later comment.
Responding now to David Phillips’s contribution above, in which he argues that genealogy on my account is (just) a discovery procedure for analysis. I am inclined to agree but think that the competitors count as discovery procedures as well.
Discovery procedures are usually contrasted with confirmation procedures. Now I think that no matter how a candidate for the analysis of a concept is discovered, it is confirmed or defended as successful on two sorts of grounds: one, that it answers to enough of our shared assumptions (in common or, if relevant, specialized) usage not to ‘change the subject’; and two, that it allows us to allocate a referent to the concept that is significant in a way that answers to our ordinary sense of its significance; criteria of significance will vary, of course, across analyses in different areas: say analyses of causation or free will or respect. (I discuss this in a paper referenced in my first contribution above: Analyzing Concepts and Allocating Referents).
The method of genealogy represents a way of generating a candidate analysis as the examples of money, convention and law were meant to demonstrate. And, as I suggest, it is a particularly promising and satisfying approach with concepts that we master on the (background, perhaps hidden) basis of being party to certain practices or subject to certain subjective responses (sensations, desires).
But what are the alternative ways of generating an analysis? The most prominent, which can stand in as an exemplar, is the method of cases. Assuming a simple non-holistic concept, this would require us to survey a variety of possible scenarios and to identify in independent terms (all and only) those under which, intuitively, we think the term or concept would apply, It would then have us sort out the conditions in those scenarios that are relevant to the applicability of the term, and select the appropriate set (and the appropriate analysis) to put forward as a candidate. That candidate, like one that a genealogy might suggest, would then have to be tested for how far it succeeds in not ‘changing the subject’, while delivering a suitably significant referent.