. 6
( 10)


phenomenal badness. Why is the view suggested here, that there is one
kind of phenomenal painfulness which is badness, to be preferred?
Reply: Aversion alone cannot constitute painfulness, since even plea-
sure of some sorts is something to which someone might have an aversion.
But nonetheless there are a variety of things that might properly be called
“pain” in English, some perhaps constituted merely by aversion at least to
a large degree. And there are a variety of phenomenal properties in which
distinct pains do differ. Indeed, there are rubbing feelings or the like that
are characteristic parts of the phenomenal component of certain sorts of
pains. So my claim is limited: that the basic phenomenal component of
the paradigmatic physical pains “ as colors of particular sorts are the basic
phenomenal components of visual experiences “ is the phenomenal bad-
ness I™ve noted. Aversion plus other phenomenal properties cannot fully
capture paradigmatic pains. There is a variety of things that might properly
be called pains, given the various meanings of the term, but my claim is not
primarily a claim about the meaning of “pain”, nor even about its essential
properties, but rather about the nature of a certain phenomenal property
that we in fact experience in paradigmatic instances of physical pain.
Objection Eight: Under the in¬‚uence of certain drugs, we can
experience pain without minding it or ¬nding disvalue in it. So even
paradigmatic physical pain doesn™t have to include disvalue, and it is only
when we mind pain and want to be free of it that we ¬nd comparative
disvalue in the experience. But why then say that disvalue is an additional
element in the experience of having a pain and being bothered by it?
Why not say that all that distinguishes pain we mind from pain we don™t
is the fact of minding or not minding it?
Reply: But it is not the minding of even disliked physical pain
that constitutes its phenomenal component to involve disvalue of the
appropriately objective sort we are discussing. Badness in my sense

needn™t be minded, and indeed that is part of the evidence that it is
objective badness in the relevant sense.
Objection Nine: We cannot understand the anguish of the red
knight without realizing that he is disposed to avoid it, and we cannot
understand it without ourselves being disposed to avoid such a thing.
This provides grounds for the objective vindication of a desire-based
conception of the good.
Reply: We cannot understand bloodthirstiness without realizing that
it involves a positive preference for horrible things. But that does not
provide a direct vindication of the value of such horrible things. What™s
more, we can in fact understand the preference of the red knight without
sharing it, and even understand his painful hedonic tone without caring
about it. As the previous objection noted, we can even feel our own
painful hedonic tone, and hence suitably understand it, without minding
it. Still, it is of such a nature that we ought to care about it.
Objection Ten: There are some philosophical circles in which any
appeal to phenomenal properties is immediately suspect. Some object
that Wittgenstein™s private language argument has shown that there are
no such things as phenomenal properties.39
Reply: But we should not be so easily intimidated. Aristotelian colors,
the most familiar analogs of phenomenal value, are not necessarily private
in any sense. And whatever the notoriously evasive private language
argument is, it is controversial. Qualia, phenomenal properties by another
name, are a central topic in contemporary philosophy of mind, and
hence no established orthodoxy within the philosophical discipline that
centrally concerns them now rules them out of consideration. So it
certainly isn™t appropriate for ethicists to rule them out of consideration
by dropping a name.
Nevertheless, the private language argument does point to an
important practical dif¬culty. If I am right so far, then it is crucial to
develop a more speci¬c characterization of the general sort of normative
properties under consideration. And it is clear that introspection, or at
least observation that is not measurement of ordinary kinds, must play
a crucial role in this. Of course, it might be introspection by more than
one individual, and hence might involve the standard empirical studies of
quality spaces that play an important role even in some parts of chemistry
and psychology. But introspection is quite fallible, and we must proceed
with caution and modesty.

39 One place this objection is made is Korsgaard (1996a: 131“166).


I have claimed that there is a type of phenomenal and hence unconstituted
natural property that is present, though not in all forms of suffering or
even in all cases that we call “pain”, at least in paradigmatic sorts of physical
pain. This is a type of normative property. In particular, it is a kind of
disvalue or badness, though not by any means the only form of disvalue
or badness. It is objective disvalue, which is bad from the point of view
of the universe, in Sidgwick™s inevitable phrase.
We have focused so far primarily on pain and hence phenomenal bad-
ness, but there is obviously another type of phenomenal property present
in some cases of physical pleasure that is likewise a type of normative
property, a form of positive value or goodness. This claim may be slightly
more controversial than my claims about pain. Some hold that only pain
is a sensation, that only pain involves a special kind of sensory experience,
while pleasure does not. But this is wrong. A special sort of sensory expe-
rience also in part constitutes pleasure, at least in a range of central cases.
In at least some cases we call pleasure, which include some of the very
down-to-earth pleasures of eating, scratching, defecating, and sex, there
is an experience of localized pleasantness that is closely analogous to the
experience of localized painfulness characteristic of physical pain. This
is a kind of objective goodness. Lots of things we call pleasures are not
robustly physical and localizable in this way. Some involve no special phe-
nomenal properties at all. Some things we call pains are also not robustly
physical. But there are at least the clear and vivid cases of some physical
pains and pleasures, which involve positive and negative hedonic value.
Classical hedonists conceived pleasures and pains to differ as do
positive and negative value, and also always to be comparable in value
despite being felt by different persons. But they further conceived the
value of pleasures and pain to have an arithmetical or cardinal structure,
to involve quantities that one might add and subtract or even multiply
in familiar ways. Pains were supposed to differ in amount, but in several
different ways that were implicitly reconciled in that cardinal structure.
Bentham held that pains differed in intensity and duration.40 But we
might also reasonably consider spatial differences “ between, say, a pain
in my whole arm and one just in my elbow. And perhaps there is another

40 Bentham (1970: Chapter IV, 38). In this passage, Bentham also mentions the certainty
and propinquity of a pleasure as relevant to its value. But this is because of his psycholog-
ical hedonism coupled with his recognition that the certainty and temporal rapidity of a
punishment or reward are relevant to its motivational effect. See, for instance, pp. 170“171.

sort of phenomenal difference in quantity, between the vividness of
actual pains and the faintness of remembered or anticipated physical
pains. Perhaps those are all instances of one type of quantity.
But it seems to me that the truth about hedonic value is slightly more
complex, and it will take a while to explain it.41
First, there is one dimension of value and disvalue levels, with three
subdivisions: positive value, disvalue, and the null state in between.
Second, both positive value and disvalue come in greater and lesser
Third, each discriminable momentary bit of phenomenal experience
has either a particular degree of value, a particular degree of disvalue, or
a null valence. The null valence is had by many discriminable bits. But
there are also many bits with value or disvalue, which sum to provide
pleasures and pains that are temporally or spatially extended.
Fourth, these degrees of value are “interpersonally comparable”. It
makes sense to say that two persons experience bits with the same level
of value, though of course we may lack conclusive evidence that they do.
A ¬fth and crucial point is more debatable: While a pleasantness can
present itself as better than a lesser pleasantness, it does not present itself,
as the classical utilitarian requires, as twice or three times better. In this
respect, phenomenal value is analogous to phenomenal color. It makes
sense to say that something is bluer than something else, but not that it is
two or three times so without reference to some arbitrary scale. Intrinsic
phenomenal value comes, as economists and mathematicians say, ordinally.
Let me explain. I claim that phenomenal value is ordinal. Consider
four levels of phenomenal value. Call them A, B, C, and D. Sometimes
it will make sense to say that the difference in intrinsic value between
A and D is greater than that between B and C. For instance, it might be
that A is better than B, which is better than C, which is better than D. B
and C might be in between. But when one difference is not enclosed in
another in this manner, it is senseless to say that one is a greater difference
than the other. There is nothing in the phenomenal facts to ground that
comparison. Intrinsic phenomenal value comes ordinally. It makes sense
to speak of inter- and intrapersonal comparisons of value levels, hence to
ascribe numbers for ease of reference in this way: to the indifferent level
of phenomenal value a 0, to each level of positive value some arbitrary
positive rational number with greater numbers assigned to better levels,

41 This discussion is an improvement over the less adequate treatment in Mendola (1997:
Chapter 6).

and to each level of disvalue some arbitrary negative rational with lesser
numbers assigned to worse levels. It will not make sense to take these
representatives in other than an ordinal sense, as bearing information
about the relative size of value differences. Any assignment of rationals
meeting the stipulations noted will be as good as any other.
Let me stress the two crucial and controversial features of this initial
characterization of hedonic value, both of which will matter in the next
chapter. First, I am claiming that there are objective facts not only about
intrapersonal but also about interpersonal comparisons of phenomenal
value levels. The nature of phenomenal value is very simple. The value
I experience presents itself not as value for me but as value simpliciter.
And there is an objective fact whether or not a person experiences
phenomenal value of level X, and whether or not another experiences
value of the same level, just as there is an objective fact whether those two
persons experience the same shade of phenomenal blue. Of course it may
be hard or impossible to ¬gure out exactly what shade of phenomenal
color or level of hedonic value someone else is experiencing, and hence
whether two people are experiencing the same shade or level. But still, I
presume, there are objective facts in the relevant sense about those things,
true for all whether or not anyone has epistemic access to such facts.
Second, I am claiming that phenomenal intrinsic value comes
ordinally. Even if you come with me most of the way, you may balk at
ordinality. You may agree that objective phenomenal value exists, but
believe that it comes cardinally, so that it makes sense to speak of an expe-
rience being exactly twice better than another of the same spatial extent
and duration. I can ¬nd little phenomenal grounds for this position, and
believe it due to a confusion of our ordinary judgments about the value
of phenomenal states with their phenomenal value itself, but still it is an
obvious alternative possibility. Or you may hold an intermediate position
on the ordinality-cardinality controversy, that an ordinal comparison of
some happiness differences is rooted in phenomenal value.42 Consider,
for instance, comparisons between an intuitively huge increase in pleasure
and a nearly imperceptible increase in pain. Also, it might be that pain
presents itself “ as, for instance, negative utilitarians believed “ as more
signi¬cant than pleasure, as we will consider again in the next part.
Still, it is massively implausible that the property of being exactly
twice as good as something else could present itself in experience. That is

42 If all differences are comparable in that manner it will yield standard cardinal assessments.

not only because that is an implausibly exact feature to be present there,
but also because it seems too relational and complex a feature even if it
were made suitably vague. It is normatively relevant, for instance, only to
the evaluation of wholes and risks containing such a cardinal value. Nor
was it, I believe, the explicit contention of any classical utilitarian that
that kind of phenomenal speci¬city is in fact present in our experience,
despite the fact that they sometimes concluded as if it were. It is more
as if, in their rush to moral arithmetic, they didn™t consider plausible
alternatives. There is no obvious analogue of this alleged phenomenon
in other forms of sensory experience. Could something present itself as
exactly twice as blue as another, without regard to some arbitrary scale?
Nor do things even present themselves as twice as loud as another, or
twice as bright. Phenomenal spatial and temporal intervals may present
themselves with such a structure, but we aren™t considering at the moment
spatial and temporal features of phenomenal value. Phenomenal value
and disvalue are of the same general type as the intrinsic phenomenal
properties that are characteristic of particular senses and sense ¬elds, like
felt pressure or color. And in those cases, cardinal metrics seem out of
Nevertheless, there are forms of this general type of objection to my
simple ordinal scheme that I believe to possess signi¬cant force. It is
not just that some phenomenal loudnesses seem greater than others, but
that some differences in phenomenal loudness seem greater than other
differences. A difference in loudness may be nearly imperceptible, or intu-
itively large, even if one difference isn™t enclosed in another, even if there
are four levels of loudness involved and we are comparing differences
between levels such that those differences do not overlap. Analogously,
it is not implausible to insist that at least some such differences in value
level present themselves as greater than some other such differences even
when my ordinalism of value levels cannot underwrite it.
My response to this important objection is merely this: The relatively
simple ordinalism of value levels that I have so far presumed is apparently
consistent with my own introspection, and hence I will continue to
pursue it as our primary conception as we proceed. But I will also try to
explore the implications of alternative plausible conceptions.43 For one
thing, I will keep the traditional cardinal conception in play.

43 Indeed, one instance of the ¬nal suggestion incorporated in the objection may indeed be
useful in the next part, because that minor level of additional structure may help underwrite
one key feature of our construction.

But even if the phenomenal facts are exactly as I have claimed them
to be, there may still seem to be a way to, more or less indirectly, deliver
cardinality, and hence to undercut the mere ordinality of basic value on
which I rely in the next chapter. In fact, there may seem to be two ways.
Von Neumann and Morgenstern suggested that we derive cardinality
from rational behavior under risks.44 Presume that a person experiences
successively more positive levels of hedonic tone, from the null valence to
some positive level N. We stipulate that the value of the null valence is 0
and the value of N is 1. Now we expose them to a choice of lotteries over
possible outcomes. They are given, for instance, the choice between the
certainty of some intermediate level of hedonic tone, on one hand, and
a lottery with a 40 percent chance of the null valence and a 60 percent
chance of N, on the other. Von Neumann and Morgenstern showed that if
the individual™s choices between lotteries exhibit certain intuitively ratio-
nal patterns, we can assign cardinal values to the levels for that individual.
And remember that I am already presuming interpersonal comparability.
But, I reply, real people do not uniformly or even usually exhibit those
particular allegedly “rational” patterns for choices over lotteries. You may
be unwilling to risk any chance of torture, no matter how small, in a lottery
that will very likely generate a big increase in your hedonic value. And
that will violate the conditions of “rationality” in question. And of course
there is nothing normatively inappropriate about that. At stake here are
ultimately the proper standards for behavior under risks. The conditions
of “rationality” that von Neumann and Morgenstern presume are not
realistic, and are themselves in need of direct vindication if they are to root
appropriate normative judgments. In Chapter 6, we will return to the issue
of the proper treatment of risks if ordinal hedonic value is as I have claimed
it to be. And it will not deliver cardinality for levels of hedonic tone.45
But there is also a second possible route to cardinality. Edgeworth
developed Bentham™s proposal that we focus on just perceivable incre-
ments in levels of value.46 We might count the number of just perceivable
increments between levels to generate comparisons of differences in value
levels that will yield cardinal comparisons of a sort. If level X is four dis-
criminable levels below level Y, and Y is eight discriminable levels below
Z, we can claim that the difference between X and Y is half the difference

44 von Neumann and Morgenstern (1944).
45 It may also be that the kinds of interpersonal comparisons I presume will not mesh with
the cardinality that von Neumann and Morgenstern can deliver.
46 Edgeworth (1881: 7); Baumgardt (1952: 555).

between Y and Z. But phenomenal differences can be indiscernible in
some comparisons and discernible through others. Two shades of color
may each appear phenomenally identical to a third and intervening shade,
but not to each other. So it may be that different ways of moving from one
level of value to another will yield different numbers of just perceivable
increments between them. T¨ nnsj¨ has proposed that the relevant number
a o
of increments between two levels for an individual is the maximum num-
ber it is possible for that individual to discern between the two.47 But he
does not deploy objective hedonic value with interpersonal comparisons
of the sort I have suggested, and so evades a problem of coherence created
by the interaction of his mechanism and objective hedonic value. It is this:
Two different individuals may be capable of different numbers of level
discriminations between the same two levels of objective ordinal hedonic
value. So we cannot develop a coherent set of interpersonal cardinal
comparisons for the particular sort of hedonic value that I am proposing
in this way. For the same reason, if phenomenal value in fact comes
cardinally, that might yield a different and con¬‚icting sort of cardinality
than T¨ nnsj¨ ™s mechanism. Since single individuals can plausibly differ in
a o
their discriminatory capacities over time and with education, this problem
also probably undercuts the use of this mechanism to deliver cardinal
comparisons of phenomenal value within a life. We might alternatively ¬x
the absolute number of increments between value levels by the maximum
number of differences that anyone can perceive between them. But
that leads to the following dif¬culty: Under that conception, we might
increase the hedonic tone of all the less than fully discriminating individ-
uals in the world by a single increment that they cannot even perceive,
and hence create a huge number of positive level differences unnoticeable
by those who experience them that yet normatively trump the relatively
fewer but very noticeable negative increments involved in the torture of
a single individual. T¨ nnsj¨ endorses something analogous to this.48 But
a o
it is hard to see how such a judgment might be objectively required. That
seems to me the greatest dif¬culty with this general mechanism when
deployed in our current argumentative context. The cardinal judgments
rooted in these various sorts of discriminable and indiscriminable
differences do not seem subject to suitable and objective asymmetric
vindication relative to possible competitors. It is not only that there are
different possible detailed mechanisms of the general sort in question,

47 T¨ nnsj¨ (1998: 68).
a o
48 Ibid., 72“75.

but also that it is not obvious why appeals to numbers of discriminable
and subdiscriminable increments suf¬ce to provide a direct argument for
cardinal judgments that trade off greater pains for some to gain greater
pleasures for others “ for instance, in the torture case I just mentioned.
But let me be clear. I don™t claim to know with any certainty that
hedonic value itself is not cardinal even in intrinsic phenomenal nature.
And I certainly don™t claim to know with any certainty that it is fully
ordinal, that there are no objective facts about differences in value levels
that go beyond ordinality. So while I will continue to develop the
implications of the simple ordinal conception, I will also try to track the
implications for my central claims if a more traditional cardinal concep-
tion is correct. I will also attend as well as I can, which isn™t very well, to
the complications that ensue if some intermediate conception is correct.


Even if you grant ordinality and my other presumptions about phenom-
enal value so far, you may wonder whether we need to introduce some
other sort of structure into phenomenal value. In particular, are there
different sorts of phenomenal value of the general sort I have suggested,
and are they perhaps not fully comparable? Are there different “qualities”
of phenomenal value and disvalue “ for instance, those associated with
pushpin and poetry, or satis¬ed pigs and discontented philosophers?
Let me begin my response to this crucial cluster of questions by admit-
ting one small complication into our model of value. If you can see a
color, then, at least if you are among the majority, you can see something
at least roughly like it in your mind™s eye, in memory or imagination. Call
this “quasi-experience” of the color. Likewise, if you can feel pain, in
memory or imagination you can feel a kind of quasi-pain. If we call the
phenomenal difference between felt and remembered pains a difference
in vividness, then we might ask if vividness is another sort of “quantity”
possessed by phenomenal value, other than ordinal intensity and spatial
and temporal extent.
I think it would not be completely implausible to presume that phe-
nomenal painfulness or pleasantness is instantiated in the world only in
cases where there is fully vivid experience of it, of the sort involved in our
ordinary sensory experience. My own memories, imaginings, and dreams
do not obviously involve experienced badness, though perhaps they more
plausibly involve experienced goodness. But perhaps this is an individual
idiosyncracy or just dumb luck. And in any case, my best overall guess is

that phenomenal badness can be present in certain experiences that are
not in fact physical pains, and yet be of the same phenomenal type as
that present in physical pains, just of minimal intensity. So I will presume
here that phenomenal value and disvalue can be present in these less than
fully vivid ways, in experience or quasi-experience, but will also presume
that these differences in vividness are in fact differences in intensity of
the sort we have already discussed. This might not seem to be a terri-
bly plausible response if we ¬rst consider analogs like the intensity of a
blue, but it seems more plausible if we think about felt pressure. So while
I admit that there are phenomenal value properties beyond the central
cases of paradigmatic sensory pains and pleasures we have so far consid-
ered, I believe that they in fact require no modi¬cation of our general
characterization of phenomenal value properties. They just involve less
intense versions of phenomenal painfulness and pleasantness.
There are other possible complications in this general vicinity that
might be relevant. One might presume, for instance, that hallucinatory
pains involve the presence of genuine objective painfulness, but that the
imagination of past pain does not.49 But in any case, let me turn to the
general issue of whether there are different sorts of phenomenal value
beyond those I have sketched, whether in the case of the imagination or
more generally. Even if my general model of objective value and disvalue
as present within phenomenal experience is accepted, and even my treat-
ment of the quasi-experience of pain and pleasure, still to maintain that a
single continuum of painfulness and pleasantness is the only sort of norma-
tive property present within our phenomenal experience may seem grossly
oversimpli¬ed. An objector may plausibly postulate a range of different
normative or quasi-normative properties in a variety of different classes:
(a) Even by ancient writers like Plato, the bodily pleasures and pains were
augmented with mental pleasures “ for instance, the pleasures of anticipation
and memory.50 These may involve a different kind of pleasure than physical
pleasure, and not just a fainter or less intense pleasure of the same kind. And
it may involve different phenomenal value properties.

49 In any case, it seems implausible that the memory or imagination of a pain really is itself a very
signi¬cant pain, though of course imagination and memory of all sorts of situations can be
accompanied by present uncomfortable feelings that are present pains of some signi¬cance.
On the ethical view developed here, whether or not hallucinatory pains and dream pains
count normatively in the same manner as ordinary pains comes down to whether they
involve the presence in the world of objective painfulness, at least on sense data, and in
exactly what form. That is the natural implication of normative realism.
50 See, for instance, the Philebus.

(b) The previous distinction may or may not be identical to the distinction that
some authors have drawn between localized pains and pleasures of
paradigmatically physical sorts, and nonlocalized pleasures such as that of
reading a good book.51
(c) There is a long utilitarian tradition that suggests that there are differences in
quality among phenomenal pleasures that affect their phenomenal nature, so
that the pleasures of pushpin and poetry are not commensurable in the way I
have presumed.52
(d) There is neurophysiological evidence for speci¬c itch receptors, which
when stimulated may present a sensory experience demanding itching in
particular, while, one might argue, ordinary painfulness demands succor
(e) There is a very wide range of emotional states, including anger, joy,
depression, guilt, shame, and anxiety. These arguably involve complex
phenomenal presentations that may involve subtly different sorts of
phenomenal normative properties. To focus on a speci¬c vivid case, think
about grief. This is a highly negative state that involves a distinctive
phenomenal component of emotional pain, but doesn™t seem to involve the
same sort of phenomenal badness as physical pain.
(f) A number of sensibility theorists, such as Wiggins and McDowell, have held
that properties like humorousness may be present in our experience,
properties presenting things as properly laughable. This may involve yet
other sorts of phenomenal normative properties, of very complex varieties.
(g) One may read the continental phenomenological tradition, leading down
through Heidegger and Sartre, as proposing a yet more expansive
conception of phenomenal experience that provides grounds for yet subtler
forms of normativity.53 According to Heidegger, perhaps even the basic sort
of experience of things is emotional, so that the anxiety of our situation is
presented to us but our knowledge is mostly know-how. But more to the
immediate point, perhaps tools present themselves as at hand in their full
teleological nature, or other people present themselves in experience as
radically different from the inanimate and as deserving relevant care. In
general, tools or other Dasein may present themselves in our phenomenal
experience as demanding care of some speci¬c sort, and not merely
avoidance or pursuit in the manner of pleasantness and painfulness.
According to Sartre, there are speci¬c negativities present in our experience,

51 Hospers (1961); Edwards (1979: 35“45).
52 Mill™s Utilitarianism is one obvious source. It may well be that quality in his sense is de¬ned
by our tendency to choose certain pleasures over others when we™ve had experience of both,
so that it isn™t a phenomenal feature of pleasures at all. But presumably there are phenomenal
differences among pleasures to which the differential judgments that constitute quality are
partially sensitive. See also Hutcheson (1968: vol. I, 117“119) and Edwards (1979).
53 Heidegger (1996: Division One); Sartre (1953: Part One).

such as Simone™s not being as expected in the Flore, and such absences may
present themselves as normative lacks to be ¬lled in quite speci¬c ways.

I grant the intuitive force of some of these cases, and admit that this
is a point that requires considerable attention if the general conception I
have sketched is roughly correct. Nevertheless, let me defend my simple
conception of phenomenal value as more than just a ¬rst and oversimple
model. There is an understanding of the nature of our phenomenal
experience that is plausibly adequate to all of these complex phenomena,
and that yet involves the postulation of no additional and complex
normative properties. It deploys the following elements:

(a) There is our ordinary sensory experience of material entities “ for instance,
tools and other people™s behaving bodies.
(b) There is our complex and subtle sensory experience of our own bodily
(c) Either (a) or (b) may include the experience of localized pleasantness and
painfulness of various degrees and spatial distributions.
(d) We can have a kind of quasi-experience of all these concrete sorts, in the
mind™s eye (or the mind™s throat or body), in imagination or in memory. In
some cases, this will be quasi-pain and hence involve a low intensity of
phenomenal disvalue.54 But there are other important phenomena of this
general sort. For instance, we can experience in imagination the picking up
of a tool or a laugh, an action that we are disposed to perform but aren™t

This is a complex set of resources, and it provides, I believe, enough to
account for all the complex phenomena noted in the objection, at least
when these experiences and quasi-experiences are linked up in the appro-
priate way “ say, by neural mechanisms underwriting what we would once
have called “association”. This is, roughly, how: The pleasures of antici-
pation and memory may often include the quasi-experience, in imagina-
tion, of bodily pleasures and pains, but only vaguely localized. A lot of the
complex phenomenal differences between what are alleged to be differ-
ent qualities of pleasure, or that are involved in different emotional states,
may be, as James famously suggested, differences in the experience of
one™s associated bodily states,55 or perhaps differences in quasi-experience
of analogous sorts. For instance, there may be special, even evolutionarily

54 This quasi-experience may involve a certain degree of phenomenal vagueness “ say, regard-
ing the location of a pleasure.
55 James (1950: Chapter 25).

¬xed, bodily responses triggered by whatever emotional stimulus also gen-
erates whatever pleasure or pain happens to be a component of an emo-
tion. And so one™s special phenomenal experience of an emotion may
include not merely an experience of that pleasure or pain but the expe-
rience of one™s bodily response. And indeed, there may be a kind of asso-
ciation between analogous sorts of quasi-experience as well. Also, one™s
dispositions to pick up what is present at hand and use it in some way, or to
laugh at the humorous, or to act toward something or someone in a caring
way, can be present in one™s faint quasi-experience of what one might do
but isn™t doing, when the activated disposition for such an act moves one
to a degree but isn™t suf¬ciently powerful to actually generate the act.
What supports this model over more expansive alternatives? I adopt it
largely because it is a relatively simple model that is not obviously false,
and that seems adequate to at least my own introspection. But there are
other reasons to prefer it. Painfulness and pleasantness of the sort I have
presumed are quite familiar and concrete properties, quite analogous
to traditional sensory properties. And other candidate phenomenal
normative properties seem quite different. It seems that our experience is
experience of the concrete, even if sometimes in faint imagination, that
the concrete indeed exhausts what we seem capable even of introspecting.
For instance, our genuine introspection of what we can think seems to
involve merely the faint presence to us of words that we hear in our
mind™s ear or speak in our mind™s throat, and perhaps concrete elements of
their meaning. If the content of an experience were concretely identical
to mine, it would plausibly be at least qualitatively identical in content.
But it is plausible that most of the other candidate normative properties
proposed as phenomenal additions either supervene on the more ordinary
sensory presentations of experience, or are clearly due to additional and
idiosyncratic associations brought to that experience that our model can
capture in exactly that manner. But in either case, our simple model is
preferable as a model of the phenomenal content of experience.
Compare, for instance, humorousness and painfulness: One can
imagine two otherwise concretely identical experiences of things, one
that is painful, and one that is not (perhaps because of some drug one has
taken). Painfulness varies independent of other phenomenal properties in
our experience, so we can understand what it is for objective painfulness
to be suitably independent of other sensory properties in the world.
We can understand the possibility of two otherwise concretely identical
¬‚owers, one that is painful to touch and one that is not. But humorousness
seems to be a different sort of case. Either it is the case that two otherwise

concretely identical situations can fail to be equally humorous, as they
can fail to be equally painful, or it is not. But the ¬rst half of the fork,
which supports the view that there are additional phenomenal normative
properties, is implausible. It is clear that two otherwise concretely
identical situations must be equivalent in humorousness, and that the
supervenience of any allegedly distinct humorousness that can be squared
with this fact could not be plausibly explained. Other than causation,
there are plausibly no necessary connections between distinct existences.
So if humorousness supervenes on the ordinary concrete features of
things, which it plausibly must, it cannot really be a distinct phenomenal
property. Perhaps Heidegger held that in a situation otherwise concretely
identical to this one, objective anxiety could fail to be present, or that
an animal concretely identical to you or me could fail to present that
je ne se quai characteristic of full humanity. But we shouldn™t take such
possibilities seriously. Of course, different people™s experience of the same
situation can be distinct. Perhaps a caf´ concretely identical to Les Deux
Maggots on a certain day could fail to present to Sartre the absence of
Pierre. But that isn™t really the ¬rst half of the fork. Whatever anxiety or
absence is present in experience in such cases clearly involves idiosyncratic
reactions to identical concrete presentations. Our model can explain
these by deploying concrete quasi-experience of idiosyncratic actions
we are inclined to perform or of idiosyncratic fantasies we undergo in
such situations. If, for instance, an irreducible telos can be present in our
phenomenal experience of things, then it should be possible that an exact
concrete duplicate of my tool yet not be ready to hand in the same manner,
and not merely that I take one to be ready to hand and not the other. Don™t
believe it.
None of this is conclusive. If the picture I have presented here is even
roughly correct, then this is one place where further work would be
most useful. But let me give a little more attention to what I take to
be the most troubling speci¬c cases. There are in effect three sorts of
phenomenal complications that seem most plausibly to require revisions
and complications in my model.
First, there may be types of phenomenal value that are not fully
comparable “ for instance, those associated with sex and poetry. But,
as indeed Mill™s classic discussion suggests, differences in “quality” of
pleasures may be differences in the degree to which we prefer them,
rather than phenomenal differences. Of course, it may be objected, these
differences in judgments might re¬‚ect some difference in phenomenal
quality. But it need not be a difference in a value property.

Second, different acts may seem to be the appropriate result of distinct
sorts of phenomenal value. Consider the phenomenal experience of an
itch. Here we have something very like other sorts of sensory experience,
which is, for instance, obviously spatial, and which in fact involves a
distinct set of sensory receptors. And it seems to specify a certain sort of
appropriate response beyond the general to-be-eliminatedness that pain
presents. But, I reply, the way in which scratching seems appropriate
to an itch, if not simply a matter of our dispositions in response to the
phenomenal experience, seems to re¬‚ect some sort of quasi-experience
of the scratching, and that seems something that is the presentation not of
another value property but of an imagined response to a value property.
There is, apparently, something phenomenally special about itches, but
not something that involves a different sort of phenomenal value.
Third, there are some emotional states that apparently involve distinct
value phenomenologies very different from localized physical pains and
pleasures. Consider, for instance, depression or grief. But some of the
most characteristic elements of these emotions are behavioral, or at least
involve dispositions to behave. They involve distinctive phenomenologies
only in the sense that the appropriate range of bodily responses is felt, at
least in quasi-experience. This may sometimes involve no phenomenal
value at all. Or it may involve what Edwards called a “universally localized
feeling” that is felt all over, as in the pains of fatigue and chill. But I think
it paradigmatically involves a somewhat vaguely localized pain, of the
sort characteristic of quasi-experience, and in a way that is not constant,
at least across individuals. Someone has a vague sinking feeling in their
stomach, someone else a gnawing in their chest, and that may involve
quasi-experience of painfulness.56


Our consideration of the structural details of the natural normative prop-
erties is now concluded. But we still have a little way to go in sketching
this conception. A second important set of complications involves how
these objective phenomenal unconstituted normative properties might ¬t
into the world.
There are two ways. According to the ¬rst, sensory properties just as we
experience them to be “ for instance, the irreducible and unconstituted

56 Edwards (1979: 41).

phenomenal yellow of a certain ¬‚ower or taxi, or of the wrapping on a
package “ are right there on the surfaces of the ordinary material objects
that we experience to have such properties. The yellow is on the surface
of the ¬‚ower itself.
But what objects do we experience to be painful in the relevant way?
What plays the role in the case of phenomenal disvalue that the ¬‚ower
plays in the case of yellow, according to this ¬rst way of discharging
our metaphysical presumption? Two possibilities: When we experience a
throbbing pain in our arm, it might be that the nasty painfulness or disvalue
is right in there, a property of the ¬‚esh inside our arm, or perhaps of some
ghostly (and ghastly) object that comes and goes in our arm as the pain
throbs. Another possibility: Sometimes we experience physical pain as if it
is on the surface of an ordinary external concrete object that is hurting us,
the hot stove burner or the sharp edge of a Cuisinart blade or the leaves of
plants that burn everyone who touches them. And it might be, on this
conception, that these experiences are veridical, that hedonic disvalue is
on the objects.
This ¬rst way (or pair of ways) in which phenomenal properties gener-
ally and hence phenomenal value and disvalue in particular may be present
in the world is re¬‚ected in the natural and naive human conception of
things. This is what we fall into believing when we are least philosoph-
ically re¬‚ective, and it probably underlies ordinary talk that there is a
pain in my arm, right there where it hurts. Notice the exact symmetry
on this conception between ordinary sensory properties like phenomenal
yellow, on the one hand, and physical painfulness and pleasantness, on the
other. We can coherently conceive that a schoolbus remains yellow even
in the dark or when there is no one around to see it. Likewise, on this
conception, in some broad sense of possibility, phenomenal disvalue could
exist “ say, on the surface of a plant or the edge of a Cuisinart blade “
when no one was around to experience it. The analogy with yellow also
suggests that there may be cases of the mere imagination of pain or the
nonveridical perception of it, as of golden skyscrapers. In such cases, it
may be, on this conception, that nothing really has phenomenal disvalue.
Or it may be that there is something suitably painful, but merely the kind
of thing deployed by the second manner of discharging our metaphysical
presumption that there is hedonic value, which we are about to consider.
Despite the fact that this ¬rst way to ¬t objective hedonic value into
the world is the most naive and straightforward, it will no doubt seem
objectionable to many analytic ethicists. It suggests that sensory experi-
ence in general involves the presence of irreducible and unconstituted

phenomenal properties on external objects, which may seem unscien-
ti¬c and implausible, and certainly unfashionable and out of date. And
it presumes a kind of symmetry between the case of ordinary sensory
experience and the experience of pain that some may ¬nd misleading.
So some may prefer the second conception of how phenomenal value
or disvalue ¬ts into the concrete world, which, while more baroque and
less natural to humans, as are all alternatives to the naive conception, yet
probably has more living partisans among analytic philosophers.
On that second conception, phenomenal properties, the properties
that appear in our immediate experience, are not out there in the exter-
nal world of ordinary material objects. Or at least such properties that are
traditional “secondary” properties are not out there. There are no irre-
ducible phenomenal colors splashed on trees and tables. Such properties
are, rather, had by odd corners of dualist substance, or at least by sense
data. In idealist variants of this view, such sense data or immaterial minds
constitute everything there is. But in what is probably the paradigmatic
such conception, sense data or internally colored dualist substances are
somehow causally secreted by physical brains that have the right level of
active complexity.
On this second way of discharging our key metaphysical presump-
tion, hedonic value and disvalue are really in the world, but are lodged in
rather unusual regions of it, on the physically unconstituted entities that,
according to such conceptions, help to make up creatures with compli-
cated psychologies. On such a view, it may well be that the imagination or
imaginative recollection of pain involves some genuine disvalue “ say, on
the surface of the sense data involved in that quasi-experience “ and such
a model of quasi-experience may be favored even by those who favor our
¬rst manner of discharging our metaphysical presumption for the case of
ordinary sensory experience.
According to either of these two conceptions, there are unconstituted
but natural phenomenal properties of physical painfulness and pleasant-
ness and hence badness and goodness instantiated in the world. To feel
paradigmatic physical pain is literally to feel bad. To feel such pleasure is
literally to feel good. When you have a throbbing pain in your arm you
literally have a throbbing badness in your arm, or at least on the surface
of your sense data.
There is an obvious pressing objection, which goes like this: Such
painfulness and pleasantness are not plausible properties of real objective
things. Certainly there are states of people™s experiencing pain. But there
are no genuine things pains in the world, with genuine, full-¬‚edged, and

unconstituted concrete properties of painfulness “ for instance, located in
our arms. Nor do ordinary concrete objects like Cuisinart blades really
have painfulness on their surfaces. Nor is it at all plausible to think that
there are sense data under our hats with properties of painfulness either,
nor other dualist claptrap. Physicalism is the correct conception of the
world. Everything is constituted out of the little particles and universal
forces that physics has discovered or will discover. Such resources certainly
underwrite no properties of unconstituted painfulness, nor even of irre-
ducible phenomenal yellowness. Either yellow is constituted somehow
out of the physical properties of ordinary things “ for instance, by their
surface spectral re¬‚ectances “ or there is no yellow in the world at all
but merely the experience of yellow, which is itself somehow constituted
out of the machinations of the microphysical. Ever since Galileo, it has
seemed radically implausible to suppose that color just as we experience
it to be, color unreduced, phenomenal color “ in which, for example,
reddish purple is very similar to bluish purple even though they corre-
spond to wildly different wavelengths of light “ is really out there on the
surfaces of things. What™s more, even if there were phenomenal properties
like that out in the world, still there is a crucial disanalogy between yellow
and the case of hedonic value. Perhaps yellow presents itself as a prop-
erty of external objects in our experience in a way that may mislead the
unwary and unre¬‚ective, but painfulness is a very different sort of thing.
Painfulness isn™t on the edge of sharp blades, nor even literally inside of
our hurt arms, and this was evident even to prescienti¬c common sense.
Pain is much too subjective a thing for that. No one feels the pain in
someone else™s leg, even if their hand is nearby, even if they are touching
the leg, indeed even if their hand is inside the leg. Far from being included
in the very naive and pre-theoretical view of the world, hedonic value is
really at home only in massively implausible dualist, idealist, or sense data
conceptions of the world.
Reply: Not everyone, even among analytic philosophers, is a physical-
ist. Dualists and others who believe in unconstituted qualia may plausibly
incorporate hedonic value and disvalue and hence objective normative
properties into their conceptions of the world. Indeed, qualia are gener-
ally accepted to be one of the greatest dif¬culties facing physicalism. It is
certainly not inconsistent with any current consensus to hold that qualia
are unconstituted by the physical and that qualitative experience cannot
be constituted by the physical. That is, it is not inconsistent with any cur-
rent consensus to hold that dualism, or neutral monism, or even idealism
of appropriate sorts involving unconstituted qualia, is correct. You can call

this merely “property dualism”, if that makes you feel better. A number
of current authors even endorse a more or less naive direct realism, with
unconstituted qualitative properties on objects, and though this is probably
not within the central consensus of analytic philosophers, it is at least the
standard unre¬‚ective view. And it certainly avoids the baroque machin-
ery of idealism and dualism. If unconstituted qualia exist in any of these
ways, then there is no barrier to the model presumed here. What™s more,
physicalists who accept that phenomenal experiences can be constituted
by the physical often do so because they reject two-dimensionalism in
its paradigmatic forms, and embrace necessary a posteriori propositions,
which link certain physical states to qualitative experiences of certain
kinds, but not via meaning links assessable a priori. There are two ways in
which they might think this works. It might be that the mere experience
of certain phenomenal properties is constituted by the physical, as thought
of unicorns is constituted by conditions that include no unicorns, or it
might be that phenomenal properties are themselves constituted by the
physical. In the second case, on something like the model suggested by
constitutive naturalism, they can accept that the natural normative phe-
nomenon that I have identi¬ed exists, and insist only that it is constituted
out of the physical in a way forbidden by paradigmatic forms of two-
dimensionalism. It would still remain the only at once plausibly natural
and normative phenomenon.
But my general point is that the hedonic value and disvalue I am sug-
gesting should be treated just as any other qualia are treated. There are
physicalist accounts of qualitative experience that reject the conditions
necessary for the plausibility of my proposal, but they are far from a con-
sensus view. What™s more, I repeat, the most naive and natural human
conception of the world, to which even Galileo would have reverted
when hauled into court to testify, is not only one in which phenomenal
color, color just as we experience it to be and not as constituted by a
tendency to re¬‚ect certain wavelengths of light, is found on the surfaces
of ordinary material things, but also one in which physical pain, a really
bad object, can literally be found in hurting legs and broken ¬ngers and
drilled teeth.57 Perhaps, at least before we think about it re¬‚ectively, we
think things that imply that anything else inside a hurting ¬nger or tooth
would confront the pain, that it is not really a subjective entity. Or per-
haps we pre-re¬‚ectively believe in subjective objects, which only single

57 In Mendola (1997), I called this the “forensic truth”.

individuals can experience. Berkeley not implausibly pointed out that the
sensory experience of intense heat or cold seems to be an experience of
painfulness, so perhaps, before we think re¬‚ectively about it, we presume
not only that phenomenal heat is to be found on activated stove burners
but also that intense painfulness is there, which everyone would feel if
only they touched the burners. And naive conceptions like this may have
a certain sort of natural and appropriate epistemic weight, however far
behind we™ve left them in our more philosophically re¬‚ective moments.
What do I think? That probably isn™t relevant, but in any case there is
more than one answer. I once was a dualist, and once a naive realist. But
sometimes now I am a physicalist who thinks that experience of qualia is
like belief in unicorns, which is a problem. Still, in Human Thought I argue
in this way:58 Neither dualism, idealism, neutral monism, nor naive realism
is adequately plausible, but many contemporary physicalist philosophers
are wrong to think that their particular view of the world is any more
plausible. No positive and coherent conception of the world of which
humans are capable can be privileged over those that deploy unconstituted
qualia, and hence hedonic value and disvalue. The central normative
claims made in this chapter have as good an epistemic status as any other
positive claims about the world that we humans can coherently conceive;
they are equally re¬‚ective of the shared, though perhaps hallucinatory,
conception of the world that our shared human experience underwrites.
And indeed, hedonic value and disvalue are the only sort of unconstituted
normative phenomena that are coherently conceivable by human beings.
That is one sort of vindication of the normative in general and of my
proposal in particular, at least relative to anything else we might plan to
privilege over them. Of course that may be less vindication than seems
desirable, but perhaps it is the best we can get.
The conception proposed in this chapter is reminiscent of the sensi-
bility views rejected in section I. But I claim that we can conceive of
hedonic value and disvalue as fully objective properties, and that they are
part of a natural human conception of things that would vindicate our
normative practice if only that natural human conception turned out to
be right. The sensibility theorists do not and, I believe, cannot plausi-
bly make such a claim about the normative properties they favor. What™s
more, no human, nor for that matter any rational shark or pain-feeling
dragon or Martian, is beyond the range of the particular “sensibility”

58 Ibid.

I deploy, beyond pain and pleasure, which is not the case of the more
socially idiosyncratic normative sensibilities proposed by contemporary
sensibility theorists. Consider the analogy of humorousness. This is what-
ever it is that the suitably re¬ned ¬nd humorous, according to sensibility
theories. But the relationship alleged to exist between the feeling or sen-
sory experience involved in humor and objective humorousness is not the
same as that between the experience of pain and hedonic disvalue. Even
the unre¬ned can feel genuine pain. What™s more, people differ not only
in what things they ¬nd humorous, but also in having much of a sense
of humor at all. While not everybody feels physical pain from the same
causes, everybody knows what it™s like.59 There is no need for a direct
vindication of pain sensibility over others available to humans, for in the
suitable sense there are no alternative sensibilities available to humans.
These are crucial differences between my proposal and the view of the
sensibility theorists.
Mackie™s famous queerness objection seems to this degree correct: No
ordinary machinations of the physical stuff of which we believe at least
most of the world to consist can generate normative properties. Some-
thing as strange and as plausibly nonphysical as the phenomenal must be
considered before objective value becomes plausible. But that is not so
strange as the nonconcrete non-natural properties of Moore. We are stuck
with the problem of qualia in any case, and so we can have objective hedo-
nic value for free. Well, maybe not quite for free. The crucial normativity
of value may seem queerer still than ordinary phenomenal properties.
But if I am right that the characteristic phenomenal property involved in
paradigmatic physical pains is in fact a kind of nastiness, that when we
feel bad we literally feel one fundamental sort of badness, then we can
plausibly explain the crucial supervenience of the normative on the nat-
ural in a way in which those who favor Moorean non-natural normative
properties cannot, and we can form some genuine understanding of what
such otherwise spooky unconstituted normative properties might be.
Where are we? We have isolated the sole type of unconstituted natu-
ral normative property, ordinal hedonic value or disvalue, though I have
also promised to consider plausible close variants, which deploy cardi-
nal structure or something intermediate. Instances of such a property
must support proper justi¬catory reason giving in our world, probably
even if two-dimensionalism is false and constitutive naturalism is true, but

59 Or rather, while some very unfortunate and often short-lived humans don™t feel pain, they
agree that this is a defect.

almost certainly if two-dimensionalism60 is true. The only other reason-
ably live competitors are outright skepticism about justi¬catory reasons,
and some complicated phenomenologically based extension of my pro-
posal. Whether we like or not, some kind of hedonism is the sole theory
of basic value plausibly consistent with the facts, and my speci¬c ordinal
variant seems the most plausible and simplest form. We also saw in the last
chapter that hedonism is in proper accord with our normative intuitions
of the same level of generality and abstraction. So the ¬rst element of the
Hedonic Maximin Principle is suitably secure.

60 Or a close analogue like my own view.

Part Three
Just Construction

The second element of the Hedonic Maximin Principle “ its maximin
structure “ is the central subject of this part. Chapter 7 will argue that
it is suitably intuitive. This chapter develops a direct argument for HMP
on the basis of the ordinal hedonic value discerned in the last chapter.
We will also consider the implications of other conceptions of hedonic
value “ for instance, a cardinal conception.
“Possible worlds” are fully detailed ways the universe might be. If there
are other sorts of basic value than ordinal hedonic value found in other
wildly possible worlds, still I will presume, for most of our discussion here,
that such worlds are too unrealistic to be of concern to ethics. Call the
worlds that are in question for us, which have merely ordinal hedonic
value, “feasible worlds”.
The Hedonic Maximin Principle speci¬es the proper ordering, from
worst to best, allowing for ties, of lotteries over feasible worlds.
That this is an ordering of things from worst to best means that (a) each
member of the set of those things is either better, worse, or as good as any
other speci¬c member; (b) the “better than”, “worse than”, and “as good
as” relations are transitive; and (c) nothing in the set holds more than one
of the “worse than”, “better than”, or “as good as” relations to any other
speci¬c member.
Lotteries over feasible worlds are simply collections of the worlds, each
world with a probability attached, and such that the sum of the proba-
bilities in each collection is one. We face risky alternatives, and lotteries
over worlds are a way to model risky alternatives. The point is that we
must assess actions, and actions result in lotteries over outcomes and not
merely in particular outcomes.

The question of the proper ordering of lotteries over feasible worlds is
only one normative question. I begin with this question because it is the
one question that I know how to answer directly on the basis of ordinal
hedonic value that yet can plausibly root ethics, through the mechanism
of Multiple-Act Consequentialism.
As we will see, the ordinality of hedonic value has signi¬cant implica-
tions regarding this question. That is not a surprise. There are developed
literatures on the implications of analogous sorts of ordinality involving
preferences for the summary evaluation of states of affairs, which make
the principal conclusions of this chapter quite expectable.1 Ordinality, we
know, in the context of other plausible constraints, leads quite naturally to
maximin. Indeed, it is only peculiar details of our construction “ in par-
ticular, the differences among null, positive, and negative hedonic value “
that require that I develop my own treatment.
That maximin is an obvious implication of ordinality is intuitively
advantageous at least in some ways. One standard objection to classical
utilitarianism can be evaded by HMP through its maximin structure.
The sum of individual pleasures, which classical utilitarianism aims to
maximize “ and, more generally, the summation of value in states of
affairs to determine their overall value “ seems intuitively objectionable.
Here is Samuel Schef¬‚er:

[U]tilitarianism gives no direct weight to considerations of justice or fairness
in the distribution of goods. Provided that net aggregate satisfaction is maxi-
mized, in other words, utilitarianism is indifferent as to how satisfactions and
dissatisfactions are distributed among distinct individuals. So if overall satisfac-
tion will be maximized under an arrangement in which goods and resources are
channeled to people whose circumstances are already comfortable, while other
people are allowed to languish in abject poverty, then that arrangement is pre-
cisely the one that utilitarianism will recommend. . . . [Since utilitarianism] makes
the importance of distributive considerations entirely dependent on their ef¬cacy
in promoting maximum satisfaction, it both denies their intrinsic moral signif-
icance and requires them to be set aside if ever and whenever their ef¬cacy

This intuitive worry supports the modi¬cation of classical utilitarianism
that we adopt in this part.

1 D™Aspremont and Gevers (1971); Hammond (1976); Strasnick (1975).
2 Schef¬‚er (1988: 2“3).

Nevertheless, our concern in this chapter will not be normative intu-
itions. Indeed, despite the general intuitive structural advantage that the
distributionally sensitive form of HMP provides, some of its particular
structural details may initially seem quite counterintuitive. We will see
in the next chapter that such a structure is in fact intuitively defensible,
at least when allied with hedonism. And we will see in Chapter 8 that,
in conjunction with MAC, it delivers a properly intuitive detailed moral
code. So I request your patience on this crucial point. But our immediate
concern is to provide a direct vindication of the maximin structure of
HMP, independent of moral intuitions.
I must also request your patience in a second way. My argument here
will be a dry thicket, and I lack the art to sugarcoat it. But there is
a positive and somewhat intuitive thread through this labyrinth. Marx,
Gandhi, De Beauvoir, and King each reminded us of obvious inequalities.
And in a world where it makes no sense to add and subtract value in the
manner of classical utilitarians, where it makes no sense to claim that
transferring resources from the well-off to the badly-off could entail that
the well-off would lose more than the badly-off would gain, this is perhaps
especially inexcusable. But again, our focus here is not on normative
This chapter develops an ordering of lotteries over feasible worlds by
deploying construction principles that, in conjunction with ordinal hedo-
nic value, imply it. These construction principles, which are capable of
asymmetrical vindication relative to competitors, give my metaethical
view some af¬nities with practical reasoning and constructivist views.
They also assure that HMP re¬‚ects basic distributional justice of at least
one kind, that it is in one sense a just good that we specify here.
The three immediately succeeding sections focus speci¬cally on the
proper principle for ordering feasible worlds, which I will hereafter call
“worlds”, from worst to best. Section I articulates certain constraints on
the proper ordering of feasible worlds. Section II shows that they are
capable of a properly deep legitimation in the world we inhabit. Section
III argues that the constraints, which re¬‚ect the nature of ordinal hedo-
nic value as well as the relevant construction principles, entail a speci¬c
ordering of feasible worlds.3 Section IV examines the dependent ques-
tion of the proper ordering of lotteries over feasible worlds. Section V

3 This central argument of the chapter is adapted with many modi¬cations from Mendola

considers the effects of relaxing some of the constraints, and also considers
what follows if hedonic value is cardinal or has some intermediate sort of


I will argue in this and the following two sections that the proper ordering
of feasible worlds follows a maximin rule with the following form: Given
two worlds that contain the same number of momentary bits of phenom-
enal experience, which I will call “¬‚ecks” of phenomenal experience,
the better world is the one that has the better (namely, more hedonically
valuable or least disvaluable) worst (namely, most hedonically disvaluable
or least valuable) ¬‚eck of phenomenal experience, or, in case of a tie, the
better second-worst ¬‚eck, or, in case of a further tie, the better third-worst
¬‚eck, and so on. This is a lexical maximin rule: We maximize the state of
the worst-off, and in case of ties move on to the second-worst-off, and
so on. I will argue that, since there is a null level of hedonic value, any
world is equal in value to any other world that has the same number of
¬‚ecks of phenomenal experience as the ¬rst at each level of hedonic value
and disvalue plus any number of null-valenced ¬‚ecks. So if we can rank
feasible worlds containing the same number of ¬‚ecks of experience, then
we can also rank those that do not.
I will close the gap between the ordinal hedonic value found in feasible
worlds and the proper ordering of those worlds by deploying various
construction principles, principles that constrain the way in which the
relative value of a whole feasible world is a function of the value contained
within it. The next section will legitimate the construction principles
I deploy. This section merely articulates those principles, and also two
further constraints that capture the nature of hedonic value and that of
our question. I will provide in this section only the minimal motivation
and discussion required to allow the constraints to be understood.
Review our situation. Feasible worlds contain only one sort of basic
value and disvalue, we presume. Each ¬‚eck of experience in such a
world exhibits a single level of positive hedonic value and hence good-
ness, or a single level of negative hedonic value and hence badness,
or it exhibits no positive hedonic value or disvalue and hence a null
value. All hedonic disvalue is worse than all hedonic value, and the null
valence is in between. And hedonic value and disvalue come merely ordi-
nally, though interpersonal comparisons are possible. There are various
objective levels of ordinal hedonic value and disvalue. Hedonic value

or disvalue with that general character is the ¬rst key element of our
What™s more, we face a certain question. We are attempting to deter-
mine the proper ordering of feasible worlds, on the way to answering a
dependent question about the proper ordering of lotteries over worlds.
I will assume that nine constraints govern the construction of an order-
ing of feasible worlds containing merely the hedonic value and disvalue
that I have noted.4 These restrictions are:
Restriction 0 (Completeness): There is a complete ordering from worst to
best (allowing for ties) of feasible worlds.
This restriction is set by our question, by the problem we address. It
might seem quite optimistic, but it is a kind of practical necessity if full
normative evaluation is to be possible, and we will see that this restriction
can be met. Of course, lots of moral assessments might be possible without
completeness, even if there is no complete moral ordering of worlds. But
we will see that there is such an ordering.
Because this is an ordering from worst to best, it will have the following
properties: If one world is better than (respectively worse than, equal in
value to) another, and that world is better than (respectively worse than,
equal in value to) still another, then the ¬rst is better than (respectively
worse than, equal in value to) the third. If one world is better than another
that is equal in value to a third, then the ¬rst is better than the third. I also
presume that the relations of being better than, worse than, and equal in
value to are transitive, and that no world bears more than one of those
relations to any given other. Because our ordering is complete, any given
feasible world is better than, worse than, or equal in value to any other.
But also note a limitation on the feasible worlds to be considered here,
which quali¬es this claim of completeness: I will restrict our consideration
to feasible worlds each of which contains only a ¬nite number of ¬‚ecks of
phenomenal experience. Hereafter by “feasible worlds” or “worlds” these
are what I mean. The reason for this restriction is straightforward: Since
in¬nite collections can be put in one-to-one correspondence with parts
of themselves, any ordering of such collections based on purely ordinal
information is a breeding ground for paradox. In¬nite sums are subject to
the same sorts of dif¬culties, so this isn™t a special problem for the view to
be developed here. One might object that even if there plausibly isn™t an
in¬nite number of creatures with experience in any feasible world, still a

4 As I™ve noted, this conception re¬‚ects developments in the social choice literature. See, for
example, Sen (1986).

given individual may have an in¬nity of in¬nitesimal ¬‚ecks of experience. I
don™t believe this, because experience has a ¬nite grain and hence ¬‚ecks are
not in¬nitesimal. But in any case, by deploying temporal and spatial grains
of some arbitrary but still ¬nite size, we can model even a temporal and
spatial continuum of in¬nitesimal ¬‚ecks to whatever degree of precision
is required.
The next constraint is Restriction A (Abstraction): The information
relevant to determining the relative value of a world is merely the number of ¬‚ecks
of phenomenal experience of each degree of objective hedonic value or disvalue (or
null value) that it contains.
Hedonic value is the only basic value that there is in feasible worlds.
Worlds that contain the same numbers of ¬‚ecks of experience at each
level of value and disvalue are equivalent in value. How such moments
are located in the world that contains them “ for instance, how they are
bundled together into individual lives “ does not affect the relative value
of the worlds. This assumption may seem problematic, and we will return
to the issue of its legitimacy in the next section.
Restriction A raises the issue of how to represent the hedonic value
of ¬‚ecks of experience. Such value comes ordinally, but it also involves
a three-way difference among null, positive, and negative hedonic value.
We can properly speak of inter- and intrapersonal comparisons of value
levels, hence properly ascribe numbers to those levels in this way: To
the null level of value a 0, to each level of positive value some arbitrary
positive rational number with greater numbers assigned to better levels,
and to each level of negative hedonic value some arbitrary negative ratio-
nal number with lower numbers assigned to worse levels. But we cannot
properly take these representations in other than an ordinal sense, as bear-
ing information about the relative size of value differences. Any assign-
ment of rationals meeting the stipulations noted will be as good as any
By Restriction A, and given a particular assignment of rational numbers
to levels, the information about a world relevant to the determination of
its relative value is born by a set of rationals, which contains for each ¬‚eck
of experience in the world an element representing its level of hedonic
value or disvalue. Hence our problem of ordering worlds from worst
to best reduces to that of ordering, in regard to the value of what they
represent, sets of rationals. This will allow a second and quasi-formal
expression of the constraints we have yet to consider, which will be useful
in our quasi-formal proof in section III that they entail the ordering of
worlds I have promised. Let such a set representing a world be said to

be “>” (respectively =, <) another iff the world the ¬rst represents is
better than (respectively equal in value to, worse than) that the second
represents. I will consider the problem in the guise of ordering not sets
but rather ordered n-tuples of rationals, one n-tuple generated from each
set of rationals by arranging it from the least rational on the left to the
greatest on the right, allowing of course for ties.
With this background in place, we can go on to the other constraints, in
¬rst an informal and then in a quasi-formal guise. The informal guise will
be important when we legitimate the constraints in the next section, and
the quasi-formal guise will be important when we trace their implications
in section III.
Restriction B (Null Addition): Adding to or deleting from a world a ¬‚eck
of experience with null hedonic value yields a new world equal in value to the
In other words, the null valence is just that, a valence of value that
makes no difference to the value of the worlds in which it is contained.
Adding or removing a ¬‚eck of phenomenal experience of null value to
or from a world yields a new world equal in value to the ¬rst. There is
nothing in the new state of affairs that makes it better or worse than the
original. Here is the quasi-formal expression of this restriction: Adding a
zero or deleting one from an n-tuple yields a (n + 1)-tuple or (n ’ 1)-tuple that
is equal to the original.
Restriction C (Generality): The proper principle for ordering worlds must
be insensitive to the particular numbers of ¬‚ecks of experience they contain, in the
sense that there isn™t a distinct normative principle for ordering distinct
groups of worlds where all members of a group are of the same size in this
respect but where different groups have different sized members. In other
words, we can™t deploy different sorts of ordering principles for different
sorts of cases, for different sized worlds. Note that this does not rule out
averaging principles, since n would not vary if we were given a particular
set of n-tuples of the same n to order. Why this constraint? We need a
general principle, which isn™t just a summary of different principles for
each size n. Nothing else could have a plausible general rationale. And it
is also plausible that the number of parts of a whole cannot in¬‚uence the
proper method of deriving the value of the whole from the value of those
parts. Some fans of organic unity (in other words, some fans of the thesis
that the value of the whole may depend crucially on the arrangement
and relations of parts that in other contexts would not contribute any
value, or at least not the same sort of value, to a whole) may ¬nd this
questionable, but we will return to that point in the next section. Here is

the quasi-formal expression of this restriction: The proper ordering of a set
of n-tuples of equal n must not be sensitive to the value of n.
Given that ¬‚ecks of phenomenal experience are the only value-bearing
elements of worlds, and given traditional conceptions of how the value
of a whole is a function of the value of its parts, three more restrictions
also seem obviously appropriate:
Restriction D (Value Responsiveness): If one world is better than a second,
then that is because some ¬‚eck of experience in the ¬rst is of greater hedonic value
than some ¬‚eck (or some ¬‚ecks) in the second.
In other words, a positive difference in the value of two whole worlds
must be a function of a positive difference in the hedonic value that is
contained in them. If one world A is better than another B that contains
the same number of ¬‚ecks of experience, then there is some pairing of
the ¬‚ecks of experience in A to the ¬‚ecks in B that meets two conditions:
(i) In it, A contributes the better member of one or more pairs, and
(ii) A is better because it contributes the better member or members
of that pair or those pairs. I gloss (ii) as follows: “A is better because
it contributes the better members of those pairs” means that if A had
contributed the worse members of those pairs while B and everything
else about A were unchanged, then A would have been worse than B.
The reasons for this restriction are straightforward. Only the ¬‚ecks of
phenomenal experience in worlds determine the value of those worlds,
but certainly a world does not get to be better than a second world because
it has a less valuable element or only equally valuable elements. It might
seem that there are other routes to being better, but we will return to
that issue in section II. There is a slightly stronger constraint than D that
would serve our purposes equally well and is easier to understand and
motivate: For any two worlds, if one is better than the other, then some
¬‚eck in the ¬rst world is of greater hedonic value than some ¬‚eck in the
second. But I will presume the weaker condition, since I can. Here is the
alternative quasi-formal formulation of Restriction D: If an n-tuple A is >
an n-tuple B, then there is some pairing of elements of A and B such that (i) A
contributes the greater member of one or more pairs and (ii) if A had contributed
the lesser member of that pair or those pairs while B remained unchanged and A
was otherwise unchanged, then A would be < B.
Restriction E (Weak Pareto): Making every ¬‚eck of experience better off in
regard to the hedonic value it presents would yield a better world. It would be
better if every ¬‚eck of experience had a higher level of hedonic value.
Making each and every ¬‚eck of experience in a world better would yield
a better world. The stronger Pareto principle that if some ¬‚ecks are better

off and none are worse off then the world is better is also plausible, but
we will see that it is implied by Restriction E in the context of our other
restrictions. Alternative formulation: For n-tuples A and B of equal n, if the
ith member of A is greater than that of B for all i, then A > B.
Restriction F (Separability): Adding to a pair of worlds (or deleting from
them) ¬‚ecks of phenomenal experience with equal hedonic values doesn™t affect the
relative value of the resulting wholes. An analogous condition in the economics
literature is known as the irrelevance of unaffected individuals. In other
words, if you remove from each of two worlds ¬‚ecks of experience that
are equal, then the ordering of the worlds that result preserves the order
of their predecessors. In other words: If A is > (respectively =, <) B, then
for all pairs of n-tuples A— , B— formed from the originals by the addition to (or
deletion from) each of a single element equal to that added to (or deleted from) the
other, then A— is > (respectively =, <) B— .
The next constraint is built into our characterization of the hedonic
value found in the world, so it isn™t really a new construction principle but
rather a familiar constraint set by the world and by the nature of hedonic
Restriction G (Strong Ordinality): A complete representation of hedonic
value must respect merely the three-part organization provided by positive, neg-
ative, and null value, and its merely ordinal nature. In other words, hedonic
value is merely ordinal (beyond the difference between positive, null, and
negative value), and any representation that captures that ordinal structure
is as good as any other. No principle for ordering worlds that depends
on privileging one such representation over another is legitimate. The
social choice literature calls such a condition the irrelevance of cardi-
nal information. This constraint constitutes the major difference between
the view developed here and classical utilitarianism, but it is required by
the kind of objective value found in feasible worlds. In our alternative
style of expression it becomes this: A correct principle for ordering worlds
must yield the same ordering of the worlds for any possible assignment of rationals
to hedonic value and disvalue levels meeting the restrictions on such assignments
noted earlier. The proper ordering principle must not be sensitive to dif-
ferences in assignments that are, given the ordinality of hedonic value,
The previous restrictions underdetermine an ordering of worlds, but
not by much. There is only one ordering that meets our former eight
restrictions and also holds that it would be better if a relatively better ¬‚eck
were worse off to the gain of a still worse ¬‚eck in at least one case of each
of three sorts “ involving all positively valued, all negatively valued, or

mixed ¬‚ecks. I will assume, as the ¬nal constraint on our construction,
the very weak equity assumption that picks out this ordering.
Restriction H (Weak Equity): Consider pairs of worlds that cannot be
ordered given our previous constraints. For at least one such pair of worlds
that contain only ¬‚ecks of experience with negative hedonic value, and for at least
one such pair of worlds that contain only ¬‚ecks of experience with positive hedonic
value, and for at least one such pair of worlds each of which contains some ¬‚ecks
with negative value and some with positive value, the better world of the pair is the
world whose worse ¬‚eck is better than the worse ¬‚eck in the other. In other words,
it would sometimes be better to make a worse ¬‚eck in a world better off
at the expense of a relatively better-off ¬‚eck in at least one case of each of
these three natural types. This constraint has an unintuitively complicated
form, but that is because it is very weak and speci¬cally addressed to
our situation, which includes the differences among null, positive, and
negative value. We will return to the legitimation of this constraint in the
next section. In quasi-formal form for two-¬‚eck worlds it is: For some x,
y, w, z such that x < y < w < z < 0, <x,z> > <y,w>, and for some x, y,
w, z such that 0 < x < y < w < z, <x,z> > <y,w>, and for some x, y, w,
z such that x < y < 0 < w < z, <x,z> > <y,w>.


This section legitimates the construction principles introduced in the
last section. Normative evaluation is justi¬catory reason giving, so the
normative principle to which the construction principles will lead in
section III must be vindicated relative to possible competitors. This in
turn requires a reasoned vindication of the construction principles that
support it.
Restriction 0 (Completeness) is set by the question we seek to answer. If
a complete ordering is unavailable, we might hope to act at least sometimes
on the basis of a partial ordering of worlds. But the rest of our constraints
do deliver a single complete ordering. And it is a practical desideratum
that this constraint obtain.
Restriction A (Abstraction) allows us to characterize feasible worlds
for purposes of basic normative evaluation simply by reference to the
number of ¬‚ecks of experience of each level of hedonic value and disvalue
they contain, by their number of instances of each distinguishable basic
normative property.
Hedonic value and disvalue are the only basic normative properties that
are instanced in feasible worlds. So a characterization of a world by those

instances is its characterization by all the elements in it that can plausibly
be thought to constitute its basic normative status. So the legitimation of
Restriction A is quite straightforward.
Given that the ordinal hedonic values of ¬‚ecks of phenomenal expe-
rience are the only unconstituted normative elements in feasible worlds,
they are the only things relevant to the normative value of those wholes.
Or, to put it more precisely, they constitute everything about such worlds
that can be so relevant. What else could be relevant to the value of a whole
but all its basic value-bearing elements? To be still more precise, what else
could be relevant to the value of a whole but all its basic value-bearing
elements when there are no unconstituted normative elements of other
sorts that might somehow help to constitute its overall value?
Any practice of justi¬catory reason giving must ultimately vindicate
judgments about normative status by appeal to objective and basic nor-
mative phenomena, as we saw in Part Two. But we presume here that the
ordinal hedonic values and disvalues of ¬‚ecks of experience constitute all
the objective and basic normative phenomena that there are. If anything
else were treated as relevant to the value of a whole feasible world, then it
would be a feature whose relevance could not be legitimated. A practice
of so-called justi¬catory reason giving that treated as relevant to the value
of a whole elements of that sort would not be similar enough to genuine
and coherent justi¬catory reason giving to be properly worthy of that
name. It would deploy as legitimations of certain evaluations phenomena
that are, we know from the last part, inappropriate to ground such justi¬-
cations. It would not be in the full sense evaluation as justi¬catory reason
giving. It would suffer from a kind of incoherence, treating as relevant to
value what at the same time it implied is not relevant.


. 6
( 10)