NOTE: This essay is a response to an August 4, 2011 post on The New Atlantis’s Futurisms blog.

January 23, 2012

I want to begin by thanking Ari Schulman for the careful and insightful reading he’s given my paper, and for raising some important questions about it. I’d also like to thank The New Atlantis for giving me space on their blog to discuss it.

I described my talk as a “preliminary dialectical exploration” of a complex topic, by which I meant that it was an initial attempt to get clear on the issues rather than to defend a thesis or come to a conclusion. I spent most of the paper lining up arguments in an attempt to give them their due, and identifying pertinent facts that would have to play a role in any finished account of the topic. I ended the talk with a mere gesture at a verdict, on which Schulman properly presses me for elaboration. Though Schulman and I agree on many things, I’m going to devote most of what I say here to our disagreements. One of them concerns a basically moral question: how strong is the connection between appearance and character? The other is a more theoretical or epistemological question: assuming that there is such a connection, what form should it take?

The moral question

As Schulman notes, the question that motivates my talk is, “Does focal visual perception ever disclose evidence relevant to judgments about moral character, where the relevant aspect of moral character is under the agent’s control?” Put more simply, the question is: can we ever learn anything about a person’s moral character by looking at him? In my view, the default answer to this question—established by the projectivist, mismatch, and danger arguments—is “no.” But there are, I admit, possible exceptions to that default rule. What I insist on is that the exceptions preserve the default character of the rule by satisfying a very high burden of proof.

This statement of my thesis underscores my first, somewhat subtle, disagreement with Schulman. The difference, I think, might be described as a difference in the degree of suspicion that we have about the appearance-character connection. He describes himself as “hesitant” about it. I would describe myself as positively suspicious of it. If we could grade attitudes on a scale of 1 to 10, with “1” denoting outright complacency, and “10” denoting downright paranoia, I’d give myself an “8” and give Schulman a “6.”

That’s a very abstract way of putting things, so it might help to consider cases. Take for instance what I called the “concomitants argument,” which Schulman thinks that I underplayed a bit. He says:

First, the “concomitants” argument deserves more emphasis. It’s not just that happiness might indicate a life well lived, which can be a sign of good character. People are very good at reading facial expressions — especially within their own cultures, but many expressions have been found by anthropologists to be universally recognizable. And there are greater depths to facial expressions that strongly indicate personality, and so character. For example, there is a subtle but apparent difference between how happiness looks on the face of a person when it comes from kindness, charity, and good humor as opposed to when it comes from smugness, greed, and pridefulness.

I disagree with just about all of that. I tried to present the concomitants argument in a sympathetic way in my paper, but I think it makes two problematic assumptions. For one, it presupposes that moral character looks a certain way. For another, it presupposes that we can pick out the way that moral character looks from the dense network of chance factors that influence physical appearance. I doubt the truth of both claims.

To challenge the first assumption, I’d suggest doing some sustained people watching in public places, as I’ve done for the year or so since I first conceived this project. What one discovers, I think, is how physically non-descript people turn out to be. The vast majority of people are neither ugly nor beautiful. They look neither virtuous nor vicious. In fact, they don’t look like anything in particular. If we assume that they all have moral characters of one sort or another, the problem is that their physical appearance tells us very little about it. It would be an obvious mistake to infer that since most people are physically non-descript, they are morally mediocre. They could be better than that or worse. One can’t tell by looking.

But suppose that there were some way that good and bad character looked. I think the obstacles to identifying it even in fairly intimate contexts are almost insurmountable. Schulman says that “people are very good at reading facial expressions.” I find that very dubious. To see why, consider the distinction between anxiety and perplexity. They’re obviously very different emotions. Anxiety is a negative; perplexity is not. Chronic anxiety is incompatible with happiness; perplexity is not. And yet anxiety and perplexity can look very similar to an external observer—even to an intimate observer. How do you know that the look on the face of the person before you is anxiety rather than perplexity or vice versa? Not, I think, by looking. You have to ask. But in that case, it’s the person’s answer that does the real work, not his physical appearance. If this phenomenon generalizes, as I think it does, it will often be difficult (even impossible) to differentiate similar-looking but normatively distinct states of mind and character by visual inspection. And here I’m setting aside the deeper question of what the expressions of anxiety and perplexity really tell us about someone’s moral character.

Schulman says that there is a subtle but “apparent” difference between how happiness looks on the face of a person of virtuous character versus how it looks on the face of a vicious person. I doubt that, too. There are certainly stereotypes about the difference in appearance between virtuous happiness and vicious glee, stereotypes that artists and moralists have exploited for centuries. But do these images or stereotypes reflect a real difference between virtue and vice? I’m very skeptical.

To see why, consider the following three videos. Watch them with the sound off (covering up any subtitles), and try your best to set aside any prior knowledge you have of the protagonists. The first is from Leni Riefenstal’s “Triumph of the Will,” depicting a speech by Adolf Hitler (1934). The second is a music video by the heavy metal group Judas Priest (1984). The third is Howard Dean’s notorious “scream speech” (2004).

All three videos feature histrionic, gesticulating men of roughly the same age engaged in a dramatic and impassioned performance for an eager and adoring audience. As far as sheer appearance is concerned, they look rather similar. But whatever one thinks of Judas Priest or Howard Dean (I’m a fan of the former, not the latter), there is no moral comparison to be made between either of them and Adolph Hitler. We tend to think that Hitler has a specifically “crazed look” that justifies our moral judgment of him, but I think that puts the cart before the horse. Hitler “looks crazed” because we know, independently of how he looks, that he was a murderous tyrant with psychopathic beliefs. In such cases, I think we falsely project our subsequent knowledge of his evil onto an appearance that is itself neutral.

I would be interested in hearing more about the anthropological literature to which Schulman alludes, but my experience with the psychological literature suggests that there is a lot of very questionable social scientific research out there on this and related topics, which we would do well to scrutinize with a fine-toothed comb. So without shutting the door on what the anthropologists have to say, I confess to a bit of a priori skepticism. I’ve read too much nonsense by social scientists on this topic to be very optimistic about their findings.

In my view, the most plausible of the arguments in favor of the appearance-character connection is the argument from social appropriateness, which brings us to complex and difficult concepts like “comportment” (as well as “demeanor,” “countenance,” “modesty,” “tastefulness,” and the like). Schulman is right to say that comportment involves an amalgam of appearance and behavior, and that behavior is relevant to judging moral character. So at some level, we agree that comportment is a legitimate exception to the default rule I described earlier.

But even here, I’m skeptical. My view is not that we should have no standards of socially appropriate appearance, but that we should only insist on such standards (and hold others to them) when we’re sure why we have them. The unfortunate fact is that in our culture, people’s insistence on conformity with standards of social appropriateness typically outruns any rational justification that anyone can offer for the standards themselves.

Take for instance the job market and professional life. An enormous premium is put in such contexts on what people cavalierly describe as “professional appearance.” Job interviewers standardly assume that a “first appearance” makes or breaks a job candidate, and managers often have (and are encouraged to have) extremely rigid and arbitrary standards about what “professional appearance” requires in the way of on-the-job dress and presentation. But what does the term “professional appearance” really mean? The truth is that it means almost nothing at all. Its meaning is loose enough to give those who deploy it maximal discretion to enforce it as they please, but rigid enough to justify arbitrary demands such as that women must never wear pants in a professional setting, that men must never wear square-toed shoes, that visible tattoos are a no-no, and that suits and ties are a must. Such demands are in my view plain nonsense, and justify push-back from the rest of us. (If you think I’m exaggerating, try some of these links on for size.)

On the other hand, I agree that police officers (and many others) ought to wear uniforms, mostly so that we can pick them out from non-police-officers. In fact, I myself would probably be inclined at first glance to regard a police officer with an unkempt uniform—dirty, wrinkled, untucked—as slovenly and unreliable. So a uniform is not merely a functional way of distinguishing one set of people from another. It’s legitimately intended as a proxy indication of certain traits of character. But the connection between appearance and character even in this context is very attenuated. A police officer could easily look slovenly and be reliable, or look neat but lack reliability. The same is true of other professions where uniforms are worn. And I think police departments often go overboard in their insistence on conformity. (There are also difficult issues in criminal-law contexts about the role of “demeanor” in the courtroom, about the use of “suspicious appearance” as evidence of probable cause, and about racial profiling.)  

I don’t think reflection on the military clarifies these issues—at least not for those of us outside of the institution. Certainly, as a general proposition, the military’s emphasis on discipline in manner of appearance has something to recommend it. Military personnel have to be disciplined, and part of being disciplined is looking the part. The question is whether contemporary militaries take this emphasis to an unreasonable extreme. It would probably take a sociologist of military life to answer that question—ideally a sociologist who’d served in the military him- or herself. But since few of us have the relevant personal experience of military life to address such issues, my hunch is that any appeal to military culture will provoke as many questions as it resolves.

All of this leaves us with an ambivalent, perhaps anti-climactic answer to our original question. “How strong is the connection between appearance and character?” Not very.

The epistemological question

My second disagreement with Schulman is more philosophical, and concerns the so-called theory-dependency of perception. He says:

Visual perception always depends on theories — and might this not include theories about other people? Put simply: doesn’t the way a person appears change as you get to know him? At a basic level, when you come to admire or love someone, they become more pleasing in appearance, and when you come to dislike or hate someone, they become displeasing. (Deeper contours of character, one might think, could also be revealed in appearance as you get to know someone.)

I’m willing to agree, at least tentatively, with the thought in parentheses. Perhaps we can in some limited (and intimate) contexts come to observe how a person’s character is revealed in his appearance. I strongly disagree, however, that visual perception depends on theories. In fact, I see the two claims as inconsistent with each other. In my view, we can only see how a person’s character is revealed in appearance insofar as we look at him independently of our prior theoretical or evaluative commitments.

The issue of theory-dependency is easily misunderstood, I think, because there are two incompatible ways of understanding what it is for perception to “depend” on a theory. In one (innocuous) sense, theory draws our attention to facts about how things appear, but allows us to perceive those facts directly, without the intermediation of theory itself. In another (highly problematic) sense, theory literally determines what we see in perception, so that we never see facts except “through” our theories. Dependency in the first sense is perfectly legitimate (even salutary), but in the second sense seems to me to lead to a form of Lysenkoism that is incompatible with objective knowledge.

This latter Lysenkoist idea is to my mind an unfortunate inheritance of Thomas Kuhn’s famous book, The Structure of Scientific Revolutions. In that book, Kuhn famously defended the so-called “theory-laden” view of perception, going so far as to suggest that if Aristotle and Galileo were both to have looked at the same physical objects, they would (having different theories of physics) have literally seen different things (pp. 112-14, 120-21). Presumably a shaman with a magical conception of the world would have seen yet a third thing. Alasdair MacIntyre has been heavily influenced by Kuhn, but Kuhn’s account strikes me as a disastrously confused and wrongheaded one, whether applied to scientific or moral knowledge. On my view, Aristotle and Galileo would, on being confronted with the same physical objects, have seen exactly the same thing, regardless of the theoretical differences between them. For there was ultimately only one thing to see: reality. They may have looked at different aspects of that reality, and drawn different inferences from what they saw. But they saw the same thing. Its appearance would have been invariant across changing theoretical perspectives.

A similar point applies to perception of people. Schulman asks, “Doesn’t the way a person appears change as you get to know him?” My answer is, “no.” The way a person appears only changes as his appearance itself changes. Your getting to know him may reveal facts that you hadn’t seen before, but the facts themselves were there whether you saw them or not. Indeed, you may well have seen them but not brought them fully into conscious awareness. When you do, what changes is you the perceiver, not the person observed. It’s no different from hearing something “new” in a symphony you’ve heard a dozen times before. When you first hear what you hadn’t heard before, the symphony may seem to sound different, but the fact is, it actually sounds the same. Your sophistication as a listener may have changed, but it would be a mistake to attribute your new-found sophistication to a change in the symphony. The symphony is the object; you’re the subject. The first thing is invariant; the second is what changes.

If a person’s character is revealed in his appearance, what you see when you look at him are the theory-independent facts that justify your belief that he has a character of a certain sort. Your theory about him depends on what you see, not the other way around. If visual appearance really “depended on theories,” what you would see when you looked at someone would not be objective facts about the person’s appearance but facts constituted by your theories. That error is what I referred to as “projection” (the projection of features of the subject onto the object of perception), and as I see it, it’s the root of almost every error people make about perception. (I’ve been highly influenced on this subject by the work of philosopher David Kelley, whose 1986 book The Evidence of the Senses I highly recommend.)

For reasons like the foregoing, I don’t think that anything is gained by describing moral phenomena in quasi-aesthetic terms like “attractiveness” (or “disgust”). In my paper I insisted that I was using the term “appearance” to mean literal physical appearance as perceived by the eye. Understood in that sense, “attractiveness” always refers to physical attractiveness, and the concept of “moral attractiveness” ends up being a misnomer. On my view, people can be morally good but ugly, or beautiful but morally corrupt. The concept of “moral attractiveness” seems to me a potentially confusing way of saying that those of us who prize morality tend to admire morally good people and seek out their company.

Schulman goes out of his way to insist that the attraction in “moral attraction” is not (or need not be) physical, but I don’t think that that well-intentioned gambit can succeed. The fact is, it’s nearly impossible to decouple the concept of “attraction” from its specifically physical applications. To find someone “attractive” is not, except metaphorically, to express an abstract moral endorsement of his non-physical traits. It’s to appraise his specifically physical traits. So as I see it, there’s no way to tear the concept away from its original usage in order to use it some other way. I doubt that we can escape the physical connotations of the language of attraction, and don’t see the point in trying.

I think it’s better to admit that there are just two irreducibly different kinds of judgments here. There are aesthetic judgments about physical beauty, and moral judgments about character. Since the two sorts of judgments ultimately have very little to do with one another, there is no reason to insist that good people are in some non-physical sense beautiful, or that bad people are somehow non-physically ugly. Better to invest the time and energy in grasping more firmly than we currently do that evil people can be genuinely beautiful, and that good people can be genuinely ugly. I should note, however, that Schulman is in good philosophical company here. The view he defends goes by the name “moral aestheticism” in the philosophical literature, and has some able defenders, among them Colin McGinn of the University of Miami. (See his Ethics, Evil, and Fiction.)

I can’t resist arguing a bit with Schulman about purely aesthetic issues. He says:

It is not coincidental that some of the best portrait and figure painters choose as their subjects not the most conventionally beautiful human subjects, but ones that might be considered plain or unattractive. The Helga paintings of Andrew Wyeth are a particularly famous example.

It seems churlish to disagree with this, but I do. As it happens, I think Schulman’s claim tends to confirm my view that judgments about beauty are more subjective than we realize. For one thing, I think he’s overgeneralizing a bit. Many of the best portrait and figure painters have chosen to paint subjects who were attractive, precisely because they were attractive (think of John Singer Sargent’s Lady Agnew of Lochnaw or Thomas Eakins’s portrait of Henry Ossawa Tanner). In any case, I suspect that there are some disagreements here about beauty itself: for instance, I don’t find Andrew Wyeth’s Helga unattractive in the least. And I’ve perennially been puzzled by the concept of “plainness”: I not only don’t know how to pick out “plain” persons from “non-plain” ones, I’ve never understood why “plain” persons are thought unattractive. Ironically, the psychological literature tends to equate “plainness” (a.k.a. “averageness”) with beauty. And consider the fact that Charlotte Gainsbourg, Ruth Wilson, and Mia Wasikowska have all been cast in the role of Charlotte Bronte’s Jane Eyre. Since Jane Eyre is supposed to be “plain,” we might infer that all three actresses were cast because they are themselves “plain.” And in some sense, I suppose that they are. And yet, I personally would also be inclined to regard all three as beautiful (without making the further inference that any of the three has a moral character that resembles that of the fictional Jane Eyre). So there seems to be more to the concept of “plainness” and indeed beauty than meets the eye.

Having piled up some disagreements, I’d like to end with an agreement. I think Schulman draws the right conclusions from my paper about transhumanism (a topic I didn’t explicitly discuss). The concept of “transhumanism” is a vague and ill-defined one, so I don’t mean to be rejecting it tout court. But I think Schulman is right to single out some of the problematic features of transhumanists’ advocacy of so-called “morphological freedom”—notably their apparent acquiescence in what I called the cult of beauty, and their failure to recognize the costs of such acquiescence. As Schulman says (and as his examples well illustrate), there is a point at which the desire to “enhance” one’s appearance crosses the line from legitimate self-expression to culpable self-contempt. Contrary to the claims of the more strident transhumanists, loose rhetoric about freedom cannot abolish the imperative to draw this line in the right place. But it needs to be drawn.

We may be free, whether physically or legally, to alter our appearance in any way we wish, but we are not free to describe those alterations as “enhancements” unless we can articulate what counts as an enhancement. An enhancement is a change for the objectively better, and presupposes a standard that distinguishes the better from the worse. It also presupposes that whatever has been enhanced has been improved from an objectively defective state. What seems to have gone unnoticed is that a good deal of the talk of enhancement of appearance in transhumanist circles presupposes attributions of objective ugliness, or at least of objectively defective appearance. I think it’s worth asking those who use this language either to justify their claims or to stop using the language of enhancement altogether. They can’t have it both ways. Freedom is a great thing, but it cannot distinguish genuine enhancements from mere alterations.

We need to ask why morphological freedom is valuable by an objective standard of well-being. In cases where morphological alteration is required for clear reasons of physical or mental health, it certainly has a justification. But when it involves risky and expensive medical procedures performed merely for “self-expression,” we’re entitled to be more suspicious of the underlying agenda, and more scrutinizing of the risks and costs involved. When self-expression begins to abet outright racism, as Schulman’s examples suggest, the questions become yet more urgent.

We may not have all the answers, but we should insist, in the face of some strident claims to the contrary, on the right to pose the relevant questions. My paper was a preliminary attempt to that end, and I’m gratified to learn that it dovetails with others’ attempts to similar ends.

Irfan Khawaja
Department of Philosophy, Felician College
Co-Editor, Reason Papers: A Journal of Interdisciplinary Normative Studies