In Defense of the Mirror Thesis
Peter Brian Barry
Associate Professor of Philosophy
Saginaw Valley State University
pbbarry@svsu.edu
Published in Philosophical Studies, Vol. 155, Issue 2 (2010), pp. 199-205
1
Introduction
In this journal, Luke Russell defends a sophisticated dispositional account of evil
personhood according to which a person is evil just in case she is strongly and highly
fixedly disposed to perform evil actions in conditions that favour her autonomy (Russell
2009). While I am generally sympathetic with this result, at least some of Russell’s
discussion of evil personhood is problematic. At present, I consider his rejection of the
mirror thesis—roughly, the thesis that evil people are mirror images of the morally best
sort of persons. Having discussed and defended the mirror thesis elsewhere, I count
myself as one of its adherents (Barry 2010, 2009). Initially, I clarify just what the mirror
thesis entails. I then consider Russell’s arguments against it and argue both that his
arguments are not convincing and that he is actually committed to my favored thesis
about evil personhood.
2
The Mirror Thesis
Susan Wolf suggests that a moral saint is as “morally worthy as can be” (Wolf
1982, p. 419)—that is, the morally best sort of person. The mirror thesis suggests that
evil people are relevantly analogous to moral saints—that despite their differences, the
morally worst sort of people are relevantly similar to the morally best sort of people.
Obviously, the analogy will not be perfect: evil people fall short of moral decency while
In Defense of the Mirror Thesis
2
moral saints do not; we expect evil people, but not moral saints, to be regular sources of
harm and suffering; moral saints are praiseworthy in a way that evil people are not, and
so forth. And it would be a mistake to suppose that evil people count as the mirror image
of the moral saint in virtue of just any similarity. Recall that while Kant supposed that
there could be evil human beings, he denied that there could be “devilish” human beings
whose reason is entirely exempt from the influence of the moral law (Kant 1960, pp. 301). Thus, for Kant, even the morally worst of us are not devilish. But it would be odd to
suppose that because both evil people and moral saints are not beyond the influence of
the moral law, the former is the mirror image of the latter.
Further, evil people do not simply “mirror” moral saints but perversely mirror
them, as any number of philosophers sympathetic with the mirror thesis have suggested.
For example, we are told that the evil person is motivated “in the opposite direction” of
the virtuous person and thus counts as his “mirror image” (McNaughton 1988, p. 135),
that the moral saint is “the positive counterpart to the evil person” (Haybron, 2002b, p.
274), that certain constitutive dispositions of evil persons and moral saints are the “exact
reverse of one another” (McGinn 1997, p. 61), that the evil person occupies a space “at
the other end of the moral spectrum” from the moral saint (Garrard 1998, p. 329), and
that the sort of wickedness that marks evil persons “is the exact opposite” of the sort of
moral goodness that characterizes moral saints (Milo 1984, p. 7). So, I submit that the
mirror thesis can initially be captured as follows:
(MT): a person is evil just in case she is a perverse mirror image of the moral
saint
3
There is, of course, more to be said. As it stands, (MT) is too metaphorical to be
illuminating and there are various literal variations of (MT) that are implausible as I have
argued elsewhere (Barry 2009).
However, the following version of (MT) is plausible and arguably goes a long
way to capturing just what it is to be an evil person:
(MT*): a person is evil just in case he is a perverse image of the moral saint such
that he suffers from extremely vicious character traits
(MT*) implies that moral saints are the morally best sort of person in virtue of their
extreme moral virtue, that evil persons are the morally worst sort of person in virtue of
their extreme moral vice, and that their respective characters are constitutive of being
saintly and evil (Barry 2010 and 2009). I take it that (MT) is correct if (MT*) is, such
that if the characters of evil people and moral saints are both marked by extreme aretaic
properties then the mirror thesis is correct.
I shall have more to say about both theses below. But Russell offers some reason
to think that (MT) must be incorrect.
3
Russell on (MT)
Russell’s rejection of (MT) depends upon a crucial asymmetry between moral
saints and evil people, one that allegedly upsets the mirror thesis. On one view of things,
an evil person must be consistently and thoroughly vicious, lacking any good side
(Haybron, 2002a, p. 269). Russell rejects this sort of consistency account of evil
personhood (Russell 2009) and I concur (Barry 2010 and 2009). There are, after all,
plausible putative examples of evil persons who do seem to have some “good side”
4
insofar as they do sometimes seem to possess morally appropriate motivation and affect;
even Hitler could have cared a bit for Eva Braun. And, importantly, we are certainly not
rationally required to revise our overall judgment of Hitler’s character simply because he
had some slight morally redeeming qualities since those qualities are altogether trumped
and overwhelmed by his extreme moral vice (Barry 2010). However, while Russell
allows that evil people need not be morally depraved through and through, he rejects a
corollary thesis about moral saints—that is, he insists that moral saints must be morally
admirable through and through, lacking any moral vice. Here is Russell:
It is true that an extremely virtuous person must be morally admirable in all
respects, and could be counted on to do the right thing in all situations, and a
mirror-image of that person would be bad in all respects and could be counted on
to do the wrong thing in all situations. Even so, we have no reason to believe that
every extremely immoral person must be morally bad in all respects (Russell
2009).
The problem is not simply that evil people and moral saints are different in some respect;
presumably, two non-identical objects will differ in some respect no matter how similar
they otherwise are. The problem is that the aretaic properties possessed by evil people in
virtue of which they count as the morally worst sort of person do not mirror the aretaic
properties possessed by moral saints in virtue of which they count as the morally best sort
of people: the former need not be perfectly vicious while the latter must be perfectly
virtuous. If all this is right, the mirror thesis seems wrong.
5
Worse, there is another problem for proponents of the mirror thesis. It is
worthwhile to make explicit a thesis that Russell adopts when he suggests that “It is true
that an extremely virtuous person must be morally admirable in all respects”:
(MS): a person is a moral saint only if she is morally admirable in all respects and
is not morally bad in any respect
Apparently, the conjunction of (MT) and (MS) yields counter-intuitive results. Russell
argues as follows:
If we assumed that evil persons must be mirror-images of fully virtuous persons,
then the chillingly successful and reliable real-life evildoers would not count as
evil persons simply because they are not bad in every respect. The implausibility
of this result suggests instead that we should reject the mirror thesis (Russell
2009).
Roughly, if (MT) is correct, then, given that moral saints count as the morally best sort of
person partly because they are morally admirable in all respects and suffer from no moral
defects, evil people would have to me morally bad in all respects and not at all admirable.
But, as I concede above, we have good reason to doubt that this is the case given our best
putative examples of real-life evil persons are not necessary morally flawed though and
through. And that is some reason to think that (MT) is false.
Still, even if the conjunction of (MT) and (MS) yields an implausible result, that
result could just as easily be avoided by rejecting (MS) rather than (MT). I submit both
that there are good independent reasons to reject (MS) and that Russell himself is
committed to rejecting it.
6
Note that Russell agrees with Haybron that the concept of the evil person has a
home in ordinary moral discourse such that an analysis of that concept must be fixed, at
least to some degree, by our pre-theoretic intuitions about evil people (Russell 2009).
Two sorts of pre-theoretic intuitions play a role in Russell’s rejection of consistency
accounts of evil personhood: first, that there are some actual evil persons, and second,
that at least some of our particular judgments about putative evil people—namely, that
they are evil—are correct. Consistency accounts of evil personhood fail to accord with
both intuitions insofar as they make it too hard for any actual people to count as evil
persons such that all of our particular affirmative judgments about putative evil people
are false.
However, the concept of the moral saint surely also has a home in ordinary moral
discourse such that an analysis of that concept should also be fixed, at least to some
degree, by our pre-theoretic intuitions. Parity of reasoning suggests that if we have the
intuitions that there are some evil people and that at least some putative evil people really
are evil, we have similar intuitions that there are some moral saints and that at least some
putative moral saints really are saints. The problem with (MS) is that it is inconsistent
with our actual intuitions about moral sainthood. While there might be disagreement
about just who counts as the morally best sort of person, most if not all putative moral
saints suffer from not insignificant moral failings and vices: Martin Luther King was a
philanderer; Mother Teresa confessed that for some period of time she lost her faith and
on occasion declared suspect motives for caring for the hungry; Oskar Schindler’s
motives may not have been entirely pure; Gandhi was estranged from his family and
arguably engaged in exploitive sexual relations with women to test his ascetic resolve;
7
Socrates neglected his family; Jimmy Carter, a deeply pious and religious man, admitted
to having lust in his heart, and so forth (Flanagan, 1991; Hamilton 1999). If (MS) is
correct then none are moral saints. Indeed, (MS)’s demand for moral perfection is
demanding to the point that it seems that there simply are no actual moral saints, contrary
to our intuitions. If consistency accounts of evil personhood should be rejected because
they are inconsistent with pre-theoretic intuitions about evil people, then (MS) should
similarly be rejected because it is inconsistent with pre-theoretic intuitions about moral
saints.
I suspect that Russell endorses such a demanding view of moral sainthood
because he implicitly endorses an account of moral virtue that sets the mark for virtue
rather high. Citing Aristotle, Russell notes that “full virtue is an exacting ideal that
requires an integrated unity of good character traits” (Russell 2009). But there are
stronger and weaker readings of this claim. On its weaker reading, an integrated unity of
good character traits does not require the possession of every moral virtue to significant
degrees and the absence of moral vice entirely; an integrated unity of virtues need not be
an exhaustive one. But on a stronger reading, presumably the one that Aristotle himself
endorses, the possession of any one moral virtue requires possession of all of them and
the absence of any particular moral virtue implies the absence of all the rest. This is, of
course, the “unity of the virtues” thesis. If the unity of the virtues thesis is correct then
(MS) would have to be correct, supposing that moral saints must have full virtue. But the
truth of the unity of the virtues thesis is debatable and even some sympathetic
commentators reject it (Badhwar 1996; Flanagan 1991; Foot 1983; Hurka 2001; Williams
1985). By contrast, the weaker reading is compatible with a failure to possess some
8
significant moral virtues or a failure to possess them to sufficiently significant degrees
and, perhaps, with the possession of modest moral vice. The sort of virtue—at least, the
sort of virtue possessed by creatures like us—captured on this weaker reading only
requires possession of a sufficient number of morally significant virtues possessed to
rather significant degrees. And arguably, this weaker sense of virtue is sufficient for
moral sainthood (Barry 2010; Wolf 1982). So, absent some independent argument for
the stronger version of the unity of the virtues thesis, there seems to be little reason to
adopt, and good reason to reject, the demanding (MS).
I conclude that Russell’s arguments against the mirror thesis fail. But there is
more to be said. Despite his protests, there is a rather strong case to be made that, by
Russell’s own lights, (MT) and (MT*) are true.
4
(MT), Autonomy, and Dispositional Accounts of Evil Personhood
Again, Russell defends an account of evil personhood whereby an evil person just
is strongly and highly fixedly disposed to perform evil actions in conditions that favour
her autonomy. To say that a person is strongly and highly fixedly disposed to perform
evil actions is to say, roughly, that it is sufficiently likely that she is in the sort of
conditions conducive to the performance of evil actions or that she will frequently
perform such actions in those conditions or some such thing (Russell 2009). This is
consistent with supposing that evil people suffer from significant moral vices, as
suggested by (MT*), since vices, like the virtues, are plausibly understood as complicated
multi-track dispositional states, including dispositions that ground a tendency to
frequently engage in grave wrongdoing. The further requirement that an evil person is
9
disposed to frequently engage in grave wrongdoing in conditions that favour her
autonomy is crucial given our intuition that evil people deserve our strongest moral
condemnation (Russell 2009). After all, if someone performs an ostensibly evil action
only because of situational factors that undermine her autonomy, there is arguably a sense
in which she is not blameworthy—or perhaps not fully or sufficiently blameworthy—in
the way that we tend to think that evil people are.
Suppose for the sake of argument that Russell’s dispositional account of evil
personhood is correct. I submit that a parallel of moral sainthood is just as plausible and
for similar reasons; in particular, moral saints are strongly and highly fixedly disposed to
perform the morally best sort of actions in conditions that favour their autonomy. On one
account of things, every action of the moral saint is as morally good as possible such that
the life of the moral saint is dominated by a commitment to improving the welfare of
others or society as a whole (Wolf 1982, pp. 419-20). This suggests that saints must be
strongly and highly fixedly disposed to perform the morally best sort of actions. Further,
if evil people deserve our strongest moral condemnation in virtue of their exceptional
blameworthiness, moral saints presumably deserve our strongest moral approbation in
virtue of their exceptional praiseworthiness. And just like blameworthiness,
praiseworthiness surely depends upon what an agent would do in conditions that favor
her autonomy. In other words, the complicated disposition that Russell suggests
constitutes evil personhood at least appears to perversely mirror the complicated
disposition that I have argued constitutes moral sainthood. And that is at least some
reason to suppose that, by Russell’s own lights, (MT*) and (MT) along with it, are
correct.
10
References
Badhwar, Neera Kapur. 1996. “The Limited Unity of Virtue.” Noûs 30: 306-29.
Barry, Peter Brian. Forthcoming 2010. “Extremity of Vice and the Character of Evil.”
Journal of Philosophical Research.
_______________. 2009. “Moral Saints, Moral Monsters, and the Mirror Thesis.”
American Philosophical Quarterly 46: 163-176.
Flanagan, Owen. 1991. Varieties of Moral Personality: Ethics and Psychological
Realism. Cambridge: Harvard University Press.
Foot, Philippa. 1983. “Moral Reason and Moral Dilemma.” Journal of Philosophy 80:
379-98.
Garrard, Eve. 1998. “The Nature of Evil.” Philosophical Explorations 1: 43-60.
Hamilton, Christopher. 1999. “The Nature of Evil: A Reply to Garrard.” Philosophical
Explorations 2: 122-38.
Haybron, Daniel. 2002a. “Consistency of Character and the Character of Evil.” In
Daniel Haybron (ed.), Earth’s Abominations: Philosophical Studies of Evil. New
York: Rodopi, 63-78.
_______________. 2002b. “Moral Monsters and Saints.” The Monist 85: 260-84.
Hurka, Thomas. 2001. Virtue, Value, and Vice. Oxford: Oxford University Press.
Kant, Immanual. 1960. Religion Within the Limits of Reason Alone. New York: Harper
Torchbooks.
McGinn, Colin. 1997. Evil, Ethics, and Fiction. New York: Oxford University Press.
McNaughton, David. 1988. Moral Vision. Oxford: Blackwell Publishers.
Milo, Ronald. 1984. Immorality. Princeton: Princeton University Press.
Russell, Luke. 2009. “Dispositional Accounts of Evil Personhood.” Philosophical
Studies. Published online: 13 February 2009.
Steiner, Hillel. 2002. “Calibrating Evil.” The Monist 85: 183-93.
Williams, Bernard. 1985. Ethics and the Limits of Philosophy. Cambridge: Harvard
11
University Press.
Wolf, Susan. 1982. “Moral Saints.” The Journal of Philosophy 79: 419-39.