smith, a. d. 2010, disjunctivism and illusion
Post on 15-May-2017
215 Views
Preview:
TRANSCRIPT
Disjunctivism and Illusion
a. d. smith
Warwick University
Disjunctivism is a theory about sensory experience that, as its very
name implies, invokes a certain contrast: between being one thing or
another (and not both). The theory claims that sensory experience
comes in certain fundamentally different and incompatible psychologi-
cal forms. Everyone would accept that, of course. What is distinctive
about disjunctivism is its claim that there are such differences, even
intrinsic ones, that the experiencing subject is necessarily not in a posi-
tion to appreciate simply in virtue of having the experiences and reflect-
ing on, or ‘‘introspecting’’, them, even though the experiences are fully
conscious ones. The contrasting kinds of experience that the disjunctiv-
ist is concerned with are those that intuitively have different cognitive
and epistemological credentials. At one extreme there is veridical
perception. Here one is perceptually aware of a real object in the
world, the object appears just as it really is, and one is thereby, at least
usually, in a position to have perceptual knowledge of this object. This
is, cognitively and epistemologically, a ‘‘good’’ case. At the other
extreme, there is hallucination. Here one is not aware of any real object
in the world at all, and the experience can furnish one with no percep-
tual knowledge of that world. This is a ‘‘bad’’ case. A hallucination
might, however, be subjectively indiscriminable from some veridical
perception, in that one could hallucinate an object of just the same per-
ceptible sort as one might veridically perceive, and the hallucination
might be so lifelike that one could not tell, simply on the basis of
‘‘what it is like’’ to have the experience, that one was not veridically
perceiving such an object. Nevertheless, claims the disjunctivist, these
two sorts of experiences, even considered just as experiences, are intrin-
sically different in nature: so different, indeed, that they should be
regarded as falling into two fundamentally different psychological
kinds. The kind of experience one has when one veridically perceives
384 A. D. SMITH
Philosophy and Phenomenological ResearchVol. LXXX No. 2, May 2010� 2010 Philosophy and Phenomenological Research, LLC
Philosophy andPhenomenological Research
some object in the world is one that simply cannot be had by a subject
who is not in perceptual contact with some such object. This kind of
experience is essentially ‘‘object-involving’’, and so is a kind of experi-
ence that one cannot have hallucinatorily.1 Mere subjective indiscrimi-
nability, the disjunctivist claims, should not be used as a criterion for
determining psychological nature in such cases. There is, to be sure,
something that reflective subjects of either perceptual or hallucinatory
experiences can definitely know about their conscious life just on the
basis of reflecting on their experiences. They can, for example, defi-
nitely know that it at least seems to them that they are perceiving
something of a certain sort. In order, however, to emphasise the fact
that two fundamentally different kinds of experiential state can be the
basis of such knowledge, disjunctivists prefer to express the knowledge
in question as being of a disjunctive form: the subjects know that they
are either perceiving something or that they are merely under the
impression that they are.2 We may call such a formulation an ‘‘experi-
ential disjunction’’: one that disjoins what are commonly referred to as
a ‘‘good’’ and a ‘‘bad’’ disjunct.
In this paper I address the question of what account disjunctivism
can or should give of illusion. This is a surprisingly under-discussed
topic. Although most disjunctivists do briefly mention illusion, and
sometimes commit themselves on the question of whether it should be
regarded as falling under the good or the bad disjunct of the experien-
tial disjunction, almost all their detailed discussions of bad cases con-
cern hallucination. Perhaps they sense that illusion may harbour
difficulties for the view. Be this as it may, I shall argue that the phenom-
enon of illusion, at least of certain kinds, indeed cannot be adequately
accommodated by one influential form of disjunctivism. Having estab-
lished that, I shall explore alternative versions of disjunctivism that are
not simply refuted by the facts, and ask whether illusion should be allo-
cated to the good or the bad disjunct. In as much as illusory perceptions
are indeed perceptions—perceptual contact is made with some actual
object in the environment—they share a good feature with veridical per-
ceptions. In as much as they are illusory, however, they are in some
sense ‘‘bad’’. No doubt because of this, disjunctivists themselves are
1 Those writing in this tradition treat objects as entities. In other words, non-existent
intentional objects are not ‘‘countenanced’’. For the purposes of this paper I shall
fall in with this way of thinking.2 I have argued elsewhere (Smith, 2008) that neither the notion of being under the
impression that one is perceiving, when one isn’t, nor its variants that can be found
in the disjunctivist literature (such as being subjectively indiscriminable from a per-
ception), is adequate to pick out the class of hallucinatory experiences. For present
purposes, however, I shall leave aside this somewhat important issue.
DISJUNCTIVISM AND ILLUSION 385
divided over the issue of where to place illusions in the experiential dis-
junction—as Paul Snowdon, himself a leading disjunctivist, recognises:
‘‘In one account (that embodied in the present discussion), the disjuncts
are perception and hallucinations. Into the perception disjunct would go
both accurate and inaccurate perceptions. The other accounts form a
disjunction between accurate perception on the one side and, on the
other, all non-accurate cases. These are the disjunctions to which Hin-
ton . . . and McDowell have attended’’ (1990, 131). For reasons that we
shall see, and as Snowdon himself later indicates, the case of McDowell
is not straightforward, in that his form of disjunctivism is not directly
concerned with the nature of perceptual experience. Hinton, however, is
explicitly concerned with this issue, and he does make it clear that his
bad disjunct includes both illusion and hallucination (Hinton, 1973,
114–121). Another leading disjunctivist who regards both illusion and
hallucination as constituting a bad case is M. G. F. Martin. His contrast
is always between veridical perceptions on the one hand, and states that
are merely subjectively indiscriminable from them on the other.3 Indeed,
Snowdon is in a small minority of disjunctivists in allocating illusion to
the good disjunct. I shall argue that he is right to do so.
It may not be immediately clear, to either insiders or outsiders,
whether these different decisions concerning where to place illusion in
the experiential disjunction reveal a philosophical disagreement; or, if
they do, how significant it is. One might, for example, wonder why dis-
junctivists should be required to operate with just a twofold disjunc-
tion. If the experiential disjunction were presented in terms of veridical
perception or illusory perception or hallucination, perhaps the present
issue would simply evaporate. Even someone who does work with the
simple twofold disjunction, and who places illusions along with halluci-
nations in the bad disjunct, need not deny that illusions and hallucina-
tions differ significantly, perhaps even fundamentally and essentially,
from one another. Indeed, Snowdon himself suggests that the different
formulations of the experiential disjunction ‘‘do not necessarily repre-
sent theories between which we have to choose’’ (1990, 132). Since illu-
sions seem to have both good and bad aspects, perhaps for some
philosophical purposes it is worth stressing their good aspect and asso-
ciating them with veridical perceptions, whereas for other purposes it is
worth underlining their bad aspect and allying them with hallucina-
tions. There is, however, a substantial philosophical issue involved here
on which one does have to take a stand. It is the question whether in
both illusions and hallucinations we find a kind of experience, or
3 Martin adopts disjunctivism as a way of upholding what he terms ‘‘naıve realism’’:
a position that, as we shall see, requires illusions to be placed in the bad disjunct.
386 A. D. SMITH
intrinsic feature of experience, that is necessarily not to be found in any
veridical, non-illusory perceptual experience, or conversely.4 Anyone
who, like Hinton or Martin, places illusion with hallucination in the
bad disjunct clearly thinks that this is so. When Snowdon formulates
his own perception ⁄hallucination disjunction, he may not be expressing
disagreement with this. He may simply be interested in a different issue.
It is, however, this issue that I am concerned with in this paper. When,
therefore, I refer to a form of disjunctivism that endorses the percep-
tion ⁄hallucination disjunction, and thereby allocates illusion to the
good disjunct, I mean a theory that positively claims that there is noth-
ing to illusory experiences, qua experiences, that cannot be found in
veridical experiences.5 To place illusions together with veridical percep-
tions in the good disjunct is, therefore, to commit oneself to the view
that they have, to use John McDowell’s phrase, a ‘‘highest common
factor’’ (McDowell 1982, 472). I take this phrase to express the idea
that types of experience that have such a highest common factor are of
identical psychological kinds, differing at most only in external, rela-
tional matters. As mentioned, I shall argue that such a view is the cor-
rect one for a disjunctivist to hold. Having established that, I shall
conclude by considering how significant such a ‘‘concession’’ may be.
I
In this section I argue that a certain currently influential form of disjunc-
tivism is incapable of giving an adequate account of illusion. All disjunc-
tivists agree that a veridical perceptual experience is of a psychological
type that cannot be had unless one is perceiving some item in the real
world. Most disjunctivists go beyond this, and give a particular explana-
tion of this impossibility. When some worldly object is veridically per-
ceived, this common view has it, the object is a constituent of the
experience. Paul Snowdon goes so far as to make such constituency an
4 For reasons that we shall examine later, ‘‘veridical, non-illusory’’ is not pleonastic.
Since, until then, the issues involved are not pertinent to the discussion, I shall
sometimes write ‘‘veridical’’ where ‘‘veridical and non-illusory’’ would strictly be
accurate.5 Martin (2002, 395 n24) contrasts a view that simply allocates illusions to the bad
disjunct with one that simply allocates them to the good one, and suggests that his
own position is ‘‘somewhere in between these two approaches’’. This is because
Martin construes the terms ‘‘illusory’’ and ‘‘veridical’’—quite correctly, as we shall
see—as applying primarily to the perception of features of objects, rather than of
objects as such. Given the way I have just set up the issue that is to concern us,
however, he is to be placed squarely in the camp of those who allocate illusion to
the bad disjunct. For, as we shall soon see, he certainly supposes there to be some
feature of veridical experience that cannot be found in any illusory experience in so
far as it is illusory.
DISJUNCTIVISM AND ILLUSION 387
essential part of disjunctivism as such. Disjunctivism is, he writes, ‘‘not
exhausted . . . by the simple denial of a common nature’’ to bad and good
experiences. It ‘‘involves also the characterisation of the difference
between the perceptual and non-perceptual in terms of the different con-
stituents of the experiences involved. The experience in a perceptual case
in its nature reaches out to and involves the perceived external object,
not so the experience in other cases’’ (2005, 136–7). The specific form of
disjunctivism that we shall be considering in this section interprets such
constituency as accounting not only for the fact that an experience makes
perceptual contact with the world; it accounts also for the phenomenal
character of the experience, at least in part. Martin, who subscribes to
this view, calls it ‘‘naıve realism’’, and I shall follow him in this. He speci-
fies it as follows: ‘‘According to naıve realism, the actual objects of per-
ception, the external things such as trees, tables and rainbows, which one
can perceive, and the properties which they can manifest to one when
perceived, partly constitute one’s conscious experience, and hence deter-
mine the phenomenal character of one’s experience. This talk of constitu-
tion and determination should be taken literally’’ (1997, 83, my
emphasis).6 For an object to be a constituent of an experience, therefore,
is for it to constitute and determine the phenomenal character of the
experience, at least in part. The object does this by, as it were, importing
its very own features into experience. When, for example, I veridically see
a green square, the ‘‘phenomenal greenness’’ and ‘‘phenomenal square-
ness’’ of my experience are nothing more than the physical greenness and
shape of that square featuring in my experience by way of constituency.
The reason why illusion poses a problem for such a view is that illu-
sions can be, indeed almost always are, partial. Consider, for example, a
case where a green square looks yellow to me, though it does look square.
Since I am really seeing the square, and veridically perceiving its shape,
the square and its shape are presumably constituents of my experience.7
This square’s shape furnishes the phenomenal character of the illusory
perception in some respect: in the respect of being as of a square. What,
however, about the apparent yellowness? It, clearly, must be accounted
for by something other than the green square being a constituent of the
experience, since this is the case when a green square veridically looks
green to me. Some extra, ‘‘bad’’ factor, over and above the green square’s
being a constituent, must, therefore, be attributed to this partially illu-
sory state to account for the illusorily appearing colour: something that
6 ‘‘Determines’’ implies ‘‘is sufficient for’’, as Martin makes clear: ‘‘Difference in
presented elements between two experiences will be sufficient for difference in their
phenomenal properties’’ (Martin 1998, 174).7 We shall consider later how matters stand if this is denied.
388 A. D. SMITH
is absent in the case of completely veridical perception, where the constit-
uent object itself does all the work. A naıve realist about veridical percep-
tion must, it seems, embrace a ‘‘mixed’’ account of such partial illusions,
constituency accounting for the respects in which the experience is veridi-
cal, and some extra factor accounting for the respects in which it is
illusory.8 There are a number of candidates for such a bad factor: some
representational feature, or a quale, for instance. But whatever is postu-
lated will, in certain cases, undermine the naıve realist account of the
veridical perception of an object’s features. This is because, to continue
with our example, a certain square can, through constituency, determine
an aspect of the phenomenal character of a visual experience—making it
such that something looks square—only in virtue of being seen. A visible
shape, however, necessarily has some colour; and only because an
object’s colour is seen is an object’s shape seen.9 Indeed, the apparent
shape of an object, as far as vision is concerned, just is a matter of the dis-
tribution of apparent colour. An object is visibly square to me because
and only because of how colour is distributed in my visual field.10 Because
of this, a square’s being a constituent of my experience cannot account
for why a square seems to be visually present to me unless it also
accounts for the appearance of colour. In the case in hand, the appar-
ently distributed colour is yellow; but the square is actually green. How
could a green square’s being a constituent of my experience account for
the apparent yellowness of the square? By itself, it clearly cannot. So, by
itself it cannot account even for the appearance of a square, since this
depends on apparent colour and how it is distributed. To account for the
distribution of the apparent colour is to account for the appearance of
the shape. If something other than the real square accounts for the
appearance of the colour, something other than the real square (and its
constituency) accounts for the very appearance of the square. Generally,
whatever bad factor is postulated to account for an illusorily appearing
feature in a partially illusory perceptual experience will undermine the
naıve realist account of any veridically appearing aspect whose appearing
presupposes and is fully determined by the appearing of such a feature.
It will not do to respond to this argument by claiming that, although
something other than the square accounts for which colour appears, it
8 Martin (2002, 395 n24) seems to commit himself to this account of the matter.9 What, it may be asked, about glints and gleams? One possible answer is to extend
the term ‘‘colour’’ so that it covers such phenomena. Another, which I favour, is
to deny that in seeing such a glint or gleam you are seeing a part of the object or
its surface. What you see, in these cases, is light reflected from the object.10 Things, at least usually, visually appear as located and variously oriented in three-
dimensional space. The distribution of colour of which I speak is not in some
merely two-dimensional ‘‘visual array’’.
DISJUNCTIVISM AND ILLUSION 389
could be the square that accounts for how this colour is specifically dis-
tributed. The only way, on the naıve realist account, in which the square
can do this, given that the shape is veridically perceived, is by being a
constituent of the experience. It cannot, however, be a constituent with-
out being seen; and it cannot be seen without some particular colour
appearing. According to the present response, however, an object’s being
a constituent of a visual experience does not suffice for any particular col-
our to appear. A square, through being a constituent of an experience,
cannot account for the appearance of just some colour or other. To sup-
pose otherwise is to treat constituency as a weaker notion than it in fact
is. The constituency view is not merely that an object is responsible for
the phenomenal character of an experience when the latter is veridical. It
is, rather, responsible for this character specifically in virtue of importing
its own character into the experience through constituency. It must,
therefore, import everything that is necessary for its constituency, and
hence for its veridically perceived features to be seen. This is just what
does not happen in illusory cases of the sort we are considering.
An alternative way for the naıve realist to account for such illusory
perceptions would be to deny that any extra factor is responsible for the
illusorily appearing colour. This would apply to the case in hand some-
thing like the account that Martin offers of certain hallucinations.11
According to that account, the only thing that can be said about the phe-
nomenal character of these hallucinations is that they are not subjectively
discriminable from some kind of veridical perception: their ‘‘only positive
mental characteristics are negative epistemological ones—that they can-
not be told apart by the subject from veridical perception’’ (Martin,
2004, 73–4). Applied to the illusory case of a green square looking yellow,
one would say that the square’s being a constituent of the experience
accounts for a square appearance, and that its looking specifically yellow
is simply a matter of the experience’s negative epistemological character-
istic of not being subjectively distinguishable from veridically seeing
something yellow, or something along these lines. Since no positive fea-
ture is postulated to account for the apparent yellowness, there is, it may
be thought, nothing to conflict with or undermine the real square’s actual
shape accounting for the apparent squareness through constituency. This
will not help matters, however, since it still remains impossible for the
square’s actual shape to appear at all in a way compatible with naıve real-
ism. It can appear only in virtue of some colour appearing as distributed
in a particular way, and one must account for the latter in order to
account for the former. The veridical appearance of the square is
11 The hallucinations in question are those that have the same kind of proximal cause
as some (possible) veridical perception.
390 A. D. SMITH
certainly not, however, for the naıve realist, simply a negative epistemo-
logical matter, as is, supposedly, the appearance of colour.
II
Bill Brewer has recently offered an analysis of illusory perception that,
if acceptable, would block the preceding argument against the naıve
realist form of disjunctivism. Brewer has given up his earlier representa-
tionalist account of perceptual consciousness, and now holds that per-
ceived objects not only are constituents of perceptual experience, but
that they ‘‘constitute the subjective character of perceptual experience’’
(2006, 168). Moreover, he offers an account of how this can be true in
the case of illusion. His analysis postulates ‘‘two levels in the subjective
character of experience’’. At the first level, ‘‘the mind-independent
direct object itself, just as it actually is, . . . is constitutive of this sub-
jective character’’. The second level consists of the way in which an
object presented at the first level ‘‘may mistakenly be perceptually
taken’’ (2006, 172). What is significant about Brewer’s proposal is that
error enters into an illusory experience only at this second level: it is to
be attributed solely to the subject’s response to what is perceptually
presented. ‘‘In perceptual experience,’’ he writes, ‘‘a person is simply
presented with the actual constituents of the physical world themselves.
Any errors in her worldview which result are products of the subject’s
responses to this experience, however automatic, natural or understand-
able in retrospect these responses may be. Error, strictly speaking,
given how the world actually is, is never an essential feature of experi-
ence itself’’ (2006, 169). It should be clear how naıve realism is
sustained by this account of illusion. Objects in the world can be con-
stituents of illusory experiences, and determinative of their phenomenal
character, since how they are in such experiences, at least at the first
level, is how they actually are. Any mismatch occurs only when it
comes to our response to our (first level) experiences. In a later work
Brewer is not so explicit about the two-level nature of the account; and
yet it is still in place. He still postulates that ‘‘the core subjective char-
acter of perceptual experience is given simply by citing the physical
object which is its mind-independent direct object’’ (2008, 171). In cases
of illusion, however, this physical object, given a certain point of view
and ‘‘circumstances of perception’’, bears certain visually relevant simi-
larities to a different kind of object; and these similarities may ‘‘intelli-
gibly lead’’ one to take the object as being of this other kind. It is such
similarities and the way they provide an ‘‘intelligible ground’’ for error
that constitute the illusory ‘‘look’’ of the thing (2008, 171–2 and 178).
Such a look is, however, still regarded as something over and above a
DISJUNCTIVISM AND ILLUSION 391
‘‘more basic phenomenal presentation’’ of the object that constitutes
the ‘‘core’’ subjective character of the experience (2008, 178).
Brewer illustrates how his account is meant to work by applying it to
the case of the Muller-Lyer illusion. Brewer suggests that the two objec-
tively equal lines in the Muller-Lyer figure are present as equal lines in
our core experience when we view the figure, so that the basic phenome-
nal character of our experience is indeed as of equal lines. Nevertheless,
these lines ‘‘have the power to mislead us, in virtue of their perceptually
relevant similarities with other things’’. For these lines bear a similarity
to two unequal lines, one of which is nearer to the viewer than the other
so that they have the same projective length in the plane in which the
Muller-Lyer figure is located. The inward and outward pointing wings
at the ends of the two lines function to make this similarity ‘‘salient’’,
and bring two unequal lines ‘‘to mind’’ (2006, 168–9).
Such a two-level account is not implausible in relation to some illu-
sions: those that psychologists call ‘‘cognitive illusions’’, and of which
the Muller-Lyer illusion is one. Brewer claims, however, that his account
can be applied to ‘‘many of the most standard cases of visual illusion’’
(2008, 173). Unless it applies to all types of illusion, naıve realism will
still be inadequate, but in fact it does not apply to any illusions at all
other than the ‘‘cognitive’’ ones: to any, that is to say, that are more
physiologically or physically based. It does not, for instance, to continue
with our example of colour, apply to the illusorily appearing colours
that are due either to unusual lighting or to simultaneous or successive
colour contrast. In such cases it is one’s sensory experience itself, at the
most basic level, that is affected, not just some response to the experi-
ence. Because of this, there are in such cases no visually relevant similari-
ties of the sort that Brewer can point to in the Muller-Lyer illusion. A
green square can look yellow to me because of the peculiar lighting.12
12 A naıve realist may suggest that when, as most of us would say, illusion is caused
by the behaviour of light, or by other objective features of the environment—as in
the present example, and in the familiar example of the straight stick half immersed
in water—there is really no illusion at all, since we are accurately picking up on
real features in the ‘‘external world’’. The green square is, it may be suggested, yel-
low-in-this-light, and the stick is, as we might say, ‘‘optically bent’’. (I am grateful
to an anonymous referee of this journal for raising this objection.) I reject this sug-
gestion, because it is incapable of recognising that, in the situations in question, we
get anything wrong at all about our environment. (The error cannot be attributed
to our taking the situation to be in some sense ‘‘standard’’ when it is not, since we
need not so take it: we may be perfectly well aware of the situation for what it is.)
This inadequacy is grounded on the fact that when I, for example, see a stick, it is
the shape of the stick that seems apparent to me—not its ‘‘optical shape’’. In any
case, there are many kinds of wholly internally caused, physiologically based illu-
sions—such as those due to simultaneous or successive colour contrast— in relation
to which this naıve realist proposal is a complete non-starter.
392 A. D. SMITH
There is, however, no ‘‘relevant’’ similarity at all between green and
yellow. They are simply two quite different colours. In fact, when Brewer
explicitly considers non-cognitive illusions, he is forced to make it clear
that his visually relevant similarities cover things of which the subject
may be totally unaware. ‘‘Two objects have visually relevant similari-
ties,’’ he writes, ‘‘when they share sufficiently many properties amongst
those which have a significant involvement in the physical processes
underlying vision. Thus, and very crudely, visually relevant similarities
are identities in such things as, the way in which light is reflected and
transmitted from the objects in question, and the way in which stimuli
are handled by the visual system, given its evolutionary history and our
shared training during development’’ (2008, 172). It should be evident,
however, that such similarities as these, wholly unknown to the experi-
encing subject as they typically are, can play no role in an account of
illusion that locates error only in an ‘‘intelligible’’ response to a core
experience that perfectly matches the actual situation perceived. When I
see the yellow-looking square, there is no sense at all in which I see the
light travelling to my eyes. The light does not look any way at all to
me.13 It is the square that I see and that looks a certain way. And in
cases of colour contrast, there is not even a relevant similarity in any-
thing external to the observer. According to Brewer, illusion consists in
the fact that visually relevant similarities ‘‘may intelligibly be taken for
qualitative identities’’ (2008, 173). But nothing I am wholly unaware of
can be ‘‘taken’’ by me in any way whatever. Are we really to suppose
that a white object’s looking green to me after I have been staring at a
red surface is a matter of the state of my retinal cells providing an ‘‘intel-
ligible ground’’ for a response that brings green ‘‘to mind’’? There is no
intelligibility here at all. It is just a matter of psycho-physical causation.
The ‘‘visually relevant similarities’’ in all such non-cognitive illusions
simply concern the processes that give rise to experience. They are oper-
ative before any experience occurs, even at a ‘‘first’’ level, and they
condition the phenomenal character of that experience.
Moreover, the suggestion that, in illusion, the perceived object is
accurately registered phenomenally in a first-level experience has
dropped out of Brewer’s account when it is applied to non-cognitive
illusions, or is at least playing no explanatory role. This is not surpris-
ing, since the suggestion that, in the case of non-cognitive illusions, the
actual character of the perceived object phenomenally characterises the
experience (in the respect in which it is illusory) is wholly indefensible.
In the case of the Muller-Lyer illusion, according to Brewer, an equal-
ity of the lines characterises one’s core experience—an experience that
13 Glints and gleams aside.
DISJUNCTIVISM AND ILLUSION 393
constitutes a ‘‘more basic phenomenal presentation’’ of the object than
is to be found at the second level, where alone error inters into the pic-
ture (2008, 178)). When a green object looks yellow because of the
lighting, however, greenness does not phenomenally characterise my
experience in any way at all. It is not, in such a case, simply that my
response involves an (intelligible) error. The response is just what it
should be to a core experience that is qualitatively identical to the one
I get when a yellow object looks yellow. In order to justify his claim
that equality in length does indeed characterise the core visual experi-
ence when the Muller-Lyer figure is seen, Brewer points out that the
lines in the figure would not appear to change length if the attached
wings were to shrink to vanishing point (2006, 170). This perhaps sug-
gests that there is a certain phenomenal sameness to the lines with and
without the wings; and that since the lines without wings are certainly
phenomenally equal, so are the lines in the actual Muller-Lyer figure
with their wings. Nothing analogous to this is to be found when light-
ing changes, or when a colour becomes surrounded by another colour
that gives rise to simultaneous colour contrast. In order further to jus-
tify his claim that phenomenal equality is present in the ‘‘core’’ experi-
ence of the Muller-Lyer figure, Brewer suggests that if you were to
point to where the ends of the two lines in the figure appear to be, you
will point to their actual location—which also perhaps suggests that
equality is in some way phenomenally manifest to you. Again, nothing
analogous is to be found in the case of non-cognitive illusions. When
that green square looks yellow to me, all my non-inferential responses
are those appropriate to yellow. Or, to take a non-cognitive illusion
that is closer to the Muller-Lyer illusion in that it involves spatial prop-
erties, consider the straight stick that looks bent when half submerged
in water. Here, when I point to the ends of the stick on the basis of
how it appears to me, I will not point to their actual location.
The following sort of situation also shows Brewer’s proposal to be
unworkable as a general account of illusion. Consider a situation where
you view two equal lines through distorting lenses, so that just one of
the lines looks a little longer than it is. Choose the right distortion, and
when the Muller-Lyer illusion is exemplified by adding suitable attach-
ments to the ends of the lines, the two lines will look equal with their
attachments.14 Since the attachments function the same way in this case
as in the usual Muller-Lyer case, Brewer’s two level analysis must rec-
ognise a phenomenal difference in the length of the two lines, due to
the non-cognitive illusory effect produced by the distorting lenses, at
14 This would be a case of ‘‘veridical illusion’’: a phenomenon to which we shall
attend later.
394 A. D. SMITH
the first level of experience—something that his account is meant to
rule out.
The case against a two-level account of any non-cognitive illusion is
overwhelming. At least wherever non-cognitive illusions are concerned,
the actual character of the perceived object is not, in so far as the expe-
rience is illusory, phenomenally manifest to the subject at all, at any
‘‘level’’ of experience. All such illusions therefore stand in the way of
defending naıve realism against the arguments of the previous section.
III
The only way for the naıve realist form of disjunctivism to avoid the
problems detailed above is to claim that illusory perceptual experiences,
even when they contain veridical elements, are wholly different in kind
from completely veridical perceptual experiences. The ‘‘mixed’’ account
of partial illusion we have been considering is rejected, and it is denied
that the veridical aspects of a partially illusory appearances are to be
accounted for by constituency. The question now arises, of course, how
illusions are to be distinguished from hallucinations; but that will pose
no serious problem in principle. One possibility—in fact, as far as I can
see, the only possibility consistent with the direct realism that naıve real-
ism is intent upon defending—is for the naıve realist to conceive of both
illusions and hallucinations as representational states, with no worldly
objects as constituents, but to distinguish between hallucinations and illu-
sions on the basis of a distinction between different kinds of representa-
tional content: by, for example, denying to hallucinations, but attributing
to illusory perceptual states, object-dependent representational content.15
Such an ‘‘unmixed’’ account is, however, at odds with the funda-
mental motivation for naıve realism, and, indeed, with its literal formu-
lation. As Martin has stressed on a number of occasions, naıve realism
is adopted by its proponents as being ‘‘the best articulation of how our
experiences strike us as being to introspective reflection on them’’
(2004, 42). By contrast, both sense-datum and representationalist
accounts amount to ‘‘error-theories of sense experience’’ (2004, 84).
The naıve realist is happy to convict us of error when we hallucinate,
for then we are merely under the mistaken impression that we are
really perceiving things. But the naıve realist who endorses an
‘‘unmixed’’ account of partially illusory experience must convict us of
error even when we are veridically perceiving some aspect of the sur-
rounding world. When I misperceive that green square as yellow, I see
the object’s shape just as it is. And yet my entire visual perception of
15 More on this notion below.
DISJUNCTIVISM AND ILLUSION 395
the square has, on the present unmixed account, a fundamentally dif-
ferent nature from that of any wholly veridical perception. Anything
other than a naıve realist, ‘‘constituency’’ account of perception is,
however, supposed to constitute an error-theory. ‘‘We should,’’ writes
Martin, ‘‘be moved to this position [sc. naıve realism] in defence of a
natural conception of how our veridical perceptual experience relates
us to the world around us. That is what leads us to Naıve Realism’’
(2004, 84). If so, we should be equally moved when it comes to the
veridical aspects of partially illusory perceptions. The naıve realist
claim that veridically perceived objects are constituents of perceptual
experiences is straightforwardly falsified by a case such as the yellow-
looking green square. The square is veridically perceived as a square,
but cannot, on the unmixed account, be a constituent of the perceiver’s
visual experience, because of its illusory yellowness. And yet, when I
see that green square, I see its actual shape. I see, as Wilfrid Sellars
would have put it, its very squareness. That the square looks yellow to
me in no way conflicts with this fact.
More significantly, a naıve realist account that embraces an
unmixed, representationalist account of illusion is objectively untenable
because of a variant of the following argument. Suppose, again, that I
am misperceiving that green square as yellow (though as square). On
the present account, this is a representational state that does not have
the square as a constituent. Suppose, now, that because of some change
in the situation, my representational state becomes wholly accurate.
The naıve realist surely cannot deny that this is possible. In such a situ-
ation, however, I cease to misperceive the square’s colour. Since I am
now perceiving the square in a wholly veridical way, it should now be,
for the naıve realist, a constituent of my visual experience. But the only
change I have postulated in the original experiential state is one that
concerns that state’s intentional content and its correctness. The state
remains a representational state, not one that now has physical objects
as constituents. This, however, is impossible if naıve realism is true,
and at least all wholly veridical perceptions involve the perceived object
as a constituent of experience. Such an intentionalist account will
undercut the entire naıve realist position.
I said that it is a variant of the foregoing argument that will invali-
date a naıve realist unmixed account of illusion. A variation is
required, because the argument lacks cogency as it stands. This is
because it presupposes that a wholly veridical intentional perceptual
state will necessarily involve no illusion; and this is false, since illusions
are not necessarily non-veridical in the sense of misrepresenting the
nature of the object perceived. An experience can arise, for example,
from two independently operative illusion-inducing factors—abnormal
396 A. D. SMITH
lighting and lateral inhibition or retinal fatigue, say—that precisely off-
set one another, so that how something ends up appearing is the way it
actually is.16 Some readers may not be inclined to regard this sort of
case as one of illusion at all—so tight may be the connection between
illusion and misrepresentation in some people’s minds. In fact, whether
or not such a case is properly termed an ‘‘illusion’’ is not a matter of
great philosophical importance. Everyone will, I assume, at least agree
that such a case is not a wholly optimal case of perception. One indica-
tion that it is indeed less than perceptually optimal is that we should
deny that the subject who sees a green square that looks green to him
because of two illusion-inducing factors that conspire to produce this
effect is in a position either to know that or to see that the square is
green.17 The important point, now, is that the naıve realist will not
unreasonably apply the constituency account of the phenomenal char-
acter of experience only to ‘‘optimal’’ cases of perception. If, therefore,
intentionalism is to undercut the present ‘‘unmixed’’ version of naıve
realism, it must be able to offer an acceptable account of such
‘‘optimality’’—one that involves more than mere correctness of percep-
tual representation. I now indicate how this may be done.
The reason we judge the veridical cases now under consideration to
be less than optimal cases of perception is that the veridicality of the
experiences in question is fortuitous, because of the nature of the visual
processes involved. Given just this shade of green, and just these sur-
rounding colours producing a retinal effect that gives rise to the phe-
nomenon of simultaneous colour contrast, and just this abnormal light,
the green thing looks green; but this type of situation does not reliably
give rise to accurate perception.18 This fact certainly supports our
judgement that a subject in such a situation would not know that the
object before him is indeed green. Moreover, it supports our judgement
that such a subject would not see that the object is green.19 This is not
16 Johnston (2006, 271–4) has already argued for this. The illusions he considers,
however, all involve at least one arguably ‘‘cognitive’’ illusion. Although this in no
way affects the general point, as far as the present argument is concerned it is, for
reasons we have noted earlier in connection with Brewer, better to avoid reliance
on illusions of this kind.17 If you do not think that the sort of case now in question is even less than optimal,
then you do not think that such a case harbours any problem for intentionalism as
it stands; and nor, therefore, for the suggestion that intentionalism ‘‘undercuts’’
naıve realism.18 That these two factors merely happen to conspire to produce accurate perception is
fundamental to our judgement on such cases, for we can imagine a sensory system
that works optimally, and as a matter of course, by employing two sub-systems
that reliably off-set one another in this fashion.19 Mark Johnston (2006, 271–6) has emphasised the need for any adequate account
of ‘‘veridical illusion’’ to account for this fact.
DISJUNCTIVISM AND ILLUSION 397
because seeing that entails (visually) knowing that—so that accounting
for the absence of the latter would ipso facto account for the absence
of the former. There are two sorts of situation that render such an
entailment at least dubious. First, you can clearly see that something is
the case, and yet not believe your eyes, because of some (false) collat-
eral belief that you have. One could attempt to preserve the link
between seeing that and knowing that by severing the link between
knowing and believing; but that would itself be contestable. The second
example, however, is independent of this issue. For you can clearly see
that something is the case, and correctly believe it is, even if you are in
the middle of a period when you are having spontaneous hallucinations
or illusions. Since it is a real possibility that you are hallucinating or
having an illusion now, though in fact you are not, you do not know
to be so what you see to be so. It is not, therefore, that subjects of the
sort of non-optimal visual cases now under consideration fail to see
that things are as they appear to be simply because they do not know
this, but, rather, because of the specific reason why they do not. In the
two kinds of situation just mentioned, subjects can see that things are
as they appear to be without knowing this only because the factors that
are incompatible with knowledge are extrinsic to perception itself. One
is matter of whether or not a natural perceptual belief is inhibited by
collateral information—which is matter of how a subject epistemologi-
cally exploits a perception; and the other is a matter of whether a given
perception occurs in a situation in which there is a real possibility of
certain other (hallucinatory or illusory) perceptions occurring. By con-
trast, the intentionalist’s account of ‘‘veridical illusion’’ deals with fac-
tors that are internal to perception, since it concerns the aetiology of
perceptual experience itself and, therefore, how one stands perceptually
to a certain object. The notion of seeing that comes apart from (visu-
ally) knowing that just to the extent that the former, unlike the latter,
essentially concerns just the relation in which I stand perceptually to an
object (given certain conceptual capacities and attention).
In this way an intentionalist can give an account of ‘‘optimal’’
perception that does not involve any elements of naıve realism. It is,
therefore, this account that functions to undercut the naıve realist
account of such optimal perceptions. Note that I am here not simply
playing one perceptual theory off against another. I am not simply
saying that a representationalist has an account of wholly veridical,
‘‘optimal’’ perception that is at odds with naıve realism. The point is
that the only form of naıve realism still in the field is one that is
itself committed to endorsing a representationalist theory of illusion.
So, unless naıve realists can point out some error in the foregoing
extension of the theory to cover optimal perceptual states, they will
398 A. D. SMITH
themselves be saddled with an account that invalidates their own
constituency analysis of such states, since they cannot both endorse
representationalism as an account of illusory states and yet reject an
unobjectionable development of that position that accounts for opti-
mal cases. For this reason, and also because, as I suggested at the
beginning of this section, an ‘‘unmixed’’ account of partly veridical
perceptions is anyway in conflict with the motivation of naıve real-
ism, we must conclude that such a unmixed account must be
rejected. This was, however, naıve realism’s last stand in its attempt
to account for illusion. So, naıve realism must be entirely rejected.
Disjunctivism, if it is to survive, must be dissociated from it.
IV
In this section I consider the issue of where to place illusion in the
experiential disjunction—in the good or the bad disjunct—when
naıve realism has been rejected. This issue of where to place illusion
is not independent of the fate of naıve realism, since a disjunctivist
theory that embraces naıve realism is forced to allocate illusion to
the bad disjunct—at least as I have interpreted such an assignment.
This is because, in cases of veridical, non-illusory perception, naıve
realism accounts for the phenomenal character of experience simply
in terms of a certain physical object being a constituent of the expe-
rience. The ‘‘phenomenal yellowness’’ of my veridical perception of a
canary just is a matter of that yellow bird being a constituent of my
experience. The phenomenal yellowness of my experience when I illu-
sorily see some differently coloured object as yellow must be
accounted for in some other way—a way that necessarily has no
place in accounting for the phenomenal character of veridical percep-
tion, since here the constituent object does all the work. As Martin
writes, ‘‘The naıve realist account of perceptual experience . . . can-
not be applied to any case of delusive experience, such as illusions
where one does perceive an external object, but misperceives it as
other than it really is. If we suppose that such cases involve the
same type of mental state, perceptual experience, as veridical percep-
tion, then that will directly contradict the naıve realist account even
of those cases’’ (1997, 85).20 Since, however, naıve realism must be
20 In so far as Brewer’s form of naıve realism regards illusion as occurring subse-
quently to experience itself, in some ‘‘response’’ to such experience, he assigns illu-
sions, or at least the truly experiential components of illusions, to the good
disjunct. But that is because, in effect, Brewer does not recognise illusion as a genu-
inely sensory phenomenon at all.
DISJUNCTIVISM AND ILLUSION 399
rejected, the question of where to place illusion in the experiential
disjunction remains open.
Naıve realism involves a certain way of interpreting the common
disjunctivist claim that in good perceptual cases a certain object is a
constituent of experience. Such a constituency view does not in and
of itself amount to naıve realism, however. I have already quoted
Paul Snowdon claiming that such constituency is an essential ingredi-
ent in any theory that deserves to be regarded as a form of disjunc-
tivism; and yet he himself does not endorse naıve realism, in the
present sense, in any of his writings. Indeed, he informs me that he
rejects it. Does such a form of disjunctivism—one that accepts con-
stituency but rejects naıve realism—also determine where illusion is
to be placed in the experiential disjunction? It does: illusion must be
placed in the good disjunct—as Snowdon himself consistently does.
For in rejecting naıve realism—the view that a worldly constituent
of an experience determines the phenomenal character of that exper-
ience—the present form of disjunctivism is left with no role for
constituency other than to constitute an experience as being a
perceptual experience, rather than a mere hallucination. Constituency
simply ensures, or expresses the fact, that perceptual contact is made
with the real world. And for those disjunctivists (the vast majority)
who peddle the idea of constituency, it is this issue of whether a
perceived object is a constituent of a sensory experience or not that
alone determines the fundamental psychological kind to which the
experience is to be allocated—and, therefore, on which side of the
‘‘experiential disjunction’’ the experience falls.
Despite what Snowdon claims, however, it is not necessary for a dis-
junctivist to hold that worldly objects are constituents of ‘‘good’’ expe-
riences. He is certainly right that ‘‘the mere denial of a common
nature’’, as he puts it, does not fully capture the disjunctivist position.
It is necessary that this lack of a common nature derive from a differ-
ence in how one stands cognitively to the world. It is precisely because
certain experiences require the world to be a certain way, and require
us to stand in a certain cognitively favourable relation to that world,
that they count as good perceptual cases. With them, and with them
alone, is the world perceptually ‘‘present’’ to one in some sense that is
prized. A constituency view of experience is not required in order to
capture all of this, however. So we have not yet covered all possible
forms of disjunctivism.
What sort of theory it is that both counts as a form of disjunctivism
and yet rejects the idea that worldly objects are constituents of even good
experiences? According to many writers, any direct realist theory that
rejects a constituency account of perceptual experience amounts to some
400 A. D. SMITH
form of ‘‘representationalist’’ or ‘‘intentionalist’’ theory of experience.21
Such a theory holds that sensory experience, both perceptual and halluci-
natory, possesses ‘‘intentional (or representational) content’’ in virtue of
which such experience presents the world as being a certain way. What is
thus presented—some object or state of affairs in the world—is always
distinct from the experience that presents it. Intentional content also
determines, at least in part, the phenomenal character of experience.22 It
is, I believe, correct to hold that such a theory is the only direct realist
alternative to a constituency view—though calling it a ‘‘representational-
ist’’ theory may cause some hackles to rise. One can understand resistance
to the idea that anyone who denies that worldly objects are constituents
of experience is committed to viewing us as merely ‘‘representing’’ the
world when we plainly and directly perceive it. Perhaps the alternative
term ‘‘intentionalist’’ will help to allay such fears. Moreover, one can,
where perception is concerned, restrict the term ‘‘represent’’ and its cog-
nates to the experiences themselves and the sensory systems involved in
their production, and deny that the subject (‘‘merely’’) represents how
things are. When the world is represented by a subject’s experiences in a
certain way, the world is presented to the subject of these experiences. It
is, it may be held, the experiential character of such representations that
renders the term ‘‘represent’’ inappropriate as applied to the subject of
experience. If such scruples about ‘‘merely representing’’ the world are
respected, as I believe they can be, it does seem to me that intentionalist
(or, as I shall, in fact, also write, ‘‘representationalist’’) accounts of
experience are the only direct realist alternative to a constituency view.
Although many intentionalists are not disjunctivists at all, there is
certainly a possible marriage between the two positions.23 This will be
possible, however, only if an intentionalist account of experience recog-
nises two fundamentally different ways of experientially representing
the world—one characterising the good, and one the bad, disjunct.
These two ways must, for an intentionalist, be spelled out in terms of a
21 For example, Brewer (2006, 168) states that a constituency view is ‘‘the only alter-
native to characterizing experience by its representational content’’. Martin (2002)
concurs. Phenomenalists and certain indirect realists will fall into neither category,
of course. It is, however, only direct realist accounts of perception that need to be
considered here, since disjunctivism, whatever else it is, is essentially a direct realist
perceptual theory.22 Many intentionalist theories are reductive, in the sense that they claim that sensory
experience is nothing but a matter of representing the world in a certain way. In-
tentionalism need not be thus reductive. Husserl is a classical non-reductivist. More
recently Block (e.g., 2003) has advocated a non-reductive version. This issue is not,
however, relevant to the present discussion, however important in its own right.23 Alex Byrne (2001, 202 n7), for example, makes it clear that his own representation-
alist theory is compatible with disjunctivism.
DISJUNCTIVISM AND ILLUSION 401
difference in intentional content.24 Since it is definitive of disjunctivism
that one cannot enjoy a good experiential state without perceiving
some real object in the world, the sort of intentional content that is
required in order to confer such ‘‘good’’ status on an experience is
object-dependent content: a type of content, that is to say, that a psy-
chological state can have only when it is the case that there really is an
object that the state represents. Nothing short of this would guarantee
that the experiences that possess such content are of such a kind that
they are possible only when some real object is being perceived, as dis-
junctivism requires. More than this may, however, be required if an in-
tentionalist account is to do justice to the disjunctivist position, even
when it is specified that the psychological states in question are sensory
in character.25 Paul Snowdon has suggested to me that any case where
one hallucinates a familiar object shows that experiential object-depen-
dent content is too weak a notion to capture disjunctivist claims. The
psychological state here is, it may be thought, both sensory and object-
involving; and yet, being a hallucination, it falls under the bad disjunct.
For reasons I have given elsewhere (2002, 265–6), I reject the sugges-
tion that hallucinatory experiences themselves can contain the sort of
object-dependent content suggested by Snowdon. But let us suppose
that they can. If so, the intentionalist disjunctivist must appeal to a
more specific type of intentional content: one that requires not just the
(sometime) existence of its object, but that object’s presence. Such con-
tent, and the experiences that possess it, need to be, we might say, not
merely object-dependent, but object-involving.26 Such content is itself
24 Although, as mentioned in n22 above, intentionalism need not offer a reductive
account of sensory experience, and so can recognise something other than representa-
tional content as intrinsic to, and even as essential to, such experience, no such extra
feature is able to sustain a disjunctivist division of experiences into different funda-
mental kinds. The only non-representational variation in sensory states that any rep-
resentationalist will countenance is in some irreducible sensory features of such states:
qualia. Qualia, however, are supposed to account only for the phenomenal character
of experiences (at least in part). Since both illusions and hallucinations can be phe-
nomenally identical to veridical perceptions, no such variation in sensory character
can constitute the disjunctivist’s fundamental difference in kind of psychological state.25 According to Martin, ‘‘If one holds that the content of perceptual experience can
be singular, and also holds that singular content is object-dependent, then one will
thereby be forced to be disjunctivist’’ (2002, 395 n25). This may well be right; but,
as we are about to see, the point is not self-evident.26 Martin (2003) has signalled a different distinction by contrasting the terms ‘‘object-
involving’’ and ‘‘object-dependent’’. He draws his own distinction as part of an
exploration of a possible non-disjunctivist position that yet construes perceptual
experience as object-involving. (The account in question is Tyler Burge’s.) Since
Martin’s distinction is of significance only in relation to such an account, and since
that account is not disjunctivist, it is not relevant to the present discussion, what-
ever intrinsic interest it may have.
402 A. D. SMITH
sometimes characterised these days, employing a natural extension of
the term, as demonstrative. It is such content, the intentionalist will
hold, that underwrites successful perceptual judgement to the effect that
this object is such and such. It is the absence of such content in the hal-
lucinatory case that explains why any such demonstrative judgement
there misfires. This sketch will have to suffice here to indicate the sort
of intentionalist theory that can embody disjunctivism, since what we
are primarily concerned with are the consequences of embracing such a
theory for an account of illusion.27
Distinguishing between experiences on the basis of whether their
intentional content is or is not object-involving, or ‘‘demonstrative’’, is
clearly going to be at the heart of any intentionalist form of disjunctiv-
ism. Distinguishing experiences in this way will not, however, serve to
distinguish illusions from veridical perceptions, but only perceptions of
either sort from hallucinations. Both illusory and veridical perceptions
will have such object-involving content in virtue of the fact that both
are perceptions. The perfectly obvious essential difference between an
illusion and a hallucination is that in the former, but not the latter, a
real object is perceived—just as is the case with veridical perception.
An intentionalist will certainly attribute object-involving content to
veridical, non-illusory perceptions. But it is hardly because such ‘‘opti-
mal’’ perceptions are specifically veridical that this is the right thing to
say. My non-illusory perception of a green square does not possess
object-involving content because I get the square’s colour right: it is
simply because I am actually seeing the square—something I can do
even though I get the colour wrong as a result of illusion. My seeing
the square puts me in a position to make a certain sort of demons-
trative judgement: one to the effect that this object is green. This
judgement is possible, for the intentionalist, only because of the object-
involvingness of the experiential state of which it is an expression. The
same will hold when I misperceive the object as yellow. My natural
judgement, in such a case, to the effect that ‘‘this is yellow’’ will be
27 A final word on this issue, though. John Campbell (2002, 135–6) interprets such an
intentionalist account as construing experience in terms of grasping a demonstra-
tive, object-dependent thought. He then, quite reasonably, goes on to criticise any
such account as failing to do justice to the distinctive role that experience plays in
relation to thought. The intentionalist form of disjunctivism that we are now con-
sidering, however, views experiences themselves as possessing object-dependent con-
tent. Such experiences will put a suitably cognitively equipped subject in a position
to entertain a corresponding demonstrative thought; but they are not themselves
modes of thinking. (They could, indeed, be wholly non-conceptual in nature.)
Campbell interprets an intentionalist account in the way he does because of a
restriction of intentionality to thinking, which there is absolutely no reason to
accept.
DISJUNCTIVISM AND ILLUSION 403
mistaken, of course; but this judgement can be entertained at all only
because of the perceptual presence of the object itself—something that
the object-involving content of an experience alone guarantees.
Although intentionalist disjunctivism cannot distinguish between illu-
sory and non-illusory perceptual states in terms of the above sort of
object-involving content, it may be thought that it can do so in terms
of an element of content that relates specifically to features of objects.
When I ‘‘optimally’’ see a green square, I not only see the square, but
also, to echo Sellars again, its very greenness. On the basis of so seeing
the square and its colour I can think demonstratively of this green col-
our, and mean by that, not some repeatable shade of colour, however
determinate, but a particular instance of colour. When I hallucinate, I
can do no such thing. And perhaps I am not in a position to think
demonstratively about an object’s features when those features are illu-
sorily perceived (even veridically). The suggestion would be that in vir-
tue of the illusory nature of a perception of a thing’s colour, the
experience at best visually represents a certain colour as instantiated,
but not a particular instantiation of a colour. There is no actual
instance of a colour that it represents. This approach would give us a
neat threefold distinction. An optimal perception is demonstratively
related to its object and its veridically perceived features; an illusory
perception is demonstratively related to its object, but not to its illuso-
rily perceived features; and hallucination is demonstratively related to
nothing, though it putatively is. On this basis one could allocate illu-
sion to the bad disjunct in virtue of its not, in so far as it is illusory,
representing an object’s features in a manner suitable for sustaining
demonstrative judgement.
This proposal will not work, however, since what has just been said
of ‘‘optimal’’ perceptions, as supposedly distinctive and essential to
them, applies equally, and essentially, to illusory perceptions. When we
illusorily perceive an object, we perceive the object’s features—even the
ones that illusorily appear to us. To deny this is to deny that we see an
object at all in at least some illusory cases. For one cannot perceive a
thing at all without perceiving some feature that it has; and one cannot
consciously perceive a thing at all (so that it consciously seems a cer-
tain way to the subject) without consciously perceiving some feature
that it has (and that, therefore, itself appears a certain way to the sub-
ject). We have already seen that visually perceiving an object’s shape
requires seeing that object’s colour. So if, in seeing a green square that
looks yellow to me, I did not see the object’s colour, I would not see
its shape either. If these are the only two features of the object that
appear to me, as may be the case, I would therefore not see anything
of the object at all. Indeed, we can suppose that I misperceive both the
404 A. D. SMITH
object’s colour and its shape. If it is claimed that, because of illusion, I
fail to see either of these features of the object, then, even more obvi-
ously, I end up not seeing it at all.28 So, when that green square looks
yellow to me, I perceive its colour: its actual colour, that is, since it has
no other. Since I see the square’s colour, and that colour is green, I see
the square’s greenness (though not as green, of course). It is, indeed,
precisely this greenness that looks yellow to me. What other feature
could it be? And nothing can look a certain way to me unless I see it.
If it had had a different colour, doubtless that colour would not have
looked yellow to me in this light. Again, suppose the colour of this
green square were (really, objectively) to turn a darker shade—in
response, say, to a change in temperature. In some, perhaps most, such
cases, the apparent yellowness would seem to become darker. Here I
would be seeing the actual darkening of the square’s colour—some-
thing I could not do were I not seeing its colour. It is only because of
the fact that, when we suffer illusion, we illusorily perceive the actual
features of an object that a suitably cognitively equipped subject can,
on the basis of an illusory experience, make a perceptual demonstrative
judgement about the actual features of the object perceived. Seeing the
green square as I do—that is, as yellow—I could judge that this col-
our—that is to say, this colour instance—is qualitatively identical to
that of some canary I had recently seen. Here I would be making a per-
ceptually based demonstrative judgement about the actual colour of
the square: something I could not do if I were not perceiving it and
visually representing it in an object-involving way. That I would indeed
be making such a judgement is clear from the fact that such a judge-
ment would be false—rather than entirely lacking an object or subject
matter. And the reason for this is that the square, unlike the canary, is
not really yellow.29 Even illusion is, we may say, ‘‘feature-involving’’.
Intentionalism cannot, therefore, distinguish between illusory and
veridical perceptual states in term of an intrinsic difference in inten-
tional content. It can do so in terms of the distinction between the
accuracy and inaccuracy of the intentional content; but this is an
extrinsic matter. We saw earlier that the existence of veridical illusion
shows that a distinction between accuracy and inaccuracy is not
28 A surprisingly large number of philosophers have claimed that it is impossible for
all the perceived features of an object to be perceived illusorily: that illusion must
be partial. There is nothing whatever to be said for this view. For a brief discussion
and diagnosis of this error, see my (2002, 82–3).29 I could, of course, judge that the colour the square appears to me to have is identi-
cal to that of the canary, and then I would be correct. In the straightforward situa-
tion in question, however, I am making the simpler judgement about the colour of
the square.
DISJUNCTIVISM AND ILLUSION 405
sufficient to account for the distinction between illusory and non-illu-
sory (or, at least, optimal and non-optimal) perceptual states. Never-
theless, the extra resources that were employed there to make good this
deficiency are insufficient to locate illusion in the bad disjunct. All that
was required for a state to be non-optimal, even though it was veridi-
cal, was that it not put one in a position either to know that or to see
that the perceived object is really as it appears (in some relevant
respect), and this was spelled out in terms of the perceptual processes
involved only fortuitously resulting in veridical perceptual states (when
they do). This, too, is an extrinsic matter, at least for an intentionalist.
Indeed, for an intentionalist, the only relevant intrinsic features of
intentional states are features of intentional content.30 But on this
score, as we have just seen, there is no differentiating between optimal
perceptual states and non-optimal ones.
Nevertheless, it may occur to some readers that bringing into the in-
tentionalist account of non-optimality a reference to the possibility of
knowledge provides the basis for regarding illusory perceptions
(whether veridical or not) as intrinsically different in kind from any
non-illusory perceptions. In particular, readers may be reminded of
John McDowell’s form of disjunctivism. He contrasts psychological
states, allocating them to the good or the bad disjunct, precisely in
terms of their potentiality for giving the subject knowledge. The dis-
junction that McDowell typically employs is that of a fact being mani-
fest to one as opposed to the mere appearance of such a fact. That a
fact is manifest to one does not mean that one knows it to obtain.
What the manifestness of a fact does, rather, is to ‘‘make knowledge
of the fact available to one’’ (McDowell 1982, 457). According to
McDowell, ‘‘for some purposes the notion of being in a position to
know something is more interesting than the notion of actually know-
ing it’’ (ibid., n1). If, however, psychological states are allocated to the
good or the bad disjunct on this basis, it looks as if the intentionalist
account ought to put illusion squarely in the bad disjunct of the
perceptual disjunction.
As I mentioned earlier, however, although McDowell is certainly a
disjunctivist of sorts, his disjunctivism is not a perceptual theory at all.
That his concern is but indirectly related to the issue of the nature of
perceptual experience is made clear when he considers a subject whose
visual system properly operates only fitfully, in such a way that he is
prone to hallucinate things. On an occasion where it is working prop-
erly (though he does not know this), the subject sees a certain object,
and sees it veridically and non-illusorily. According to McDowell, this
30 See n24 above.
406 A. D. SMITH
constitutes a ‘‘bad’’ case. This subject is not in a position to acquire
knowledge, and no fact is manifest to him, since for all he knows he
could be hallucinating. Although, in McDowell’s example, the subject
genuinely sees a tomato in front of him, he ‘‘need not count as experi-
encing the presence of a tomato. . . One counts as experiencing the fact
making itself manifest only in the exercise of a (fallible) capacity to tell
how things are’’ (ibid.). For McDowell, therefore, even a veridical,
non-illusory perceptual experience is, in virtue of an unfavourable con-
text, to be allocated to the bad disjunct. It is not types of experience as
such that McDowell is interested in classifying disjunctively, but experi-
ences-in-a-context. And this is the most that a theory that works with
the notion of being in a position to know can claim.
We must conclude, therefore, that intentionalism cannot regard illu-
sion as an intrinsically different sort of fundamental psychological state
from non-illusory perception, and so is unable to place illusion any-
where other than in the good disjunct. Since this is also true of Snow-
don’s version of disjunctivism, and these are the only two disjunctivist
alternatives to the discredited naıve realist theory, we must conclude
that disjunctivism as such must allocate illusion to the good perceptual
disjunct.
V
We should finally consider what significance such a ‘‘concession’’ may
have. Presumably those philosophers who have both embraced disjunc-
tivism and have placed illusions in the bad disjunct did so because they
thought that otherwise some unfortunate cognitive consequence would
ensue; or that some cognitive virtue would otherwise be lost. In order
fully to appreciate the possible effect of our concession, we should need
to see what it precisely is that disjunctivism is supposed to uphold; and
unfortunately there are almost as many such aims as there are disjunc-
tivists. I shall, therefore, restrict myself to addressing two related con-
cerns that are, I think, fundamental. Disjunctivists, whatever other
aims and interests they may have, are concerned to defend direct real-
ism, and believe that disjunctivism is the only way to defend it. It may
be thought that the present concessive proposal puts such direct realism
in peril, because it requires us to acknowledge that there is a ‘‘highest
common factor’’ to illusory perceptual states and veridical, non-illusory
ones. This may be thought worrying, because it implies that when we
normally and non-illusorily see a thing, we do so in virtue of having a
type of experience that might have been illusory. This does not, how-
ever, in any way imperil direct realism. It might, if the common funda-
mental kind to which both illusory and non-illusory perceptions are
DISJUNCTIVISM AND ILLUSION 407
assigned were insufficient to constitute direct awareness of the world.
One can, therefore, understand a worry about the suggestion that there
is a highest common factor, a complete identity in psychological nat-
ure, in veridical perception and a possible hallucination. Since a hallu-
cinatory state involves no awareness of the world, and yet just such a
state supposedly constitutes the entire experiential dimension of a
veridical perception, even the latter can, it may be thought, afford no
experiential direct awareness of the world either. I do not say that this
line of thought is unanswerable, but one can at least see the worry.
When it comes to veridical and illusory perceptions, however, the
worry is not even visible, since the latter do give one a direct awareness
of the world, in virtue of being, unlike hallucinations, perceptions. At
least we need some argument to show that they do not; and I cannot
think of one that ought to be convincing to a disjunctivist.
There is, however, a remaining concern that may be thought not to
be adequately addressed by pointing out that allocating illusion to the
good disjunct in no way impugns direct realism. This concern is to pre-
serve the thought that in non-illusory perception, and here alone, an
object and its perceptible features are manifest to one. When I non-illu-
sorily perceive a green square, its greenness is immediately present to
my consciousness. When it illusorily looks yellow, however, something
less than this is the case—even when, as in veridical illusion, it really is
yellow. This observation, even if correct, will support the decision to
place illusion in the bad disjunct, however, only if such manifestness is
an intrinsic and essential characteristic of a certain kind of experience.
It can be this, however, only if naıve realism is true, which it is not.
For in order for the observation to be correct, manifestness must at
least require that the object be accurately perceived. So, if manifestness
is an intrinsic and essential feature of a kind of experience, so is accu-
racy. Accuracy is clearly not an essential feature of experiences accord-
ing to intentionalism. Nor is it according to Snowdon’s version of the
constituency theory. It can be an essential feature only on the naıve
realist account of things, where constituency—certainly an essential
matter—determines the phenomenological character of an experience,
and hence ensures accuracy. Naıve realism is false, however; and these
are the only direct realist options. Such manifestness cannot, therefore,
be an essential feature of experiences, and our observation loses its
force as an objection to placing illusion in the good disjunct. If there is
a sense of ‘‘manifest’’, as doubtless there is, according to which the
world is never manifest to one in virtue of perceiving illusorily, then
such manifestness is something that accrues accidentally to any intrinsi-
cally specified kind of experience. There is, perhaps, a weaker sense of
the term according to which only certain intrinsically specifiable kinds
408 A. D. SMITH
of experience allow the world to be ‘‘manifest’’ to us. In this sense,
however, for the world to be manifest is just for it to be directly per-
ceived—something that holds for perceptual experience as such,
whether it be illusory or not. It is understandable why one might want
such openness to the world to be written into the very nature of a cer-
tain kind of experience. To demand that, in addition, accuracy be built
into the very nature of such experience is to want too much. Indeed,
disjunctivists should not be at all hesitant in allocating illusion to the
good disjunct, since the clear fundamental distinction in this area is
between all perceptions, whether illusory or not, on the one hand, and
mere hallucinations on the other. Both illusory and non-illusory per-
ceptions are genuine perceptions. With both, perceptual contact is
made with some real item in the physical world, and it itself appears
some way to the perceiver.31 On their basis, demonstrative thoughts
about a perceived object are equally possible for suitable cognitively
equipped subjects. All this indicates a fundamental sameness of kind.
References
Block, N. 2003. ‘‘Mental Paint’’. In Reflections and Replies. Essays on
the Philosophy of Tyler Burge, ed. M. Hahn and B. Ramberg,
165–200. Cambridge, MA: MIT Press.
Brewer, B. 2006a. ‘‘Perception and Content’’. European Journal of
Philosophy 14: 165–181.
—— 2006b. ‘‘How to account for Illusion’’. In Disjunctivism: Percep-
tion, Action, Knowledge, ed. A Haddock and F. Macpherson,
168–180. Oxford: Oxford University Press.
Byrne, A. 2001. ‘‘Intentionalism Defended’’. Philosophical Review 100:
199–240.
Crane, T. 2006. ‘‘Is There a Perceptual Relation? ’’ In Perceptual Expe-
rience, ed. T. S. Gendler and J. Hawthorne, 126–146. Oxford:
Clarendon Press.
Hinton, J. M. 1973. Experiences. Oxford: Clarendon Press.
Johnston, M. 2006. ‘‘The Function of Sensory Awareness.’’ In Percep-
tual Experience, ed. T. S. Gendler and J. Hawthorne, 260–290.
Oxford: Clarendon Press.
31 Tim Crane (2006, 134) characterises disjunctivism as the thesis that ‘‘there is no
common fundamental kind of state—‘perceptual experience’—present in cases of
genuine perception, which is a relation to a mind-independent object, and illusion
and hallucination’’. The suggestion that illusion is not a relation to a mind-inde-
pendent object is quite mistaken (unless idealism be true). In virtue of being a kind
of perception, rather than a hallucination, that is precisely what it is. The addi-
tional suggestion that illusory perception is not ‘‘genuine perception’’ is also at
least tendentious.
DISJUNCTIVISM AND ILLUSION 409
McDowell, J. 1982. ‘‘Criteria, Defeasibility, and Knowledge’’. Proceed-
ings of the British Academy 68: 455–479.
Martin, M. G. F. 1997. ‘‘The Reality of Appearances.’’ In Thought and
Ontology, ed. M. Sainsbury, 81–106. Milan: FrancoAngeli.
—— 1998. ‘‘Setting Things Before the Mind.’’ In Contemporary Issues
in the Philosophy of Mind, ed. A. O’Hear, 157–179. Cambridge:
Cambridge University Press.
—— 2002. ‘‘The Transparency of Experience’’. Mind & Language 17:
376–425.
—— 2004. ‘‘The Limits of Self-Awareness’’. Philosophical Studies 120:
37–89.
Smith, A. D. 2002. The Problem of Perception. Cambridge, Mass.:
Harvard University Press.
—— 2008. ‘‘Disjunctivism and Discriminability.’’ In Disjunctivism: Per-
ception, Action, Knowledge, ed. A. Haddock and F. Macpherson,
181–204. Oxford: Oxford University Press.
Snowdon, P. 1990. ‘‘The Objects of Perceptual Experience I.’’ Proceed-
ings of the Aristotelian Society, Supplementary Volume LXIV,
121–150.
—— 2005. ‘‘The Formulation of Disjunctivism: A Reply to Fish.’’
Proceedings of the Aristotelian Society CV: 129–141.
410 A. D. SMITH
top related