Daniel Russell’s Practical Intelligence and the Virtues is one of the most philosophically sophisticated books I have ever read. The most impressive section lays out a framework for how to think about moral ideals, how they provide the top end of a scale and give us a sense when “virtuous enough” is what he calls “virtuous tout court“; that is, what positions on the scale that fall short of perfect virtue nevertheless are virtuous, without qualification.
Satis Concepts vs Binary Concepts
Russell is directly responding to Christine Swanton’s argument that virtues should be thought of as “threshold” concepts. Swanton argues that this is an alternative to the version of virtue ethics that relies on “what the virtuous individual would do” in a given situation, a formulation made especially widespread by Rosalind Hursthouse. Russell likes the idea of “thresholds”, but thinks it is not sufficiently precise. Moreover, he does not think it is at odds with Hursthouse’s formulation, but rather that the two are necessary for one another.
In the place of “thresholds” Russell suggests “satis concepts”—“satis” being Latin for “enough”.
In this respect, satis concepts like painful and bald are unlike, say, whole or perfect, or prime, positive, and even in the case of numbers, which we might say are ‘binary’ concepts: in the case of the latter, since there are no degrees of F-ness among F things, things are either ‘absolutely’ F or not F at all, and so it makes no sense to talk here of something’s being ‘F enough’. By contrast, satis concepts are such that there are degrees of F-ness among F things, and so since something need not be ‘absolutely’ F to be F, something can be F by being ‘F enough’.
The fact that most or all actually existing Fs in the world will be cases of “enough” is the key here. And the implication of this is that there must be some scale in which you can be more or less F. In the case of binary concepts, there is no scale and no “enough”; a prime number is either prime or it isn’t, there is no “more or less prime”, and “prime enough” is simply incoherent.
Boundary vs Vague Concepts
A further distinction that Russell introduces is between those things that have a sharp boundary between being F or not-F, and those things where no such sharp boundary exists but we can still speak meaningfully of some things being F or not-F.
‘Boundary’ satis concepts are such that while something can be F by being F enough, there is also a boundary dividing F things from not-F things. Notice that since boundaries establish ‘set-theoretically describable divisions’, boundaries by definition are sharp. For instance, we might say that painful is a boundary concept, since anything above a certain boundary of pressure on the skin, say, will count as painful.
However, though there are boundary satis concepts, most boundary concepts are binary, like the prime numbers example above.
Boundary concepts, then, are such that there is a sharp set-theoretical boundary between F and not-F things, whether or not there are degrees of F-ness among F things (i.e. whether or not F is binary). Only where there are degrees of F-ness among F things is a boundary concept also a satis concept.
Russell here points out that the problem with Swanton’s use of the word “threshold” is that it makes it sound like virtue is a boundary satis concept, though he’s quite sure that isn’t what she means. So he introduces the next term in his taxonomy:
Since virtue is a satis concept, but lacks a sharp boundary, it would seem to be a vague satis concept. A classic description of vague concepts holds that a vague concept F is such that there will be ‘borderline cases’ of F, that is, cases in which no method of making F more precise could settle in a privileged way whether the thing is F or not. Vagueness thus arises because of the concept itself, not because we happen to lack a method that would settle these cases. This account of vagueness does not go quite far enough, though, since a concept with sharp boundaries between F things, not-F things, and borderline cases is not a vague concept, despite having borderline cases; so we should say instead that a concept is vague if it lacks such boundaries
Russell offers “bald” and “tall” as examples; there are borderline cases where the vagueness arises from the concepts of baldness or tallness themselves, which cannot be completely settled except arbitrarily (and Russell argues that this does not truly settle it).
Model Concepts vs Central Cases
What makes either boundary or vague concepts also satis concepts is the fact that they exist on a continuous scale in which you can be more or less F, and because you can be F tout court without being “as F as possible.” In some cases, Russell points out, it isn’t even coherent to speak of being “as F as possible.” So while it is coherent to speak of someone being “as bald as possible” but consider cases where one is “bald” simply by being “bald enough”, for tallness, for instance, it is not coherent to contrast against being “as tall as possible.”
When considering whether someone counts as tall, and even if someone who falls short of complete baldness is nevertheless bald, we usually consult central, representative cases of tallness and baldness. In some cases, however, central cases may not be enough. For instance, a multi-dimensional vague concept like personhood is hard to pin down with merely central cases. Russell references fetuses as well as patients in “severe” comas as instances where figuring out what constitutes personhood would matter to us. In order to make use of the central concepts of personhood, we need to understand “in what ways must a person resemble the ‘central cases’?” According to Russell, what we require is a model.
When we try to say what personhood really is, we construct a theoretical model of what we take to be the essential features of personhood, in some kind of reflective equilibrium, and realized to the fullest degree, since the model must illuminate the central cases, not just join their ranks. This model, we should note, is an ideal, and therefore not merely a central case: you or I could stand as a central case of personhood, but not as a model of personhood, since particular persons always have shortcomings in some dimension or other of personhood, a shortcoming that the model is to reveal as a shortcoming.
Vague satis concepts that require such models in order to be made coherent are what Russell calls model concepts (are you following the nesting venn diagrams so far?)
One aspect of model concepts is that they are not purely interest-relative.
Some cases of ‘F enough’ require no such model because ‘F enough’ is purely interest-relative, and sometimes those interests permit—and even require—that we eliminate vagueness by simply stipulating a boundary. For instance, consider the children’s game of racing to fill a cup by carrying water across the room in a spoon. Since the object of the game is to be the first to fill one’s cup, players need to know with a fair bit of precision when a cup counts as ‘full’, and an easy way to solve this problem is to draw a line on the cup and stipulate, ‘Full is at that line’. Notice that a cup that is ‘full enough’ in this stipulated sense is not necessarily full tout court; all we can say is that it is ‘full*’, full for this rather special purpose.
What makes ‘F enough’ purely interest-relative? When we ask whether a fetus or a comatose patient is a ‘person enough’, we take a keen interest in the answer, but what we want to know is whether the fetus or patient is a person tout court, not a ‘person*’, or a ‘person’ relative to just any old interest or purpose we may have. Here our interest is in finding out whether the patient is really a person, and so here ‘person enough’ is not purely interest-relative. Depending on how the answer comes out, we may even have to revise the interests that initially prompted the question.
Russell further stipulates that model concepts must deal with a substantive concept, not just our use of language. That is, it must refer to F in the de re rather than purely de dicto sense. He contrasts the concept of personhood with that of baldness. In the latter case when it comes to borderline cases we really are just talking about the de dicto use of the word “baldness”, whereas in the case of personhood we care about what a person really is, regardless of what words you want to use.
Therefore, a concept F is a model concept just in case being F enough entails being F tout court, where ‘F tout court’ is understood neither in a purely interest-relative sense nor in a de dicto sense, but in terms of what it is to be F, de re. In the case of model concepts, the possibility of error about ‘F enough’ is more than the possibility of being out of step with how competent speakers talk about what is F enough. It is instead the possibility that one may even be in step with everyone else about the central cases, but nonetheless be mistaken in thinking that those cases really are or are not F.
Russell’s discussion of how models calibrate our understanding of the subset of vague satis concepts they are necessary for reminds me of Deirdre McCloskey’s decades-long quest to get empirical economists to answer the question of “how big is big?” Here is Russell:
Cups are full, power plants are safe, and men are bald by degrees that can be ordinally ranked, and in some cases we can even say by just how much one degree differs from another. How such scales are calibrated, then, depends on our interests, or perhaps on common usage of the related terms. By contrast, model concepts like rational and person also yield a calibrated scale, but here the scale requires a ‘standardized’ calibration: what it is to count as rational at all, and to what extent a given agent is rational, is to be determined (where it can be determined) by a scale that is calibrated by a model of ideal rationality that both sets the top end of the scale and gives meaning to the idea that a particular agent occupies a certain level on that scale. (Likewise for personhood.) Understanding what it is to be F tout court in a de re sense, then, calls for a theoretical model of ‘really F’.
It should be clear by now that Russell believes that any theory of virtue ethics must be build around model concepts. And moreover, that it is the nature of such concepts that they are both “thresholds” in Swanton’s sense and require ideals in Hursthouse’s sense.
Therefore, it is a mistake to suppose that the idea that one need only be ‘virtuous enough’ to be virtuous is an alternative to thinking of the virtues in terms of ideal models. On the contrary, thinking of virtue in terms of ideals is required on account of the very sort of satis concept that virtue is.
Ideals vs Aspirations
As if this account wasn’t intricate enough, Russell goes on to address the question of just how hard we must strive to approach our ideals, once we specify them. Many of the critiques of Hursthouse’s “virtuous person” standard boil down to the idea that it is unreasonably high; that is precisely Swanton’s motivation in attempting to provide a “threshold” alternative.
Russell meets this critique by turning to Donald Davidson’s model of rationality, which includes, among other things, the notion of “all-things-considered” judgments.
What does committing to meeting that requirement come to? Obviously, the ideal of ‘the rational person’—the person with ‘overall rational unity’—is ‘an ideal of which we are bound to fall short’, but it is my view that committing to this ideal is not the same as committing to becoming as like it as possible. Rather, to commit to the ideal of overall rational unity is to accept the principles of such rationality as principles one recognizes as one’s own principles, principles to which one can be held, by which one can be criticized, and in relation to which one must frame one’s response to such criticism—indeed, the very giving of such criticism presupposes the agent’s ability to respond in just this way.
By accepting the ideal, one has made the ‘rule’ or principle a principle to which one sees oneself as bound. But committing to making all-things-considered judgments is not the same as committing to the (rather queer) life-project of becoming the best maker of all-things-considered judgments there can be. That project, like every other, consumes resources and opportunities, and can no more be assumed to be a rational one than any other project can. That is a fact about practical rationality: when it comes to making all-things-considered judgments, at some point it is reasonable to stop considering, choose, and hope that the choice is one we can live with, or perhaps grow into. Indeed, trying to become persons who do consider all things before acting is something that we have all-things-considered reasons not to do.
In short, the question of how hard we must aspire to approximate the ideal is a separate question from merely specifying the top-of-the-value-scale model, and one that ought to be addressed in that specification itself as the model of “all-things-considered” rationality does.
Uses For This Framework
It should be obvious that I am highly impressed by this framework. I can see it already implicitly at work in places where it and its implications haven’t been fully fleshed out. Deirdre McCloskey’s Bourgeois Virtues can be seen as providing a “model concept” to both hold up as an ideal and to describe the actual virtuousness of modern middle class life. The task of reviving “a serious ethical conversation about middle-class life, the life of towns, the forum and the agora” involves in no small part specifying and defending such a model. This is nowhere more clear than in the section of the book commenting on Jane Austin:
Jane talks of virtues up and down, back and forth, on all sides. But not from a theory. Her English suspicion of theory allows for ethical complexity. Complexity, but not coherence: grace, understanding, manners, amiability, good habits; each gives its set speech, and then withdraws from the scene.
It is clear that McCloskey believes it possible to have a model that allows for both complexity and coherence—and so, indeed, does Russell.
I myself also implicitly utilized model concepts when deciding what sort of book I wanted to write.
There are two books I have in my head, with two very different implied authors.
One book is a long treatise on virtue ethics in the market, written by an implied author who is capable of addressing the main critiques that would come out of the philosophical community. This implied author is very well read in the conversations of philosophy in and out of ethics, and is well equipped to engage in that conversation. The problem is that Deirdre McCloskey has basically written this book already, and even though my version of virtue ethics may differ slightly from hers, I don’t think that this is the best contribution I could make right now, nor am I well equipped to make it. It would take me several more years of research before I could comfortably write as this implied author.
The second book has an implied audience of the people who buy books in the business section of a book store (or Amazon). Unpretentious, practical minded people who nevertheless view consulting books as both enjoyable and useful. The implied author is well read and confident in the subject he is writing about, but not an expert or a specialist in philosophy or economics. He works for a living and thus experiences the world of business forty hours a week or more, and has so for his whole career—but not as an executive or entrepreneur or anything more exciting than a salaried employee. His value proposition for his implied audience is that his life is similar to theirs, and he can offer a framework for looking at that life that is very clarifying, but also satisfying. Useful, but also enjoyable to engage with critically or from the inside.
I believe that that is an implied author I am capable of becoming within a reasonable timeframe, and that is more or less the book that I want to write.
In this case, the “implied author” is a model concept, one that I have spent most of this year working to approximate, and will no doubt continue to do so through 2015. However, this model is only useful to me insofar as it can guide me in deciding when enough is enough; when I am qualified enough to write the book and therefore qualified tout court.
Which brings me to my pseudonymous fellow Sweet Talker boatfloating, who eloquently wrote:
You do not need to shoot and kill a 7-year old girl sleeping on her grandmother’s couch yourself to support the tough-on-crime approach. You do not need to kill an innocent Iraq War veteran yourself to support the War on Drugs. You do not need toextort sex yourself to support the continued criminalization of a consensual act. You do not need to torture innocent people and informants yourself to be for the use of torture.
You do need to(or, at least, should) recognize your public policy preferences will have unintended consequences. You do need to(or, at least, should) recognize that these unintended consequences will incur costs. You do need to(or, at least, should) recognize that most of these costs will not be borne by you. You do need to(or, at least, should) be made to understand, fully, these costs. To do otherwise is moral and intellectual cowardice.
This is highly provocative, especially to a crusader against telescopic morality (another model concept) such as myself. So I ask him: what is the ideal here, and when is enough enough? This directly parallels “all-things-considered” rationality; when is a citizen’s policy position “all-things-considered” enough to make one a good citizen tout court?
It’s unfortunate that I cannot boil Russell’s framework down to less than a 3,000 word post, because I think it is very important. And I encourage everyone to read the book and in particular the fourth chapter, which covers this specific subject.
But I don’t think you have to navigate his nesting venn diagram of concepts in order to get the gist of the questions—what is the ideal? When is enough enough? How hard are we obligated to work on approaching the ideal?